docs(jfm): establish traceable manuscript planning baseline
Track the research dossiers, section freezes, supporting manuscript materials, and round-aware agent controls so future drafting decisions can be reviewed across both repository mirrors. Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
Frank14f
@@ -0,0 +1,33 @@
|
||||
# Round 4 Display and Equation Inventory
|
||||
|
||||
**Inventory only.** No candidate panel, source array, rendering, table, or publication-ready figure exists merely because it is inventoried. Text readiness and panel/artifact readiness are independent.
|
||||
|
||||
Classifications: `required`, `optional`, `dependency-gated`, `blocked`, `author-supply-gated`.
|
||||
|
||||
| Display/equation family | Classification | Owner | Argumentative job | Readiness/dependency | Fallback |
|
||||
|---|---|---|---|---|---|
|
||||
| **Fig. 1 capability/progression** | `dependency-gated` | Capability | Steady discovery → Kármán → Vortex → `ill_075L`, Kármán dominant, configurations and evidence definitions distinct | Text architecture ready; each panel needs chain/case/seed/role/comparator/coordinate/metric/source/render provenance. Vortex event L2 blocked; Illusion PPO only; no experiment. Re400 optional/drop-first | Retain sourced panels; omit blocked annotations/Re400 |
|
||||
| **Section 2 two-configuration geometry and DRL-flow schematics/tables** | `required` | Problem/method/evaluation | Decode parabolic/no-slip and uniform/free-slip configurations; body/action/sign; $x/D=10$, $y/D=0,\pm2$, radius-$0.25D$ two-component sensing; target/reward flow; ROI/roles/$Re_D$/metrics; compact CFD | Architecture frozen; sensor note is `read` and fixes geometry/ROI/metric boundaries; CelerisLab dossier is `read/full-report` and fixes K2 `partial pass` / S2 `pass` for current CelerisLab only. Redraw, per-result rows and artifact binding remain production dependencies. Does not prove observability, optimal sensing, configuration equivalence, canonical rerun completion, an exact mask, or older-solver validation; makes no Fig. 1 decision | Minimal geometry + DRL flow schematic and compact tables; provenance/detail to supplement |
|
||||
| **Kármán performance display** | `dependency-gated` | Kármán performance | L2-first parabolic/no-slip mean/phase result; DTW complementary; uniform/free-slip retention secondary | Text/numbers bound; fields need exact roles, manifests, ROI/mask/window, rendering. Separate from M06 | Compact result table; omit unsourced redraws |
|
||||
| **Section 2 method/evaluation equations** | `required` minimum | Problem/method/evaluation; reused by performance | Define bulk/mean-based $Re_D$, physical actuation, objective roles, `E_mean`, phase-resolved L2 and optional comparator-relative form; give DTW recurrence/path once because uncommon | Definition-ready at architecture level; current defaults source-bound; exact per-result settings and values pending, including learning-rate variation across retained runs. Spatial L2 is not Level/Stage; DTW is not delay/field equivalence | Keep only reused equations; move algorithm/settings detail to supplement |
|
||||
| **SR law/deletion** | law/deletion `required`; PPO/SR/Zero panel `dependency-gated` | SR | Exact executable law/symmetry; deployment; largest-tested persistent term | Law/deletion text-ready. Same-protocol panel blocked without exact roles/comparator/window. Deletion noncausal | Keep law, symmetry, deletion; omit direct role panel |
|
||||
| **CCD mean/share display** | `dependency-gated` | Field/CCD | Target/zero/constant/DRL means before residual; scalar error-difference accounting | Ledger bound; redraw needs publication arrays and exact mask/window. Cannot validate M04 | Compact scalar ledger if fields unavailable |
|
||||
| **CCD residual/modes/POD sequence** | `dependency-gated` | Field/CCD | Define non-paired residual; show action-correlated modes, phase, POD boundary | Equation/text ready; panels need corrected parent/derived arrays, orientation, lineage, captions. No causality/lag/energy/superiority | Keep residual equation and concise statement/table |
|
||||
| **Cross-case display** | overall `dependency-gated`; quantitative Vortex/Erase `blocked` | Cross-case | Event-relative Vortex, non-zero-target Illusion PPO, Erase boundary | Vortex visuals/Illusion capability are bounded candidates. Vortex L2 needs roles/offsets/ROI/mask/evaluator; Illusion L2 conditional; Erase roles absent | Use sourced visuals; omit quantitative Vortex/Erase and state gaps |
|
||||
| **Cross-case event equation** | `optional` | Cross-case | Define conditional event-relative ROI error without calling offsets phase | Definition creates no roles/mask/evaluator/result | Use prose/table semantics |
|
||||
| **Steady/theory display** | `dependency-gated` | Steady/theory | Separate uniform/free-slip endpoint/sign context; accepted/reverse boundary; failed closure | Text ledger ready. Fields need payloads and reverse amplitude-mismatch caption. New MFS field blocked | Sign schematic + endpoint table + one-circle/sign conflict; omit MFS field |
|
||||
| **Steady/theory equations** | `required` minimum | Steady/theory | Wall law/spin ratio, steady metric, and shortest outer-reference or partial-budget boundary | Existing derivation/check states distinct; no inviscid rotating no-slip closure | Full MFS/multipole/strip/CV derivations to supplement |
|
||||
| **Experiment display** | `author-supply-gated` | Experiment | Qualitative open-loop apparatus and, if gated, steady/Illusion dye observations | Needs Frank's conditions, action/sign, trials, raw media, scale/frame rate, processing, repeats, permissions. No calibrated force/PIV panel | Omit images; retain short author-written interface or bounded supplement media |
|
||||
| **Integrated conclusion aid** | evidence map `optional`; limitations table `required` | Conclusion | Synthesize ladder, compatible/non-equivalent links, failures, decisive tests | Text-ready. Map is not causal or pooled. Conclusion adds no scientific equation | Keep limitations table; omit map if misleading/cluttered |
|
||||
|
||||
## Equation ownership
|
||||
1. Problem/method/evaluation owns reusable bulk/mean-based $Re_D$, action/sign, objective/role, DTW, and field-metric definitions.
|
||||
2. SR owns reflection map and canonical law; no full reprint elsewhere.
|
||||
3. Field/CCD owns scalar shares and residual; CCD operator only if interpretively necessary.
|
||||
4. Cross-case event equation is conditional, never a result.
|
||||
5. Steady/theory owns `s`, steady metric, and selected outer-reference/budget equation.
|
||||
6. Conclusion introduces no equation, numerical result, citation, or artifact.
|
||||
7. Freeze notation once for `Kármán`, `Re_D`, `Re_code`, spatial L2, DTW, ROI/mask, action order/sign, and percentages; delete duplicates or move them to supplement.
|
||||
|
||||
## Panel readiness check
|
||||
Before calling any item panel-ready, record exact authority/artifact; chain/solver/geometry; case/seed/role; comparator; metric/estimand; ROI/domain/mask; weighting/normalization; window/phase/event coordinate; array orientation; rendering provenance; caption limitations; and artifact-audit state. Missing required fields mean `dependency-gated`, `blocked`, or `author-supply-gated`, with the listed fallback.
|
||||
@@ -0,0 +1,115 @@
|
||||
# Round 4 Global Decision Ledger
|
||||
|
||||
**Owner:** Round-4 coordinator.
|
||||
**Initial status:** `discuss`; unresolved author choices are not frozen.
|
||||
**Role:** architecture/ownership/claim/supersession source of truth, not prose or evidence.
|
||||
|
||||
## Controlling decisions carried from R2/R3
|
||||
| Decision | Disposition | Boundary |
|
||||
|---|---|---|
|
||||
| Provisional title: *Interpreting learned hydrodynamic cloaking and illusion in a fluidic pinball* | carried, provisional | No colon/opening “A”; no positive Illusion SR implied |
|
||||
| Concept = cloaking/illusion; execution = wake-signature matching; evaluation = field matching with L2 primary/DTW complementary; interpretation = control skeleton | frozen input | R2 freeze/style and R3 audit |
|
||||
| The periodic Kármán cloak under the parabolic-inflow/no-slip-wall configuration is the sole deep case | frozen input | Cross-cases do not inherit interpretation; Reynolds naming is owned by Section 2 and bare `Re100` is prohibited |
|
||||
| The parabolic-inflow/no-slip-wall and uniform-inflow/free-slip-wall configurations are separate | frozen input | Never pool, continue, validate, or imply equivalence; internal chain IDs remain provenance-only |
|
||||
| Physical progression: steady discovery → Kármán → Vortex → Illusion/Erase → qualitative experiment | frozen input | Early steady seed; later full steady/theory return; experiment is a short author-owned main-text subsection near the end |
|
||||
| Compact CFD qualification | frozen input | Selected observables only |
|
||||
| Positive SR Kármán/cloak-only; Illusion SR historical/negative | frozen input | No revived positive claims |
|
||||
| `action-correlated correction-field modes`; OID claim-free | frozen input | CCD descriptive/noncausal; no OID result |
|
||||
| Case-specific metrics | frozen input | Periodic/event/Erase/steady/DTW non-equivalent |
|
||||
| Experiment qualitative, open-loop, author-written, provenance-gated | frozen input | No quantitative validation |
|
||||
| R3 modules text-ready, not manuscript/panel-ready | frozen input | R3 final audit |
|
||||
|
||||
## Global claim ladder
|
||||
1. **Named cross-task capability:** named wake-signature or visual PPO capability across declared Kármán, Vortex, and Illusion tasks under distinct contracts.
|
||||
2. **Periodic Kármán cloak under the parabolic-inflow/no-slip-wall configuration — independent field matching:** spatial field matching only for the named target, controlled, and physical-zero roles; mean and phase L2 separate, DTW complementary.
|
||||
3. **Periodic Kármán cloak under the parabolic-inflow/no-slip-wall configuration — bounded control-skeleton interpretation:** executable SR, corrected four-role mean comparison, and non-paired residual organized by action-correlated correction-field modes support dominant persistent rear rotation plus smaller dynamic correction.
|
||||
|
||||
Level 3 is not causal, necessary, sufficient, unique, optimal, stable, robust, universal, transferable, or independent triangulation. Cross-case capability cannot inherit Levels 2–3.
|
||||
|
||||
## Frozen section order
|
||||
Round-4 freeze order:
|
||||
`problem/method/evaluation → capability → Kármán performance → SR → field/CCD → cross-case → steady/theory → experiment → conclusion → Introduction (last)`
|
||||
|
||||
Assembled manuscript order:
|
||||
`Introduction → problem/method/evaluation → capability → Kármán performance → SR → field/CCD → cross-case → steady/theory → experiment → conclusion`.
|
||||
|
||||
Introduction freezes last so framing cannot outrun settled results.
|
||||
|
||||
## Ownership and duplication
|
||||
| Owner | Owns | Allowed elsewhere | Prohibited duplication/inference |
|
||||
|---|---|---|---|
|
||||
| Introduction | Full ladder, concept/gap, contribution | Conclusion concise restatement | Detailed values, packages, priority/mechanism |
|
||||
| Problem/method/evaluation | Configurations/action/Re/roles/objective/metrics/ROI/windows/compact CFD | Results import definitions | Results narrative or configuration/evidence transfer |
|
||||
| Capability | Fig.1 progression, short Kármán preview, steady seed | Performance owns values; steady owns detail | Pooled efficacy, full result prose, experiment panel |
|
||||
| Kármán performance | Detailed parabolic-inflow/no-slip-wall L2/DTW; separate uniform-inflow/free-slip-wall seed/retention boundary | Capability headline only | M06 as replication/agreement |
|
||||
| SR | Parabolic-inflow/no-slip-wall formula/symmetry/deployment/deletion | Steady recurrence context only | Full steady/field detail, causal ablation, positive Illusion SR |
|
||||
| Field/CCD | Corrected means/shares, residual, CCD/POD boundary | Synthesis brief statement | SR derivation, causal response, M04 closure |
|
||||
| Cross-case | Vortex boundary, Illusion PPO/negative SR, Erase gap | Capability preview; conclusion envelope | Kármán transfer, common metric |
|
||||
| Steady/theory | Uniform-inflow/free-slip-wall endpoint/sign/context, theory failure/budgets | Capability seed; conclusion limit | Cross-configuration continuation, amplitude agreement, mechanism/optimum |
|
||||
| Experiment | Author provenance and qualitative observation | Conclusion gate only | Quantitative CFD validation/numbers |
|
||||
| Conclusion | Integrated answer, limits, decisive tests | No new evidence | Detailed replay/new equations/results |
|
||||
|
||||
Near-collision: M04 standardized PPO and M06 corrected mean values are **different estimands**. SR action magnitude and V5 steady spin ratio are **related contextual diagnostics**, not amplitude agreement. No arithmetic or replication/validation/agreement/triangulation language.
|
||||
|
||||
## Fixed scientific boundaries
|
||||
Cloaking/illusion is declared-reference matching, not universal invisibility. Positive SR is Kármán/cloak-only; Illusion SR is negative/historical. CCD is descriptive/noncausal; OID claim-free. Vortex is named comparable-scale transient capability; Erase quantification stays blocked. Steady is separate-chain base-bleed/gap-jet-family context, not boat-tail or mechanism. Theory constrains without finite-Re closure. Experiment is qualitative/open-loop/provenance-gated. No mechanism, momentum restoration, stability, optimality, robustness, universality, priority, or quantitative experimental validation is licensed.
|
||||
|
||||
## Open author choices
|
||||
| Choice | Recommendation / safe default | Status | Consequence |
|
||||
|---|---|---|---|
|
||||
| Final title | Retain provisional title unless reopened after section freezes | `discuss` | Any change needs terminology/claim audit |
|
||||
| Fig.1 set and optional Re400 marker | Kármán dominant; retain sourced steady/Vortex/`ill_075L`; drop Re400 first | `discuss` | Omission fallback; ladder unchanged |
|
||||
| Standardized PPO/SR/Zero panel | Omit without same-protocol roles/window | `discuss` | Text unaffected |
|
||||
| CCD manuscript label | Exact preferred derived publication plus validated parent; avoid “canonical CCD” | `discuss` | Caption/reproducibility only |
|
||||
| Quantitative Vortex/Erase panels | Default omit while blockers remain | `blocked` | Named visual capability may remain |
|
||||
| MFS field panel | Default omit until validation gates close | `blocked` | Use one-circle/sign-failure boundary |
|
||||
| Experiment image/final wording | Frank supplies prose/provenance; omit image if gate fails | `blocked` | Keep qualitative interface only |
|
||||
| Final display/equation compression | Minimum argumentative set | `discuss` | Move optional items to supplement/omit |
|
||||
|
||||
Defaults do not make unresolved rows frozen.
|
||||
|
||||
## Supersession rules
|
||||
1. Current authority/manifests/immutable artifacts/claims supersede summaries and memory.
|
||||
2. R2 `05_REVISED_STRUCTURE_FREEZE.md` supersedes conflicting R2 `01`–`04`; R2 `06` controls style.
|
||||
3. R3 `11_FINAL_CONSISTENCY_AUDIT.md` and reconciled exports control R3 conflicts; detail remains with owning module.
|
||||
4. A Round-4 change records old/new decision, reason, authority/author instruction, affected units, date, and re-audits; never erase provenance.
|
||||
5. Author choices cannot override evidence, merge chains/metrics, fabricate readiness, or raise claim ceilings.
|
||||
6. Reopen `frozen` only for new authority, discovered conflict, or explicit author redirection; dependent sections are re-audited.
|
||||
7. Missing optional panels do not reopen text when omission/supplement fallback exists; missing required evidence does.
|
||||
|
||||
## Author feedback decisions — 2026-08-10
|
||||
- Overall manuscript structure and logic are accepted.
|
||||
- Section proportions are provisionally accepted and will be refined after all section discussions.
|
||||
- A distinct short Conclusion is required.
|
||||
- The experiment remains a short author-owned main-text subsection near the end.
|
||||
- The Introduction architecture is the accepted five-paragraph sequence: fascination/gap → pinball/force mediation → nonlinear feedback/PPO/analytical hooks → realization, independent evaluation, action-side and field-side analysis → narrowed question/contribution ladder/contracts bridge.
|
||||
- Reader-facing `Legacy`/`V5` naming is prohibited. Use `parabolic-inflow/no-slip-wall channel configuration` and `uniform-inflow/free-slip-wall channel configuration`; retain internal chain IDs only for provenance and ownership separation.
|
||||
- Reynolds naming is frozen in Section 2: use article-wide $Re_D=U_{ref}D/\nu$ with bulk/mean inlet speed for the parabolic configuration and uniform inlet speed for the uniform configuration; state $U_{max}\simeq1.5U_{ref}$ for the parabolic profile, convert authoritative $2D$ code labels by one half, prohibit bare `Re100`, and never infer equality from inherited labels.
|
||||
- Section 2 reader-facing title is `Problem formulation, control framework and evaluation`; `contract` remains internal governance language only.
|
||||
- Section 2 architecture is frozen. Main text foregrounds the canonical improved method and uses one compact neutral sentence that later deep analyses use data generated with an earlier implementation under the same parabolic-inflow/no-slip-wall physical configuration; implementation differences and the absence of a complete canonical rerun are documented in the supplement. No canonical-generation, equivalence, reproduction, or rerun-completion claim is licensed.
|
||||
- The sensor note is `read`: placement is $x/D=10$, $y/D=0,\pm2$, radius $0.25D$, with both velocity components, and its ROI/metric boundaries are current. Raibaudo supports the streamwise precedent, Cornejo Maceda and Li support downstream phase/state sensing, and Marra (`mapped/full record`, DOI `10.1017/jfm.2024.593`) supports only low-dimensional controlled-flow plausibility; none establishes optimality or observability.
|
||||
- The CelerisLab validation dossier is `read/full-report`: current-solver Kan99b K2 retains the authority's exact `partial pass` label and Sah04 S2 is `pass`; the Kan99b mean-lift sign difference is understood as the CFD coordinate/force-direction convention, not unresolved physics or suspected wrong rotation. No older-solver validation is licensed.
|
||||
- Fig. 1 remains undiscussed; no capability-panel decision is inferred; Section 2 geometry and DRL-flow schematics are required method displays.
|
||||
- The Introduction may be revisited after detailed sections, but its current Round-4 architecture is `frozen`; cultural-hook and citation wording remain drafting choices.
|
||||
- Round-4/5 work must prioritize author-facing argument and design decisions over dossier-scale governance; retain only a compact controlling-read/gap ledger.
|
||||
- Future section work uses a two-section cadence: the coordinator presents two detailed minimum context packets together, records author feedback, launches two tightly scoped continuation agents in parallel, then performs a coordinator second audit before freeze or R5 handoff.
|
||||
- Every minimum context packet lists exact controlling reads, accepted author feedback, available materials and readiness, requested output decision, exclusions, target length/section share, paragraph jobs, concrete display layout, ownership and exact questions. Continuation agents use targeted Nowledge recall first when prior decisions matter, verify against repository authorities, and do not rediscover the project.
|
||||
|
||||
## Provenance separation
|
||||
Internal IDs `Legacy` and `V5` remain valid only in contracts, artifact paths, ledgers, and audit trails where they identify distinct evidence chains. Replacing manuscript-facing labels does not merge, erase, or transfer ownership between those chains.
|
||||
|
||||
## Coordinator update record
|
||||
Format: `date | item | previous | new | authority/author instruction | affected units | required audit`.
|
||||
|
||||
| Date | Item | Previous | New | Authority / author instruction | Affected units | Required audit |
|
||||
|---|---|---|---|---|---|---|
|
||||
| 2026-08-10 | Overall architecture | Under discussion | Structure/logic accepted; proportions provisional pending all section discussions | Author feedback | All Round-4 section freezes | Recheck final assembled balance |
|
||||
| 2026-08-10 | Conclusion and experiment | Conclusion/experiment placement carried but not restated as author decision | Distinct short Conclusion required; short author-owned experiment subsection retained near manuscript end | Author feedback | Experiment, Conclusion, assembled order | Preserve qualitative/provenance gates |
|
||||
| 2026-08-10 | Introduction | Freeze candidate with unresolved reader adjectives | Five-paragraph architecture frozen; exact boundary-condition configuration labels adopted | Author feedback plus repository configuration/kernel evidence | Introduction, Contracts, captions/ledgers | Revisit wording after detailed sections without reopening architecture |
|
||||
| 2026-08-10 | Reynolds naming | Configuration-local convention with parabolic reference unresolved | Section 2 freezes article-wide bulk/mean-based $Re_D=U_{ref}D/\nu$; uniform speed for uniform inlet; $U_{max}\simeq1.5U_{ref}$ for parabolic inlet; authoritative $2D$ labels convert by one half | Author feedback and repository evidence | Section 2, Introduction, labels/captions | Confirm every displayed inherited label against source authority |
|
||||
| 2026-08-10 | Fig. 1 | Open choice | Undiscussed; no new decision | Author feedback | Capability/display planning | Keep open |
|
||||
| 2026-08-10 | Section 2 formulation and provenance | Freeze candidate; sensor/validation authorities misclassified; provenance too defensive | Architecture frozen; canonical main text plus compact neutral earlier-implementation sentence; differences and no-complete-rerun fact supplementary; sensor note `read`; CelerisLab dossier `read/full-report`; no equivalence claim | Author correction plus current training, sensor and validation authorities | Section 2, captions, supplement provenance, inventories | Keep run values and production dependency-gated; no Fig. 1 decision |
|
||||
| 2026-08-10 | Agent-control priority | Round-1-style comprehensive coverage remained prominent in R4 controls | Author-facing argument/design and decision value take priority; retain compact controlling-read/gap ledger | Author process correction | All R4/R5 prompts, packets and audits | Reject dossier-scale rediscovery or governance-heavy outputs |
|
||||
| 2026-08-10 | Section cadence | Sections researched/frozen individually | Present two detailed packets together; after author feedback launch two scoped continuation agents in parallel; coordinator second-audits both | Author process decision | Remaining R4 sections and R5 handoffs | Audit pair interfaces, duplication, ownership and accepted feedback |
|
||||
| 2026-08-10 | Continuation context | Agents assembled context from broad controls/dossiers | Coordinator minimum packet freezes exact reads, feedback, materials/readiness, decision, exclusions, length/share, paragraph jobs, display layout, ownership and questions; targeted Nowledge recall precedes repository verification when relevant | Author process decision | Coordinator, all continuation agents | Memory never overrides artifacts; no whole-project rediscovery |
|
||||
| 2026-08-10 | Kan99b mean-lift sign | `partial pass` with sign/convention described as unresolved | Preserve exact `partial pass`; sign difference is understood as CFD coordinate/force-direction convention, not unresolved physics or suspected wrong rotation | Author correction and qualification interpretation | Section 2, controls, future prose/captions | Remove unresolved-sign wording while preserving dossier label |
|
||||
@@ -0,0 +1,49 @@
|
||||
# Round 4/5 Section Agent Prompt
|
||||
|
||||
Use only after the coordinator has built the minimum context packet and the author has responded to the paired section discussion.
|
||||
|
||||
## Coordinator minimum context packet
|
||||
- **Unit / round / output:** [SECTION], [R4 FREEZE or R5 DRAFT], `[EXACT FILE]`
|
||||
- **Decision requested:** [ONE BOUNDED OUTPUT DECISION]
|
||||
- **Exact controlling reads:** [R4 CONTROLS, OWNING DOSSIER, CURRENT AUTHORITIES/MANIFESTS]
|
||||
- **Accepted author feedback:** [DATED DECISIONS; distinguish accepted, provisional, open]
|
||||
- **Available materials:** [EXACT TEXT, DATA, SCRIPT, FIGURE/PANEL OR TABLE INVENTORY + `text-ready`/`artifact-ready`/`dependency-gated`/`blocked`]
|
||||
- **Target size:** [WORDS/PAGES] and [PERCENT SECTION/MANUSCRIPT SHARE]
|
||||
- **Paragraph jobs:** [P1 JOB → P2 JOB → ... → CLOSING BRIDGE]
|
||||
- **Concrete display layout:** [FIGURE/TABLE/EQUATION COUNT, PANEL GEOMETRY, PLACEMENT, EACH ARGUMENTATIVE JOB AND FALLBACK]
|
||||
- **Ownership:** imports [EXACT]; exports [EXACT]; prohibited duplication [EXACT]
|
||||
- **Exact questions:** [FINITE LIST ANSWERABLE FROM PACKET]
|
||||
- **Exclusions:** [OUT-OF-SCOPE SOURCES, CLAIMS, RUNS, EDITS, DISPLAYS]
|
||||
|
||||
## Prompt
|
||||
You are the tightly scoped **[R4 SECTION-FREEZE / R5 SECTION-DRAFT]** continuation agent for **[UNIT]** in *Interpreting learned hydrodynamic cloaking and illusion in a fluidic pinball*.
|
||||
|
||||
Start with targeted Nowledge `memory_search` for prior decisions relevant to the exact questions; use thread recall only when the exact earlier discussion is needed. Then verify operative decisions against the packet's repository authorities. Memory is decision/history navigation and never overrides current authorities, manifests, code/configuration or immutable artifacts.
|
||||
|
||||
Read the minimum context packet above and only its exact controlling reads. Do not rediscover the whole project or recreate Round-1 coverage. Use repository tools for internal evidence. Use Undermind only for a stated external-literature gap and only after `get_orientation` then `list_workspaces`; literature cannot establish internal results. For any authorized read-only derivation/script, first discover the designated conda environment from repository authority or environment listings and record it—never guess. Do not run CFD, training, artifact generation or expensive analysis unless separately authorized.
|
||||
|
||||
### Deliverable priorities
|
||||
1. Answer the section question with an evidence-bounded recommendation that gives the author a concrete decision.
|
||||
2. Honor target length/share and make every paragraph perform its assigned argumentative job.
|
||||
3. Specify the concrete display layout: panel arrangement, owner, source arrays/artifacts, placement, argumentative job, caption ceiling, readiness and fallback. Inventory is not artifact readiness.
|
||||
4. Respect result/terminology ownership, imports/exports, bridges and duplication deletions.
|
||||
5. State green/amber/red claim wording and preserve chain, configuration, Reynolds convention, role, comparator, metric/estimand, ROI/window and artifact boundaries.
|
||||
6. Include only a compact coverage ledger for the controlling reads and material gaps: `read`, `mapped`, `not-covered`, `external-inaccessible`, with verification/artifact-audit state where relevant.
|
||||
7. Separate evidence-governed conclusions from author choices. For each choice give recommendation, consequence and safe default.
|
||||
|
||||
### Fixed boundaries
|
||||
- Positive SR is Kármán/cloak-only; Illusion SR is historical/negative; CCD is descriptive/noncausal and uses `action-correlated correction-field modes`; OID is claim-free; steady is contextual; experiment is qualitative/provenance-gated.
|
||||
- Internal Legacy/V5 chains remain separate and provenance-only in manuscript prose. Case-specific metrics and near-colliding estimands are never pooled or called replication, validation, agreement or triangulation without exact authority.
|
||||
- Kan99b K2 remains `partial pass` under the qualification authority. The mean-lift sign difference is the CFD coordinate/force-direction convention, not unresolved physics or suspected wrong rotation; preserve the overall label and state its reason accurately.
|
||||
- A first pass cannot freeze itself. Return `researching`, `discuss`, `revise` or `blocked`; only the coordinator's second audit may set `frozen` or authorize R5 prose.
|
||||
|
||||
### Output order
|
||||
1. Recommended decision and bounded answer.
|
||||
2. Target size/share and paragraph-job map.
|
||||
3. Concrete display/equation layout and material readiness.
|
||||
4. Ownership, imports/exports, bridge and deletions.
|
||||
5. Claim ceiling, choices and exact answers.
|
||||
6. Compact coverage/gap ledger.
|
||||
7. Disposition.
|
||||
|
||||
Edit exactly `[EXACT FILE]`; do not edit R1–R3 dossiers, evidence packages, immutable artifacts, archives, controls or other sections.
|
||||
@@ -0,0 +1,47 @@
|
||||
# Round 4/5 Quality Gate
|
||||
|
||||
Round 4 freezes author-facing arguments and section designs; Round 5 drafts from those freezes. Neither round creates evidence. Decision value and manuscript design outrank procedural volume.
|
||||
|
||||
## Entry gate — coordinator packet
|
||||
- [ ] Exact unit, round, output decision and file are named.
|
||||
- [ ] Exact controlling reads and accepted dated author feedback are listed.
|
||||
- [ ] Available material/artifact inventory distinguishes text-, data-, script-, panel- and dependency-readiness.
|
||||
- [ ] Target words/pages and section/manuscript share are stated.
|
||||
- [ ] Paragraph jobs and closing bridge form an argument, not an inventory.
|
||||
- [ ] Concrete display layout gives count, panel geometry, placement, job, owner, source, readiness and fallback.
|
||||
- [ ] Imports, exports, exclusions and exact bounded questions are explicit.
|
||||
|
||||
A subagent failing this gate is not launched to rediscover the project.
|
||||
|
||||
## Process gate — continuation and routing
|
||||
- [ ] When prior decisions matter, agent begins with targeted Nowledge memory/thread recall, then verifies against repository authorities; memory never overrides artifacts.
|
||||
- [ ] Repository tools handle internal evidence. Undermind is literature-only after `get_orientation` and `list_workspaces`.
|
||||
- [ ] Read-only scripts/derivations use a designated conda environment discovered from authority/listing; no environment is guessed.
|
||||
- [ ] No CFD, training, artifact generation or expensive analysis ran without authorization.
|
||||
- [ ] Coverage ledger is compact and limited to controlling reads/gaps (`read`, `mapped`, `not-covered`, `external-inaccessible`).
|
||||
|
||||
## Section design/freeze gate
|
||||
- [ ] Recommendation answers the exact author-facing question and explains consequence/trade-off.
|
||||
- [ ] Length/share, paragraph progression, display cadence and evidence density are credible.
|
||||
- [ ] Claims retain owner, chain/configuration, Reynolds convention, role, comparator, metric/estimand, ROI/window and authority.
|
||||
- [ ] Imports stay brief; exports and deletion owners prevent duplication.
|
||||
- [ ] Displays/equations state what they establish and do not establish; text readiness is separate from artifact readiness.
|
||||
- [ ] Author choices remain `discuss`/`blocked` or receive an accepted decision; safe defaults are not falsely frozen.
|
||||
- [ ] Kan99b K2 stays exactly `partial pass`; the mean-lift sign difference is attributed to CFD coordinate/force-direction convention, never unresolved physics or suspected wrong rotation.
|
||||
- [ ] Fixed SR/CCD/OID/steady/experiment and cross-chain/estimand cautions pass.
|
||||
|
||||
## Two-section cadence and second audit
|
||||
1. Coordinator presents two detailed packets together.
|
||||
2. Author responds to both; coordinator records accepted, provisional and open feedback.
|
||||
3. Coordinator launches two tightly scoped continuation agents in parallel.
|
||||
4. Coordinator re-reads both outputs and controlling authorities, checks interfaces/duplication/claims/displays/choices, and records a second audit.
|
||||
5. Only then may R4 become `frozen` or R5 prose become drafting-ready.
|
||||
|
||||
## Statuses
|
||||
- `researching`: a named controlling read is incomplete.
|
||||
- `discuss`: a bounded author choice is decision-ready.
|
||||
- `revise`: targeted argument/design/interface correction is needed.
|
||||
- `frozen`: coordinator second audit locked the R4 text contract.
|
||||
- `blocked`: required authority, material, role, comparator, decision or conflict resolution is absent.
|
||||
|
||||
Final Round-4 audit checks the paired sections in the whole-manuscript order, ownership, bridges, claim ladder, display/equation inventory, accepted author decisions and Round-5 handoff. Optional omitted panels do not block text when a safe fallback preserves the argument.
|
||||
@@ -0,0 +1,133 @@
|
||||
# Round 4 Gate-A style audit and crosswalk
|
||||
|
||||
**Role:** coordinator audit of `STYLE_01_INTRODUCTION.md` through `STYLE_10_DISCUSSION_CONCLUSION.md`. This is style/architecture control, not manuscript prose, scientific evidence, artifact approval, or a section freeze.
|
||||
|
||||
**Overall Gate-A result: PASS.** All ten style memos pass after the four targeted corrections. This file records the coordinator re-audit only.
|
||||
|
||||
## 1. Scope and complete coverage ledger
|
||||
|
||||
`read` is not independently verified or artifact-audited. `mapped` is not read. External literature is **style evidence only**; internal claims remain controlled by repository authorities, manifests, immutable artifacts, the Round-2 freeze, and audited Round-3 dossiers.
|
||||
|
||||
| State | Inputs | Boundary |
|
||||
|---|---|---|
|
||||
| `read` | All Round-4 controls: `README.md`, `00_MASTER_AGENT_PROMPT.md`, `00_GLOBAL_DECISION_LEDGER.md`, `00_DISPLAY_EQUATION_INVENTORY.md`, `00_QUALITY_GATE.md` | Complete workflow/decision control set; not scientific verification |
|
||||
| `read` | All ten style memos, each in full | Style observations, recommendations, identities/read levels, interfaces and anti-patterns; no memo is internal scientific authority |
|
||||
| `read` | R2 `05_REVISED_STRUCTURE_FREEZE.md`, `06_SECTION_LEVEL_JFM_STYLE_GUIDE.md` | Author-ratified architecture/claims and controlling style synthesis |
|
||||
| `read` | R3 `00_MODULE_INTERFACE_LEDGER.md`, `00_QUALITY_GATE.md`, `11_FINAL_CONSISTENCY_AUDIT.md` | Readiness, ownership, near-collisions, blockers and assembly controls; no recomputation |
|
||||
| `mapped` | R3 Modules `01`–`10`; authorities/manifests/artifacts they name | Scientific detail remains with its owner; no mapped item is promoted to read or verified |
|
||||
| `mapped` | Literature records declared by the memos/R2 guide | Identity/read-level audit only; no new literature search or PDF reading |
|
||||
| `not-covered` | New literature; raw arrays/hashes; CFD, training, SR, CCD or OID recomputation; experiment analysis; figures; manuscript drafting | Outside Gate A; no citation, internal number, panel, or prose inferred |
|
||||
| `external-inaccessible` | Hidden/symlink payloads, model bundles, dense fields/theory arrays, raw experiment media | Authority-led artifact deferral; retained dependent panels/claims remain gated |
|
||||
|
||||
Observed comparator order, cadence and caveat practice come from the memos' declared `full-PDF`, `full-report`, `mapped`, or `abstract/snippet` literature states. Project scope, terminology, results and readiness come only from internal controls and owning authorities.
|
||||
|
||||
## 2. Gate-A result by memo
|
||||
|
||||
| Memo | Result | Exact reason | Targeted correction |
|
||||
|---|---|---|---|
|
||||
| `STYLE_01_INTRODUCTION.md` | **PASS** | Four exact DOI records and declared PDF-version limits; complete controlling reads; observed practice and recommendation separated; section-specific jobs, caveats, bridges and anti-patterns; no project value invented | None; retain publisher-version caveat and freeze Introduction last |
|
||||
| `STYLE_02_CONTRACTS_METRICS.md` | **PASS** | Four exact identities/DOIs/read levels; sampled comparator-first practice is not universalized; complete controls and owner inputs; precise display/equation/qualification ceiling | None; retain selected-observable qualification, not broad validation |
|
||||
| `STYLE_03_CAPABILITY_FIGURE.md` | **PASS** | Four exact identities/DOIs/read levels; non-pooled hierarchy, caption/body split, local failure, readiness and fallbacks are explicit; no quantitative annotation invented | None; panel selection stays `discuss`, blocked content omitted |
|
||||
| `STYLE_04_KARMAN_PERFORMANCE.md` | **PASS** | Four exact identities/DOIs/read levels; scalar-first literature practice is distinguished from authority-led L2-first recommendation; M04/M06 and Legacy/V5 boundaries explicit | None. Its first-pass `revise` label reflects no-self-freeze, not a Gate-A content defect |
|
||||
| `STYLE_05_SR_LAW.md` | **PASS** | Required R2/R3 controls are now read, and panel status, ownership and Kármán/cloak-only scope are reconciled without changing the claim ceiling | None; targeted controlling-read correction verified |
|
||||
| `STYLE_06_FIELD_CCD.md` | **PASS** | Three exact full-PDF identities plus one exact mapped/inaccessible record; read levels are not inflated; mean-before-modes, non-pairing, fixed terminology, POD boundary and M04/M06 separation explicit | None; keep Noack et al. (2016) mapped unless full text is read |
|
||||
| `STYLE_07_CROSS_CASE.md` | **PASS** | The Wang et al. author identity and DOI are corrected; structure, evidence boundaries and scientific recommendation are unchanged | None; targeted bibliographic correction verified |
|
||||
| `STYLE_08_STEADY_THEORY.md` | **PASS** | Four exact DOI-bearing full-PDF records; controls, owner dossier and relevant theory authorities read; observation, outer reference, failed route and recommendation distinct; MFS/mechanism ceilings correct | None; preserve blocked MFS field and one-circle/sign-failure fallback |
|
||||
| `STYLE_09_EXPERIMENT.md` | **PASS** | The display/equation inventory is now read, and experiment displays retain explicit author-supply gates and prose/omission fallbacks without promoting an observation | None; targeted inventory reconciliation verified |
|
||||
| `STYLE_10_DISCUSSION_CONCLUSION.md` | **PASS** | The Deng et al. and Cornejo Maceda author forms now match the controlling identities; architecture and scientific recommendation are unchanged | None; targeted identity corrections verified |
|
||||
|
||||
No memo uses literature to establish an internal result, and no memo invents a new internal number. The corrected defects were limited to exact identity and control-coverage bookkeeping; they do not alter scientific claims.
|
||||
|
||||
## 3. Shared versus section-specific JFM practice
|
||||
|
||||
### Shared observed practice
|
||||
|
||||
1. A target, baseline, reference role or simpler controller is legible before interpretation; extensions renew comparators locally.
|
||||
2. Evidence escalates from contract to principal result to richer spatial/modal interpretation; strong vocabulary follows dedicated tests.
|
||||
3. One paragraph and one display normally carry one main argumentative job.
|
||||
4. Captions decode roles, conditions and panels; body text states inference, limit and consequence.
|
||||
5. Dangerous limitations sit beside affected results and may be compressed again in Discussion.
|
||||
6. Equations define immediately/repeatedly used objects; figures carry empirical burden; tables compress contracts or comparable rows.
|
||||
7. Bridges pose the unresolved scientific question rather than announce a section number.
|
||||
8. Negative routes remain visible to motivate refinement or define the envelope.
|
||||
|
||||
### Section-specific practice
|
||||
|
||||
- **Introduction:** compact funnel; clustered citations; no equations or detailed project results; subsections only for genuine taxonomy.
|
||||
- **Contracts:** dependency order from plant/baseline through objective, metrics and qualification; tables expose contract shifts.
|
||||
- **Capability:** one dominant comparator-bearing case anchors breadth; panel order/area express hierarchy; no common score.
|
||||
- **Performance:** exact comparator and principal result precede interpretation; field and signal estimands stay separate.
|
||||
- **SR:** variables, objective and deployment precede law; execution arbitrates; deletion is not causal ablation.
|
||||
- **Field/CCD:** mean/scalar estimand precedes residual/modes; POD is a same-object boundary, not a superiority foil.
|
||||
- **Cross-case:** change one contract dimension where possible; renew comparator; label visual-only and negative evidence locally.
|
||||
- **Steady/theory:** late context is narrower than a second climax; attractive theory and failed sign/closure stay adjacent.
|
||||
- **Experiment:** apparatus/acquisition/provenance precede qualitative observation; omit media when provenance fails.
|
||||
- **Discussion/Conclusion:** complete interpretation before the short close; synthesize in evidence order; future work names decisive tests.
|
||||
|
||||
## 4. Global Round-4 rules to freeze
|
||||
|
||||
1. **Comparator-first:** bind target, controlled role, physical zero/baseline, chain and estimand before inference.
|
||||
2. **Evidence escalation:** named capability → Legacy Kármán independent field evaluation → bounded Kármán/cloak-only interpretation; cross-cases do not inherit deeper levels.
|
||||
3. **Local caveats:** put mask/window, acquisition, causal, transfer, theory and provenance limits with the affected sentence/caption.
|
||||
4. **One job per paragraph:** delete package chronology, inventory prose and panel narration.
|
||||
5. **Ownership:** Contracts owns reusable definitions; Capability Fig. 1/progression; Performance detailed Legacy L2/DTW; SR deployment/law/deletion; Field/CCD corrected means/residual/modes/POD; Cross-case changed-contract extensions; Steady/theory V5 endpoint/sign/failure; Experiment provenance/observation; Conclusion synthesis/limits/tests and no new evidence.
|
||||
6. **Caption/body division:** caption states chain, case/role, comparator, metric/coordinate, source and ceiling; body explains inferential consequence.
|
||||
7. **Display discipline:** inventory never means existence. Every retained display needs owner, job, source, readiness, caption ceiling and fallback; text and artifact readiness stay separate.
|
||||
8. **Equation discipline:** freeze reusable notation once; do not duplicate owner equations; Conclusion has none.
|
||||
9. **Bridges:** gap → contracts → capability → independent Kármán evaluation → law complexity → field accounting → non-transfer → steady context → qualitative deployability → limits/tests.
|
||||
10. **Claim ceilings:** positive SR Kármán/cloak only; Illusion SR historical/negative; CCD descriptive/noncausal using exactly **action-correlated correction-field modes**; OID claim-free; steady contextual; experiment qualitative, open-loop and provenance-gated.
|
||||
11. **Notation/chain:** preserve `Kármán`, `Re_D`, `Re_code`, spatial L2 versus Level/Stage, DTW orientation, action order/sign, ROI/mask/window and percentages. Legacy/V5 never pool, continue, validate or imply equivalence.
|
||||
12. **Metric boundary:** periodic mean/phase, event-relative, mean-only, steady profile/`E_inf_vector`, and native/normalized DTW are non-equivalent. M04/M06 means are **different estimands**; SR action magnitude/V5 spin are **related contextual diagnostics** only.
|
||||
|
||||
## 5. Author decisions for macro discussion (maximum five)
|
||||
|
||||
1. **Figure density and main/supplement split.** Recommend the minimum argumentative main set; move derivations, qualification matrices and optional diagnostics to supplement; omit before crowding. Decide final Fig. 1 panels and optional Re400/limit markers.
|
||||
2. **Gated displays.** Recommend omitting standardized PPO/SR/zero, quantitative Vortex/Erase, and MFS field panels unless exact gates close. Decide whether CCD operator/POD/evidence-map displays do distinct work.
|
||||
3. **Experiment inclusion.** Retain a short author-owned subsection only if provenance supports a named qualitative observation; otherwise use a limit/interface sentence and omit media. Quantitative validation remains prohibited.
|
||||
4. **Discussion/Conclusion split.** Recommend separate integrated Discussion and short Conclusion because multiple non-equivalent chains require synthesis before compression; safe fallback is one top-level section with clearly distinct functions.
|
||||
5. **Introduction freeze and title.** Freeze Introduction last; retain the provisional title unless reopened. Any title change triggers terminology/claim audit.
|
||||
|
||||
## 6. Section-specific crosswalk
|
||||
|
||||
| Section | Recommended sequence | Primary display contract | Ownership | Main fallback |
|
||||
|---|---|---|---|---|
|
||||
| Introduction | concept/boundary → difficulty → capability/gap → evidence-chain need → scope/bridge | No owned empirical display; optional prospective Fig. 1 reference | Full ladder, concept, gap, scope | No display; five-job funnel |
|
||||
| Contracts | comparison need → plant/chain → action/sign → observation/objective → metrics/roles/windows → qualification | Required contract table + minimal schematic; reusable metric equations | Reusable definitions and qualification | Compact table/schematic; supplement detail |
|
||||
| Capability | viewing contract → steady seed → dominant Kármán → bounded extensions → limit/deep-case selection | Kármán-dominant distinct-contract Fig. 1 | Progression, evidence levels, deep-case choice | Source-bound panels only or prose |
|
||||
| Performance | exact Legacy contract → mean/phase L2 → DTW → retention boundary → SR question | L2-first display; DTW subordinate; V5 inset optional | Detailed Legacy result and retention boundary | Compact table |
|
||||
| SR | surrogate boundary → deployment → law → execution → deletion → field question | Required deployment map/law/deletion; direct role panel gated | Executable law, imposed structure, arbitration, tested deletion | Equations + deletion table |
|
||||
| Field/CCD | four roles → means/ledger → limit → residual → CCD → POD boundary | Mean before modes; residual equation required; CCD operator optional | Corrected means, residual, descriptive modes, POD boundary | Scalar ledger + residual equation |
|
||||
| Cross-case | framing → event-relative → non-zero target → blocked/failure route → envelope | Distinct-contract display; quantitative Vortex/Erase gated | Comparator renewal, evidence levels, negative routes | Case-contract table + sourced visuals |
|
||||
| Steady/theory | return → V5 endpoint → bounded description → outer reference/failure → export | Sign contract and failure required; endpoint gated; MFS blocked | Separate-chain endpoint/sign/context/theory failure | Sign schematic + endpoint table + trace |
|
||||
| Experiment | purpose → apparatus/provenance → qualitative comparison → limit/test | Author-supply-gated media; no quantitative panel/equation | Frank-owned provenance and observation | Prose interface or omission |
|
||||
| Discussion/Conclusion | ladder → interpretation → envelope → tests; short close | Required limitations table; map optional; no new equation/result | Integrated answer, limits, tests | Discussion prose + limitations table |
|
||||
|
||||
## 7. Macro-discussion input
|
||||
|
||||
### Author-ratified decisions carried forward
|
||||
|
||||
- **Assembled order:** Introduction → contracts → capability → Kármán performance → SR → field/CCD → cross-case → steady/theory → experiment → Discussion/Conclusion.
|
||||
- **Freeze order:** contracts → capability → Kármán performance → SR → field/CCD → cross-case → steady/theory → experiment → Discussion/Conclusion → Introduction last.
|
||||
- **Claim ladder:** named cross-task capability; Legacy Kármán Re100-only independent field matching; bounded Legacy Kármán/cloak-only control-skeleton interpretation.
|
||||
- **Planning shares:** Introduction 10%, contracts 13%, capability 11%, Kármán performance 12%, SR 17%, field/CCD 17%, cross-case 10%, steady/theory 6%, experiment/limitations/conclusion 4%. These are carried R2 allocations, not new measurements.
|
||||
- **Physical chronology:** steady discovery → Kármán → Vortex → Illusion/Erase → qualitative experiment, with steady as an early seed and later contextual return.
|
||||
|
||||
### Coordinator recommendation; not author-ratified
|
||||
|
||||
1. Preserve both orders and keep SR, field/CCD, capability and cross-case as distinct ownership units.
|
||||
2. Protect the Kármán block as the analytical centre; cross-case and steady define breadth/boundaries, not competing interpretation.
|
||||
3. Apply the claim ladder sentence by sentence; interpretation remains Kármán/cloak-only and noncausal.
|
||||
4. Retain the minimum display inventory: contract table/schematic; Kármán-dominant Fig. 1; Kármán performance; SR law/deletion; mean then residual/CCD; bounded cross-case display/table; steady sign/endpoint/failure; provenance-gated experiment if available; limitations table. Optional operators, insets, maps and blocked panels are drop-first.
|
||||
5. Retain the minimum equation inventory: reusable definitions in Contracts; SR deployment/law; scalar identity only if retained plus residual in Field/CCD; event equation only with a retained quantitative result; steady definition only if not imported plus one outer-reference boundary. Full derivations go to supplement.
|
||||
6. Keep result ownership exclusive: previews name conclusions without values; owners carry exact comparisons; Discussion compresses without replay.
|
||||
|
||||
These recommendations remain `discuss` until recorded as author decisions; they do not become frozen by appearing here.
|
||||
|
||||
## 8. Handoff and quality gate
|
||||
|
||||
1. Present the five decision bundles to the author for macro discussion; record decisions or safe omissions in the global ledger. Do not falsely freeze unresolved choices.
|
||||
2. Freeze architecture only when ownership, bridges, ladder, notation, space and display/equation inventory agree. Optional panels may remain gated; retained required panels with missing source or roles remain unavailable.
|
||||
3. For section freezes, reread the owning R3 dossier and current authorities, not only this crosswalk; run local near-collision and caption-ceiling checks and require coordinator second audit.
|
||||
4. Before Round 5, audit `Kármán`, Legacy/V5, `Re_D`/`Re_code`, spatial L2 versus Level/Stage, DTW orientation, action order/sign, ROI/mask/window, percentages, equation ownership, duplicate detail and whitespace.
|
||||
|
||||
**Gate-A handoff:** the ten-memo style set is **PASS** and ready for macro discussion through the five author-decision bundles. Optional display dependencies and later section-freeze checks retain their recorded gates. No manuscript prose, literature, internal number, artifact or panel is created by this audit.
|
||||
@@ -0,0 +1,24 @@
|
||||
# Round 4 — Top-Down Decisions and Section Freezes
|
||||
|
||||
## Purpose
|
||||
Round 4 converts the R2 architecture and audited R3 dossiers into author-facing section arguments, concrete display designs and frozen R5 drafting packets. It is not manuscript prose, new evidence, artifact generation, CFD or training. Round 4 should be lighter than the dossiers: preserve traceability in a compact ledger while spending most effort on argument, ownership, displays and decisions.
|
||||
|
||||
## Authority and tool routing
|
||||
Current repository authorities, manifests, code/configuration and immutable artifacts override summaries and memory. A continuation agent begins with targeted Nowledge memory/thread recall when prior decisions matter, then verifies against exact repository authorities. Nowledge is for decisions/history; repository tools are for internal evidence; Undermind is for external literature only after `get_orientation` and `list_workspaces`. Read-only derivations/scripts use only a designated conda environment discovered from current authority or environment listings. Never guess an environment or run CFD/training without authorization.
|
||||
|
||||
## Coordinator workflow
|
||||
1. Maintain the global decision ledger, whole-manuscript ownership and author feedback.
|
||||
2. Build a minimum context packet for each section: exact controlling reads, accepted feedback, available material/readiness inventory, requested output decision, exclusions, target length/share, paragraph jobs, concrete display layout, ownership and exact questions.
|
||||
3. Present two detailed section packets together to the author.
|
||||
4. After feedback, launch two tightly scoped continuation agents in parallel. Agents use the packet and targeted checks; they do not rediscover the project.
|
||||
5. Perform a coordinator second audit across both outputs before freezing either section.
|
||||
6. Repeat by section pairs; freeze the Introduction last, then run the whole-manuscript Round-4 audit.
|
||||
|
||||
## Controlling inputs
|
||||
- `ROUND2_TOP_DOWN/05_REVISED_STRUCTURE_FREEZE.md` and `06_SECTION_LEVEL_JFM_STYLE_GUIDE.md`.
|
||||
- Relevant `ROUND3_BOTTOM_UP` owning dossiers, interface ledger, quality gate and final consistency audit.
|
||||
- Current repository authorities/manifests named by the packet.
|
||||
- This directory's master prompt, quality gate, global ledger and display/equation inventory.
|
||||
|
||||
## Round-5 handoff
|
||||
Each frozen packet carries its bounded answer, target length/share, paragraph jobs, exact displays/equations, ownership/imports/exports, claim ceiling, accepted author decisions, exclusions, compact coverage/gap ledger and separate text/artifact readiness. Round 5 drafts only from that packet and its controlling authorities; it does not silently reopen decisions or fill evidence gaps.
|
||||
@@ -0,0 +1,245 @@
|
||||
# Round 4 frozen section architecture — Introduction
|
||||
|
||||
## 1. Status and scope
|
||||
|
||||
**Status:** `frozen` for Round-4 architecture; **not final manuscript prose**, not a citation-priority audit, and not scientific or artifact verification. Cultural-hook and sentence-level citation wording remain drafting choices and do not reopen the architecture.
|
||||
|
||||
**Scope:** architecture for an unsplit five-paragraph Introduction. This file controls paragraph jobs, literature roles, terminology choices, evidence boundaries, ownership, and the bridge into the contracts section. It does not draft publishable paragraphs, define reusable equations, narrate results, or make any panel publication-ready. Reader-facing configuration labels are frozen below from inlet and lateral-wall boundary conditions; additional solver, outlet, geometry, and action differences remain owned by configuration-local contracts.
|
||||
|
||||
**Bounded recommendation:** begin with the human idea of concealment and deception, move immediately to why direct linear/transformation cloaking does not transfer trivially to separated finite-Reynolds-number Navier–Stokes flow, introduce the fluidic pinball as the controlled channel-flow object, formulate force-mediated cloaking as a nonstationary feedback problem, and preview a layered realization–evaluation–interpretation design. Close with the narrowed question and three contribution levels, then hand the reader to configuration-local contracts. Do not organize the Introduction around internal chain labels, package history, or defensive nonclaims.
|
||||
|
||||
## 2. Section question and answer ceiling
|
||||
|
||||
**Section question.** Can rotational control of a fluidic pinball in channel flow use force feedback and sparse downstream observations to approach a declared hydrodynamic reference, and what evidence is required to distinguish learned execution, independent field evaluation, and bounded physical interpretation?
|
||||
|
||||
**Answer ceiling.** The Introduction may motivate and announce three nested levels: (1) named learned-control capability under distinct task contracts; (2) independent spatial field matching for the periodic Kármán cloak under the parabolic-inflow/no-slip-wall channel configuration only; and (3) a bounded Kármán/cloak-only control-skeleton interpretation assembled from executable action-law and descriptive field-organization evidence. It may not imply universal invisibility, linear-cloak/Navier–Stokes equivalence, cross-chain continuation, a causal mechanism, transfer, robustness, optimality, priority, positive Illusion symbolic regression, or quantitative experimental validation.
|
||||
|
||||
**Green:** define hydrodynamic cloaking as approach to a declared background and illusion as retargeting toward a declared non-zero signature; motivate force-plus-sparse-observation control; state the three levels without values.
|
||||
|
||||
**Amber:** say that the later layered evidence is designed to test field matching beyond the policy observer/reward space and to support a bounded Kármán/cloak-only organization of action and field evidence.
|
||||
|
||||
**Prohibited red:** all-observer invisibility; direct equivalence between transformation/linear cloaks and finite-Re Navier–Stokes control; force reduction as proof of field matching; causality, necessity, sufficiency, uniqueness, stability, robustness, optimality, universality, transfer, priority, independent triangulation, positive Illusion SR, causal CCD, an OID result, or quantitative experiment validation.
|
||||
|
||||
## 3. Author feedback incorporated
|
||||
|
||||
- **Reader-facing configuration terminology:** use the exact pair `parabolic-inflow/no-slip-wall channel configuration` and `uniform-inflow/free-slip-wall channel configuration`. Do not use historical/current, earlier/revised, or internal chain IDs as reader-facing adjectives. The configurations remain separate because solver, outlet, geometry, actuation, Reynolds convention, roles, and windows also differ; those details remain in contracts. Internal chain IDs (`Legacy`, `V5`) are provenance only.
|
||||
- **Reynolds language:** never use bare `Re100`. Prefer the reader-facing basis $Re_D=U_{ref}D/\nu$, with $U_{ref}$ defined separately for each configuration. For the uniform-inflow/free-slip-wall configuration, the configured uniform inlet speed supplies $U_{ref}$. For the parabolic-inflow/no-slip-wall configuration, repository code shows that configured `U0` is the bulk/mean scale and the profile peaks at approximately $1.5U0$; the authoritative manuscript choice between centreline/maximum and mean/bulk convention, and any exact conversion of inherited labels, remains a contracts-section decision. Equal code labels must never be presented as equal physical Reynolds numbers.
|
||||
- **P1:** open from concealment as longstanding fiction/human fascination; use Harry Potter only with a precise source. Move to hydrodynamic cloaking and then illusion as hiding-to-deception. Radar stealth/deception is optional. End on the finite-Re gap after naming nonlinear advection, separated wakes, dissipation, boundary conditions, active forcing, and observer/target dependence. Delete literature disclaimers.
|
||||
- **P2:** start from the scale gap between shaping an extended Navier–Stokes field and controlling a compact rotating object. Select the pinball for rich nonlinear dynamics, prior rotation knowledge, and broad force/action authority. Distinguish force-target matching from scalar drag minimization. Include Noack lineage and Feng et al. only under exact roles. End with the natural force-mediated control question.
|
||||
- **P3:** bridge to nonstationary feedback; motivate DRL/PPO; treat sparse sensing as application-driven. Force feedback complements sparse observations without novelty or observability claims. Preview objective execution, target-field evaluation, and extractable/checkable action/field organization as reader hooks.
|
||||
- **P4:** use learned realization → independent evaluation → action-side and field-side analysis. Explain each layer through exact closed-loop ML, sparse/reduced-order, and modal/field precedents. Remove package names and defensive nonclaim lists.
|
||||
- **P5:** close confidently with the narrowed question, three contribution levels, and positive rationale for Kármán depth. Compress asymmetric scope into one sentence and bridge to contracts. Do not lead with limitations, Illusion SR, Erase, experiment caveats, or application disclaimers.
|
||||
- **Experiment:** reserve a short, author-owned qualitative open-loop subsection near manuscript end. Mention it in the Introduction only if structurally useful.
|
||||
- **Scientific cautions retained:** positive SR is Kármán/cloak-only; Illusion SR is historical/negative; CCD is descriptive/noncausal and uses exactly `action-correlated correction-field modes`; OID is claim-free; steady is contextual; experiment is qualitative, open-loop, author-owned, and provenance-gated.
|
||||
|
||||
## 4. Recommended five-paragraph architecture
|
||||
|
||||
### Paragraph 1 — fascination, hydrodynamic cloak, and finite-Re boundary
|
||||
|
||||
**Rhetorical job.** Turn a familiar desire to hide or deceive into a precise fluid-mechanical gap without beginning with methods or internal evidence.
|
||||
|
||||
**Logical moves.**
|
||||
1. Open with concealment/invisibility as a longstanding human and science-fiction fascination.
|
||||
2. Use Harry Potter only as a brief cultural hook if the author approves an exact citable source; otherwise keep the opening generic.
|
||||
3. Move from optical/field cloaking to hydrodynamic cloaking: an object seeks to leave a declared background flow/signature minimally disturbed for a specified observer or comparison.
|
||||
4. Introduce illusion as the natural extension from hiding to deception: replace the expected signature with a declared non-zero target. A compact radar analogy is optional.
|
||||
5. Explain why transformation-based or active linear-field constructions do not transfer directly: nonlinear advection couples scales and states; body boundaries generate separated shear layers and wakes; dissipation changes cancellation logic; boundary conditions and active forcing belong to the plant; success depends on declared observer and target.
|
||||
6. End with the gap: how can an actively forced body in nonlinear channel flow be controlled toward a declared background or target, and how should success be established?
|
||||
|
||||
**Literature roles.** Pendry et al. (2006), Leonhardt (2006), Vasquez et al. (2009), Ma et al. (2013), and Urzhumov & Smith (2012) support cloak/active-cloak genealogy and restricted assumptions. Chen, Shen & Xu (2022) can supply an exact recent porous-medium hydrodynamic example, not finite-Re channel-flow evidence. Harry Potter/radar sources remain optional verification tasks.
|
||||
|
||||
**Internal project imports.** Declared-background and declared-non-zero-target definitions; nonlinear channel-flow framing; no result values or chain IDs.
|
||||
|
||||
**Excluded content.** History catalogue; packages; linear/NS equivalence; universal invisibility; priority; applications list; literature disclaimer.
|
||||
|
||||
**Bridge/output.** The nonlinear-flow gap needs a concrete controllable body whose force and wake signatures are coupled.
|
||||
|
||||
### Paragraph 2 — fluidic pinball and force-mediated cloaking
|
||||
|
||||
**Rhetorical job.** Introduce the fluidic pinball as the object implementing the concept and distinguish the objective from ordinary drag reduction.
|
||||
|
||||
**Logical moves.**
|
||||
1. Start from the scale mismatch: the desired outcome is an extended Navier–Stokes field, whereas control acts through three compact rotating bodies.
|
||||
2. Define a three-cylinder fluidic pinball in channel flow; defer solver, boundaries, Reynolds, geometry, sign, and action details.
|
||||
3. Explain that approaching a declared background may require reducing or reshaping the body's force signature, making force a meaningful signal and intermediary.
|
||||
4. Separate scalar drag minimization from potentially time-dependent, multi-component force-target matching; neither alone proves downstream field matching.
|
||||
5. Establish the Noack/Morzyński lineage: rich nonlinear regimes, three independent rotations, and reduced-order/open-loop/closed-loop/ML control history.
|
||||
6. Add exact rotational-force precedent: Feng et al. (2023) controls force extrema and references of a rotating pinball using DRL. This establishes nearby precedent and accessible authority, not this field-cloaking result.
|
||||
7. Explain selection by rich nonlinear dynamics, prior rotation knowledge, and broad accessible force/action range in a compact MIMO plant.
|
||||
8. End naturally: can rotations use force-mediated feedback to make the pinball approach the declared hydrodynamic background while preserving a checkable downstream objective?
|
||||
|
||||
**Literature roles.** Noack & Morzyński toolkit lineage; Deng et al. (2020) for dynamics; Cornejo Maceda et al. (2021) and Raibaudo et al. (2020) for rotational/ML control; Feng, Wang, Xiang, Jin & Fan (2023) for force extrema/tracking. Group by role.
|
||||
|
||||
**Internal project imports.** Rotating-body object; force plus sparse downstream observations; declared references; channel-flow wording.
|
||||
|
||||
**Excluded content.** Bare `Re100`; solver/wall details; action signs; drag reduction as cloaking; force match as field match; unverified claims about Dixia Fan.
|
||||
|
||||
**Bridge/output.** The force-mediated objective becomes a feedback problem when the reference is nonstationary and the full field is not observed online.
|
||||
|
||||
### Paragraph 3 — nonlinear feedback, PPO, and three analytical hooks
|
||||
|
||||
**Rhetorical job.** Motivate the learned controller and sparse execution interface, then convert achieved control into three scientific questions.
|
||||
|
||||
**Logical moves.**
|
||||
1. Acknowledge pinball/control precedent while identifying the harder feature: nonstationary reference following with coupled force and sparse downstream information is less intuitive than steady actuation or scalar optimization.
|
||||
2. Motivate DRL for high-dimensional nonlinear closed-loop decisions without a tractable controlled-flow model; identify PPO locally as the selected optimizer, not a contribution by itself. If PPO is explained, say only that clipped policy updates regularize step size relative to the previous policy, supported by the original PPO paper or a suitable review.
|
||||
3. Present sparse downstream sensing as application-driven limited remote observation.
|
||||
4. Explain that body-integrated force feedback complements the sparse downstream signature. Avoid novelty, optimality, sufficiency, and observability claims.
|
||||
5. Pose three hooks: does execution achieve its declared objective; does the flow match the target field beyond observer/reward space; what recurring action and field organizations can be extracted and checked?
|
||||
6. End by requiring layered analysis rather than treating reward convergence as the scientific answer.
|
||||
|
||||
**Literature roles.** Rabault et al. (2019), Paris et al. (2021), Li & Zhang (2022), Xia et al. (2024), and Wang et al. (2024) establish DRL/partial-observation precedents. Schulman et al. (2017), “Proximal Policy Optimization Algorithms,” arXiv:1707.06347, is the original PPO citation; Garnier et al. (2021) is a verified general review. The Introduction need not tutorialize PPO.
|
||||
|
||||
**Internal project imports.** Force plus sparse velocities; independent field evaluation; bounded Kármán-specific interpretation.
|
||||
|
||||
**Excluded content.** Any claim that PPO mathematically guarantees gradient descent, monotonic improvement, convergence, or learning success; generic AI claims; sparse sensing as deficiency; priority; optimal sensing/full observability; reward/DTW as field equivalence; training or architecture details; negative-limit list. PPO is motivation, not a result or contribution.
|
||||
|
||||
**Bridge/output.** The three hooks determine the realization, independent-evaluation, and interpretation layers.
|
||||
|
||||
### Paragraph 4 — layered approach and why each layer exists
|
||||
|
||||
**Rhetorical job.** Preview the scientific method as a dependency chain, with a reason for each layer.
|
||||
|
||||
**Logical moves.**
|
||||
1. **Realize learned control:** establish what the closed-loop policy can do under named task contracts; capability precedes interpretation.
|
||||
2. **Evaluate independently:** compare registered downstream fields with target and physical-reference roles outside the online reward; separate execution-space success from field-space evidence.
|
||||
3. **Analyze from the action side:** check a reduced policy through executable closed-loop surrogate/deletion tests; ask whether compact organization survives deployment, not merely offline fitting.
|
||||
4. **Analyze from the field side:** separate persistent and dynamic contributions, organize the residual through descriptive action correlation, and compare conventional modal structure; connect organization without causality.
|
||||
5. Connect prior closed-loop ML control to layer 1, sparse/reduced-order reconstruction to fuller-state interpretation, and modal analysis to layer 4. Literature motivates roles, not internal findings.
|
||||
6. End attractively: one compact controlled object, sparse executable feedback, independent field testing, and paired action/field interpretation under explicit contracts.
|
||||
|
||||
**Literature roles.** Gautier et al. (2015) and Cornejo Maceda et al. (2021) for deployed ML laws; Loiseau et al. (2018) for sparse sensing to fuller-state representation; Deng et al. (2020) and verified modal/reduced-order records for field interpretation.
|
||||
|
||||
**Internal project imports.** Three-level ladder; spatial L2 owner downstream; executable SR; corrected four-role mean comparison; separately centred non-paired residual; `action-correlated correction-field modes`; POD boundary.
|
||||
|
||||
**Excluded content.** Packages; formulas/values/windows/masks/panels; independent triangulation; mechanism; causal/unique/superior CCD; OID result; defensive inventory.
|
||||
|
||||
**Bridge/output.** The approach prepares the narrowed question and explains why one case receives depth.
|
||||
|
||||
### Paragraph 5 — narrowed question, contribution levels, and contracts bridge
|
||||
|
||||
**Rhetorical job.** Close calmly with the paper question, three levels, Kármán-depth rationale, and the next definitions.
|
||||
|
||||
**Logical moves.**
|
||||
1. Ask whether force-plus-sparse-observation rotational control under the separately declared channel-flow contracts can realize hydrodynamic cloak/illusion objectives, and how far the periodic Kármán cloak under the parabolic-inflow/no-slip-wall configuration can be evaluated and interpreted.
|
||||
2. Name: learned-control capability across distinct tasks; independent spatial field matching for the named periodic Kármán cloak; bounded Kármán/cloak-only action-and-field control-skeleton interpretation.
|
||||
3. Explain Kármán depth positively: it has the richest aligned intersection of policy execution, independent fields, executable action-law tests, corrected mean/dynamic comparisons, and descriptive field/modal organization.
|
||||
4. Use one asymmetric-scope sentence: cross-case, steady, and experiment material provides breadth, context, or qualitative feasibility but does not inherit Kármán interpretation.
|
||||
5. Mention the author-owned qualitative experiment only if navigation requires it.
|
||||
6. End by requiring the next section to define channel plants, boundary/solver configurations, actions, observations, configuration-local Reynolds conventions, roles, comparators, metrics, regions, and windows.
|
||||
|
||||
**Literature roles.** Normally light/uncited; contributions come from internal authorities. A capability-figure reference is optional and must not narrate panels.
|
||||
|
||||
**Internal project imports.** Full ladder; Kármán-depth rationale; asymmetric scope; contracts ownership.
|
||||
|
||||
**Excluded content.** Limitations inventory; negative Illusion SR story; Erase blockers; application disclaimers; experiment caveat list; values; package labels; section-number ending.
|
||||
|
||||
**Bridge/output.** Gap/question → explicit plant, boundary, solver, Reynolds, role, comparator, metric, ROI, and window contracts.
|
||||
|
||||
## 5. Frozen configuration terminology and Reynolds naming policy
|
||||
|
||||
### Reader-facing configuration labels
|
||||
|
||||
**Frozen pair:** `parabolic-inflow/no-slip-wall channel configuration` and `uniform-inflow/free-slip-wall channel configuration`. Repository configuration and kernel evidence confirm the first uses a parabolic inlet with bounce-back lateral walls (the no-slip realization) and the second uses a uniform inlet with free-slip lateral walls. These labels describe the principal reader-facing distinction without implying chronology or continuation. Other solver, inlet-scheme, outlet, geometry, body-boundary, action, and observation differences remain explicit in contracts.
|
||||
|
||||
Internal chain IDs (`Legacy`, `V5`) are retained only for provenance, ownership, and artifact tracing; they are not manuscript adjectives.
|
||||
|
||||
### Reynolds naming policy
|
||||
|
||||
- Prefer the reader-facing definition $Re_D=U_{ref}D/\nu$ and define $U_{ref}$ per configuration, together with diameter $D$, viscosity $\nu$, and any relation to a code label.
|
||||
- For the uniform-inflow/free-slip-wall configuration, $U_{ref}$ is the declared uniform inlet speed.
|
||||
- For the parabolic-inflow/no-slip-wall configuration, source evidence implements a profile with bulk/mean scale `U0` and centreline/maximum approximately $1.5U0$. Contracts must decide which is authoritative for manuscript $U_{ref}$ and state the exact conversion before a concise Reynolds label is used.
|
||||
- Do not use bare `Re100`, `Re=100`, or `Kármán Re100`. Use `the periodic Kármán cloak case` or the exact configuration label until contracts define Reynolds naming.
|
||||
- Equal code labels do not imply equal physical Reynolds numbers; never unify them by label coincidence.
|
||||
- Exact case and chain IDs remain provenance, not reader-facing physical definitions.
|
||||
|
||||
## 6. Candidate citations requiring verification
|
||||
|
||||
`full-text web/preprint` means accessible text was inspected for architecture; it is not scientific verification or artifact audit.
|
||||
|
||||
### Cloaking and cultural opening
|
||||
|
||||
- Pendry, Schurig & Smith, “Controlling Electromagnetic Fields,” *Science* **312** (2006), DOI `10.1126/science.1125907`; Leonhardt, “Optical Conformal Mapping,” *Science* **312** (2006), DOI `10.1126/science.1126493`; Vasquez, Milton & Onofrei, “Active exterior cloaking for the 2D Laplace and Helmholtz equations,” *PRL* **103** (2009), DOI `10.1103/PhysRevLett.103.073901`; Ma et al., “Experiments on active cloaking and illusion for Laplace equation,” *PRL* **111** (2013), DOI `10.1103/PhysRevLett.111.173901`; Urzhumov & Smith, “Flow stabilization with active hydrodynamic cloaks,” *PRE* **86** (2012), DOI `10.1103/PhysRevE.86.056313`. Existing repository state: `full-PDF`; roles limited to genealogy/restricted assumptions.
|
||||
- **Mengyao Chen, Xiangying Shen & Lei Xu**, “Realizing the thinnest hydrodynamic cloak in porous medium flow,” *The Innovation* **3**(4) (2022), 100263, DOI `10.1016/j.xinn.2022.100263`. **Identity resolved; full-text open-access web read.** Exact modern porous-medium example; not finite-Re channel evidence.
|
||||
- **Harry Potter hook:** unresolved as a suitable manuscript citation. Search found literary theses/secondary cultural studies, but no precise high-authority source selected for the intended claim. Citation-search task: choose primary novel, scholarly history of invisibility, or omit. Safe default: generic fascination, no named hook.
|
||||
- **Radar analogy:** not verified. Search only if retained; safe default omission.
|
||||
|
||||
### Pinball, force control, and DRL
|
||||
|
||||
- Nan Deng, Bernd R. Noack, Marek Morzyński & Luc R. Pastur, “Low-order model for successive bifurcations of the fluidic pinball,” *JFM* **884** (2020), A37, DOI `10.1017/jfm.2019.959`. Existing R4 state: direct `full-PDF` manuscript/preprint read.
|
||||
- Guy Y. Cornejo Maceda, Yiqing Li, François Lusseyran, Marek Morzyński & Bernd R. Noack, “Stabilization of the fluidic pinball with gradient-enriched machine learning control,” *JFM* **917** (2021), A42, DOI `10.1017/jfm.2021.301`. Existing state: `full-PDF`.
|
||||
- Cédric Raibaudo, Peng Zhong, Bernd R. Noack & Robert J. Martinuzzi, “Machine learning strategies applied to the control of a fluidic pinball,” *Physics of Fluids* **32** (2020), 015108, DOI `10.1063/1.5127202`. Accessible web full-text record inspected; confirm final accenting in bibliography manager.
|
||||
- **Haodong Feng, Yue Wang, Hui Xiang, Zhiyang Jin & Dixia Fan**, “How to control hydrodynamic force on fluidic pinball via deep reinforcement learning,” *Physics of Fluids* **35**(4) (2023), 045137, DOI `10.1063/5.0142949`. **Identity/role resolved; full preprint text and final metadata checked.** Paper controls force extrema/reference tracking. Dixia Fan is a coauthor; accessible CRediT lists conceptualization, funding acquisition, and review/editing. Infer no sole-originator or experimental role.
|
||||
- **Paul Garnier, Jonathan Viquerat, Jean Rabault, Aurélien Larcher, Alexander Kuhnle & Elie Hachem**, “A review on deep reinforcement learning for fluid mechanics,” *Computers & Fluids* **225** (2021), 104973, DOI `10.1016/j.compfluid.2021.104973`. **Exact identity verified; accessible full text inspected.** Optional general review.
|
||||
- Rabault et al. (2019), DOI `10.1017/jfm.2019.62`; Paris et al. (2021), DOI `10.1017/jfm.2020.1170`; Loiseau et al. (2018), DOI `10.1017/jfm.2018.147`; Gautier et al. (2015), DOI `10.1017/jfm.2015.95`; Li & Zhang (2022), DOI `10.1017/jfm.2021.1045`; Xia et al. (2024), DOI `10.1017/jfm.2024.69`; Wang et al. (2024), DOI `10.1017/jfm.2024.333`. Existing style-guide state: `full-PDF`.
|
||||
|
||||
**Policy:** citations support transitions, never internal results or priority. Names are retained only if their sentence-level roles survive drafting.
|
||||
|
||||
## 7. Figure/equation/table/citation contracts for Introduction
|
||||
|
||||
- **Introduction empirical figure:** none; `omit`.
|
||||
- **Prospective capability figure:** Capability-owned; optional/dependency-gated; may foreshadow named breadth/Kármán depth; no pooled score, chain equivalence, positive Illusion SR, or experiment; fallback omit.
|
||||
- **Conceptual workflow:** optional/currently unnecessary; realization → evaluation → action/field analysis; dependency map, never causal graph; fallback P4 prose architecture.
|
||||
- **Equations:** none. Contracts owns Reynolds, action, objective, L2, DTW, role, ROI, and window definitions. No unified Reynolds equation.
|
||||
- **Tables:** none in Introduction. Chain/metric contract table belongs downstream; planning ledgers are not manuscript tables.
|
||||
- **Citations:** P1 compact cloak cluster; P2 pinball/rotation cluster with Feng et al. isolated for force tracking; P3 DRL/partial observation with optional review; P4 closed-loop ML/sparse ROM/field analysis by role; P5 normally uncited.
|
||||
|
||||
## 8. Ownership/import/export and duplication rules
|
||||
|
||||
**Introduction owns:** fascination-to-gap funnel; pinball selection; force-target versus drag versus field distinction; three hooks; full high-level ladder; Kármán-depth rationale; contracts bridge.
|
||||
|
||||
**Imports:** R2/R4 declared-reference concepts and claim ceiling; M03 separate configuration/action/observation/Reynolds/comparator/metric contracts; M01/M10/final-audit ladder and duplication rules; exact literature precedent only.
|
||||
|
||||
**Exports:** background/non-zero target distinction; force-mediated sparse-observation question; realization/evaluation/interpretation separation; configuration-local contract requirement; Kármán as sole deep case without bare Reynolds label.
|
||||
|
||||
**Opening dependency:** cultural hook/adjective choices do not block architecture; use safe defaults.
|
||||
|
||||
**Closing bridge:** define plants and configuration-local contracts before capability.
|
||||
|
||||
**Duplication:** Contracts owns definitions/equations; Capability owns Fig. 1/progression; Performance owns field values/windows; SR owns formulas/deletions; Field/CCD owns corrected means/residual/modes; Cross-case owns extensions/negative routes; Steady owns separate-chain context; Experiment owns provenance/observation; Conclusion restates the ladder once. M04/M06 are different estimands; SR magnitude and steady spin are related contextual diagnostics only—never replication, agreement, validation, or triangulation.
|
||||
|
||||
## 9. Author decisions required
|
||||
|
||||
1. **Cultural opening — recommendation: generic fascination, no Harry Potter.** Named hook is memorable but lacks a selected JFM-suitable citation. Safe default is one clause and immediate hydrodynamic turn.
|
||||
2. **Configuration labels — frozen.** Use `parabolic-inflow/no-slip-wall channel configuration` and `uniform-inflow/free-slip-wall channel configuration`; contracts retain all additional solver/action differences.
|
||||
3. **Mengyao Chen example — recommendation: retain only if P1 needs a recent concrete bridge.** Exact but porous-medium scope may slow the funnel. Safe default compact cluster or omit.
|
||||
4. **Feng et al./Dixia Fan precedent — recommendation: retain P2 paper collectively.** It directly supports force tracking versus drag optimization. Do not foreground individual authorship.
|
||||
5. **Experiment in P5 — recommendation: omit unless navigation is incomplete.** The downstream author-owned qualitative open-loop subsection remains reserved.
|
||||
|
||||
## 10. Coverage ledger
|
||||
|
||||
`mapped` is not `read`; `read` is not independently verified; neither implies artifact audit.
|
||||
|
||||
| State | Source/input | Boundary |
|
||||
|---|---|---|
|
||||
| `read` | R4 `STYLE_01_INTRODUCTION.md`, `00_STYLE_CROSSWALK.md`, `00_GLOBAL_DECISION_LEDGER.md`, `00_MASTER_AGENT_PROMPT.md` | Full repository reads for style/ownership/claims; not scientific verification or artifact audit |
|
||||
| `read` | R2 `05_REVISED_STRUCTURE_FREEZE.md`, `06_SECTION_LEVEL_JFM_STYLE_GUIDE.md` | Full-report reads; architecture and literature-state records |
|
||||
| `read` | R3 M01 `01_INTRODUCTION.md`, M03 `03_CONTRACTS_METRICS.md`, M10 `10_LIMITATIONS_CONCLUSION.md`, `11_FINAL_CONSISTENCY_AUDIT.md` | Full-report reads for architecture/contracts/ladder/audit; no recomputation |
|
||||
| `read` | Chen et al. (2022) open article; Feng et al. (2023) full accessible preprint; Garnier et al. (2021) full accessible preprint/web text | Targeted identity, DOI, role, framing verification; not reproduction |
|
||||
| `mapped` | Exact JFM/cloak/control records in R2/R4, including Pendry, Leonhardt, Vasquez, Ma, Urzhumov, Deng, Cornejo Maceda, Rabault, Paris, Loiseau, Gautier, Li, Xia, Wang | Existing `full-PDF` states accepted as repository records; not reread here |
|
||||
| `mapped` | Remaining active R3 owners and artifact authorities | Ownership/boundary navigation only |
|
||||
| `not-covered` | Exhaustive cloak history; radar review; Harry Potter source adjudication; full PPO bibliography | Unneeded for architecture; retained as citation/author choices |
|
||||
| `not-covered` | Raw fields/models/hashes; CFD/training/SR/CCD/OID recomputation; figures; experiment media | Outside scope; no result, number, or panel inferred |
|
||||
| `external-inaccessible` | Undermind workspace listing/search after orientation | Workspace discovery unavailable in this scoped run; targeted public records supplied identity checks |
|
||||
| `external-inaccessible` | Hidden/symlink project artifacts and raw experiment material | Authority-led deferral; no artifact-dependent Introduction claim |
|
||||
|
||||
## 11. Quality gate and disposition
|
||||
|
||||
- [x] Exactly five paragraph jobs; no final manuscript paragraphs.
|
||||
- [x] P1 reaches the nonlinear finite-Re gap from fascination.
|
||||
- [x] P2 selects the pinball and distinguishes force-target matching from drag.
|
||||
- [x] P3 motivates DRL/PPO, application-driven sparse sensing, and three positive hooks.
|
||||
- [x] P4 supplies realization → independent evaluation → action/field analysis with literature roles.
|
||||
- [x] P5 closes with narrowed question, three levels, Kármán rationale, compact asymmetric scope, and contracts bridge.
|
||||
- [x] Internal chain IDs are provenance-only; the exact boundary-condition configuration pair is frozen for readers.
|
||||
- [x] $Re_D=U_{ref}D/\nu$ is preferred, $U_{ref}$ is configuration-local, no bare `Re100` is allowed, and the parabolic reference/conversion decision is assigned to Contracts.
|
||||
- [x] PPO wording is bounded to clipped-update step-size regularization relative to the previous policy and rejects mathematical guarantees of descent, monotonic improvement, convergence, or success.
|
||||
- [x] Mengyao Chen and Dixia Fan identities/roles checked; Harry Potter/radar remain explicit tasks.
|
||||
- [x] Fixed SR/Illusion/CCD/OID/steady/experiment cautions preserved.
|
||||
- [x] No new internal number, result, priority, mechanism, or panel readiness.
|
||||
- [x] Display/equation/table/citation ownership and fallbacks stated.
|
||||
- [x] Coverage states explicit.
|
||||
|
||||
**Near-collision:** M04 policy field metrics and M06 corrected mean-field accounting remain **different estimands**. Periodic, transient, steady, native/normalized DTW, and field metrics remain non-equivalent. Separate chains are not pooled or called replication, validation, agreement, or triangulation.
|
||||
|
||||
**Disposition: `frozen` for Round-4 architecture.** The five-paragraph logic, configuration labels, Reynolds policy, PPO boundary, evidence ladder, and contracts bridge are settled. This is still not manuscript prose; cultural-hook choice, optional Chen example, experiment mention, and sentence-level citation wording remain drafting choices and do not block or reopen the architecture.
|
||||
@@ -0,0 +1,237 @@
|
||||
# Round 4 frozen architecture — Problem formulation, control framework and evaluation
|
||||
|
||||
## 1. Status
|
||||
|
||||
**Status:** `frozen` for Round-4 architecture; not final manuscript prose, a result dossier, or panel/artifact approval. Run-specific table values, bibliography completion and display production remain dependency-gated without reopening the architecture.
|
||||
|
||||
**Scope and claim ceiling.** Section 2 establishes the physical, numerical, control and evaluation setup needed to decode Fig. 1 and all later comparisons. It may define configurations, methods, roles, objectives, metrics and qualification boundaries. It may not narrate performance, imply that current results were regenerated with the improved workflow, merge the two configurations, or use solver qualification to validate control or interpretation.
|
||||
|
||||
## 2. Reader-facing title and purpose
|
||||
|
||||
**Frozen reader-facing title:** **Problem formulation, control framework and evaluation**.
|
||||
|
||||
This is a methodology/problem-formulation section. Its sequence is configuration → numerical realization → closed-loop control → target/objective → independent evaluation → compact numerical qualification. Reader-facing prose uses `configuration`, `formulation`, `definition`, `protocol`, `procedure`, `framework`, `specified cases/conditions`, and precise physical names. `Contract` remains internal governance vocabulary only.
|
||||
|
||||
**Section question.** What physical and numerical problem is solved, what information and actuation define the learned controller, how are cloak and illusion targets constructed, and how is performance evaluated independently?
|
||||
|
||||
**Bounded answer.** The main text presents one canonical formulation and improved workflow while identifying which existing evidence belongs to the parabolic-inflow/no-slip-wall configuration. It defines signal-level online objectives and field-level independent criteria separately, retains only current CelerisLab solver qualification, and defers implementation history and exhaustive matrices to provenance/supplement material.
|
||||
|
||||
## 3. Author feedback and decisions incorporated
|
||||
|
||||
- Replaced reader-facing `contracts` and `named task` language with academically natural physical/method terminology.
|
||||
- Foregrounded the canonical no-bias, physically scaled, observation-normalized PPO workflow; removed software-development chronology and reader-facing internal chain names.
|
||||
- Preserved scientific provenance: core quantitative SR, CCD and field evidence belongs to the parabolic-inflow/no-slip-wall configuration and is not represented as post-improvement retraining.
|
||||
- Required one geometry figure showing both parabolic/no-slip and uniform/free-slip configurations, fixed pinball/sensor relative positions, and case-dependent disturbance/target size.
|
||||
- Frozen article-wide bulk/mean-based $Re_D$ and exact conversion policy in §7.
|
||||
- Retained D2Q9 MRT LBM, boundaries, curved/moving-body treatment, resolution/time-step reporting, and compact selected-observable qualification in the main text.
|
||||
- Elevated sensor placement, target construction, DRL flow, DTW, registered spatial L2, and mean/phase semantics to explicit subsection jobs.
|
||||
- Preserved all fixed cautions: positive SR is Kármán/cloak-only; Illusion SR is historical/negative; CCD is descriptive/noncausal and uses `action-correlated correction-field modes`; OID is claim-free; steady evidence is contextual and separate; experiment remains qualitative/open-loop/provenance-gated.
|
||||
|
||||
## 4. Recommended subsection architecture and logical sequence
|
||||
|
||||
### 2.1 Flow configurations and numerical method
|
||||
|
||||
**P1 — physical object and coordinates.** Define the three-cylinder fluidic pinball, $D$, the streamwise/transverse coordinates, fixed relative cylinder centres, and independent cylinder rotation. Explain that lateral confinement stabilizes and organizes the wake by bounding cross-stream motion and fixing a reproducible channel environment; do not imply universality or that confinement alone establishes a control mechanism.
|
||||
|
||||
**P2 — two configurations and ownership.** Decode the parabolic-inflow/no-slip-wall and uniform-inflow/free-slip-wall configurations side by side. Define inlet, outlet, lateral boundaries, cylinder treatment and case-dependent upstream disturbance or target body. State early that the parabolic/no-slip configuration supplies the principal Kármán field/SR/CCD analysis, while the uniform/free-slip setting supplies improved-workflow capability and separate steady/contextual evidence.
|
||||
|
||||
**P3 — nondimensionalization and Reynolds number.** Define $U_{ref}$, $D$, $\nu$, $t^*=tU_{ref}/D$, velocities and forces as needed. Use bulk/mean inlet speed for the parabolic profile, with $U_{max}\simeq1.5U_{ref}$, and uniform inlet speed for the uniform profile. Record exact inherited-label conversion without using bare `Re100`.
|
||||
|
||||
**P4 — CFD realization.** State D2Q9 lattice Boltzmann formulation with MRT collision, collision/streaming precision, regularized or parabolic inlet as configuration-specific, outlet treatment, no-slip or free-slip lateral walls, and Bouzidi treatment for curved stationary/moving boundaries. Give grid, diameter resolution, lattice time step or mapping, sampling interval and compact adequacy statement only where current authority supports each value.
|
||||
|
||||
**Display cadence.** Place the two-configuration geometry/sensor schematic beside P1–P2. Put a compact configuration/method table after P4 only if it compresses otherwise repeated definitions; historical implementation differences go to supplement/provenance.
|
||||
|
||||
**Bridge.** Once the physical plants and numerical realization are decodable, define how body rotation and limited wake information close the control loop.
|
||||
|
||||
### 2.2 Rotational actuation, observations and PPO training
|
||||
|
||||
**P1 — actuation in physical variables.** Define the three angular velocities or nondimensional surface-speed commands in fixed body order and sign convention. Present the canonical no-preset-bias mapping and smoothing/persistence only to the degree needed for physical reproduction. Avoid code-level upload/readback order unless it changes the physical command.
|
||||
|
||||
**P2 — observations and sensor placement.** Define body-force components and both velocity components at three circular sensing regions centred at $x/D=10$ and $y/D=0,\pm2$, each of radius $0.25D$. Explain the rationale: downstream enough to carry a developed wake signature, transversely distributed to capture symmetry/deflection and shedding, and sparse enough to represent limited remote sensing. Raibaudo et al. provide the direct $x/D=10$ experimental precedent; Cornejo Maceda et al. and Li et al. support downstream multi-location wake sensing and phase/state information. These sources do **not** establish optimality, sufficiency or observability for the present arrangement.
|
||||
|
||||
**P3 — normalization and policy representation.** Separate solver area/time averaging, fixed physical pre-scaling and online observation whitening. State that the selected policy must travel with its matching frozen normalizer. The canonical formulation has zero preset steady-derived bias; do not expose inherited factors or historical affine maps in main text.
|
||||
|
||||
**P4 — PPO principle and project-specific settings.** In one short explanation, PPO updates a stochastic policy using a clipped surrogate objective that limits changes relative to the preceding policy; it does not guarantee gradient descent, monotonic improvement, convergence or control success. Current training authorities bind the recommended/current defaults to SB3 PPO, sinusoidal activation, two hidden layers of 64 units, rollout length 2048, batch 64, ten epochs, discount $0.995$, and default learning rate $3\times10^{-4}$. These are defaults, not universal selected-run metadata: retained run records include learning rates $10^{-4}$ and $3\times10^{-4}$. Main text reports final canonical choices and scientifically relevant project-specific deviations; exact per-result settings and full optimizer metadata go to supplement/provenance.
|
||||
|
||||
**Design lesson boundary.** The no-bias canonical design avoids injecting a steady solution before learning. Phrase this as a design rationale, not a tested superiority result, unless a same-protocol authority is supplied; any claim that preset bias prematurely constrains learning remains evidence-gated.
|
||||
|
||||
**Bridge.** The controller interface is now fixed; the next subsection defines what signal it is asked to approach and how that request differs from independent field evaluation.
|
||||
|
||||
### 2.3 Target construction and control objective
|
||||
|
||||
**P1 — role definitions.** Define `target/reference`, `controlled`, `physical zero/uncontrolled`, and, only where used later, `constant` as distinct acquisition roles. Roles are not interchangeable and do not travel between configurations merely because labels match.
|
||||
|
||||
**P2 — cloak/background target.** Explain that a cloak objective uses a declared background or zero-body/disturbance reference appropriate to the specified case. Identify the roles needed to create the target, train/control the pinball and evaluate against an independently acquired physical-zero comparator.
|
||||
|
||||
**P3 — illusion/non-zero target.** Explain that illusion uses a separately generated non-zero target signature, including the target-body acquisition where applicable. Target-body size and upstream disturbance/target geometry are case-dependent; no target parameter is generalized silently across cases.
|
||||
|
||||
**P4 — online objective.** Define the reward conceptually as a weighted combination of (i) body-force-to-target discrepancy and (ii) multichannel sensor-to-target discrepancy. Explain case-dependent force aggregation and conditional/similarity mapping in prose; retain exact implementation weights only if required to reproduce a selected run and bind them to that run in the supplement. State explicitly that the online objective is an intermediary wake-signature objective, not a full-field metric.
|
||||
|
||||
**Required schematic.** A DRL flow schematic shows: physical configuration/state → force and sparse velocity observations → normalization → PPO policy → three rotations → CFD evolution, with target signals entering the force and sensor reward terms. Caption ceiling: information/action/reward flow only; no convergence, observability, optimal sensing, field equivalence or causality.
|
||||
|
||||
**Bridge.** Because the reward observes only integrated forces and sparse wake signals, evaluation must ask separately whether signals and registered fields approach the declared reference.
|
||||
|
||||
### 2.4 Independent signal- and field-level evaluation
|
||||
|
||||
**P1 — DTW algorithm and role.** Give DTW meaningful main-text space. Describe a dynamic-programming alignment path through two sampled time series under monotonic ordering and declared endpoint/window constraints. Its advantage is comparative tolerance to local time reparameterization: it can register oscillations that differ in timing and expose amplitude, frequency, phase and waveform discrepancies without privileging a single fixed phase shift. For multiple channels, compute declared per-channel costs/similarities and aggregate with explicit scaling/weights.
|
||||
|
||||
**DTW limits.** DTW is a signal-level comparative diagnostic, not physical delay, causality, field equivalence, observability, or a universal cross-case score. Warping can conceal timing errors if unconstrained; therefore orientation (distance versus similarity), channel normalization, window, path constraints and aggregation must accompany every value. Native and normalized DTW are distinct estimands.
|
||||
|
||||
**P2 — registered spatial L2.** Define the registered ROI, same-case roles, mask/geometry guard, velocity normalization and area weighting before results. Spatial L2 is independent field-level evaluation and is not a pipeline `Level` or `Stage`. Use the current registered ROI, $-6\leq(x-x_s)/D\leq14$ and $|(y-y_0)/D|\leq5$, a $20D\times10D$ geometry-guarded downstream rectangle registered to the sensor plane and centreline. It supersedes the earlier $[-8D,12D]$ candidate. Current artifacts do not persist solver-exact masks, so do not call it an exact common-fluid-mask metric. Use a complete-cycle mean-field error and a comparator-relative form only where target, controlled and physical-zero roles share the same acquisition definition.
|
||||
|
||||
**P3 — phase-resolved and case-specific semantics.** Define periodic phase-resolved error separately from mean-field error. The current eight-slot diagnostic uses independently phased nearest snapshots and is not a repeated-cycle ensemble. Vortex uses event-relative offsets, not phase; Illusion uses its own periodic target contract; Erase is mean-field only if exact roles exist; steady uses late/profile criteria. No pooled efficacy score is licensed.
|
||||
|
||||
**P4 — division of labour.** DTW evaluates sparse multichannel temporal signatures close to the controller's information space. Mean and phase spatial L2 evaluate registered fields beyond that space. Agreement is not assumed; their difference is scientifically useful because signal alignment can coexist with spatial mismatch and vice versa.
|
||||
|
||||
**Main/supplement split.** Main text retains the DTW idea/limits, spatial definitions, registered ROI semantics and compact case-semantics table. Algorithm pseudocode, path constraints, channel-by-channel formulas, mask construction, registration details and sensitivity checks move to supplement.
|
||||
|
||||
**Bridge.** The evaluation definitions license later comparisons but not the numerical solver itself; the selected observables therefore receive a compact configuration-bound qualification.
|
||||
|
||||
### 2.5 Compact CFD qualification
|
||||
|
||||
**P1 — purpose and ceiling.** State that qualification checks whether selected CelerisLab observables are sufficiently resolved under specified benchmark configurations. It does not validate PPO, SR, CCD, target construction, field interpretation or hydrodynamic cloaking.
|
||||
|
||||
**P2 — retained anchors.** Use only the current CelerisLab dossier (`read/full-report`). Kan99b K2 is a `partial pass`: Strouhal number, mean drag and lift/drag fluctuation amplitudes pass their declared bands. The reported mean-lift sign differs because the CFD coordinate/force-direction convention is opposite to the comparison convention; this is not unresolved physics or suspected wrong rotation. Preserve the authority's aggregate `partial pass` label rather than relabelling the dossier. Sah04 S2 is a `pass` for its declared confined-channel Strouhal gate. Only K2 and S2 have retained complete physics runs in the qualification directory; other specified cases are not completed evidence there, and smoke runs are diagnostics only. Report exact values only after direct artifact/manifest binding. Do not state that an older solver was validated.
|
||||
|
||||
**Compact table.** Columns: benchmark/configuration; selected observable; comparison authority; resolution/window; status; local caveat. Main text carries a small result/table; full matrices, hashes, sensitivity, precision and smoke-test diagnostics belong to supplement.
|
||||
|
||||
**Closing bridge.** With configurations, controller, objectives, independent metrics and selected numerical observables defined, Fig. 1 may compare prescribed cases without pooling their evidence or anticipating the Kármán result.
|
||||
|
||||
## 5. Detailed content controls: equations, displays, citations, split and exclusions
|
||||
|
||||
### Required/reusable equations
|
||||
|
||||
1. **Article-wide Reynolds number** (`required`, owner Section 2):
|
||||
\[
|
||||
Re_D=\frac{U_{ref}D}{\nu},\qquad U_{max}\simeq1.5U_{ref}\ \text{for the parabolic inlet}.
|
||||
\]
|
||||
Definition only; exact conversion note accompanies inherited labels.
|
||||
2. **Nondimensional actuation** (`required`, exact sign/order source-gated): use $s_i=\omega_iR/U_{ref}$ or an equivalent physical map. Avoid reproducing obsolete affine maps in main text.
|
||||
3. **Online objective** (`required concept; formula optional`): a force discrepancy plus a sensor discrepancy, with case-specific weighting. Prefer one compact symbolic equation if reused; implementation-specific constants belong to supplement.
|
||||
4. **Mean-field L2** (`required`): area-weighted, $U_{ref}$-normalised velocity-vector error over the registered ROI, with target/control/zero roles declared.
|
||||
5. **Phase-resolved L2** (`required if later reused`): same spatial definition indexed by periodic phase slots; do not imply ensemble averaging.
|
||||
6. **Comparator-relative reduction** (`optional`): retain only if later results use exactly the same role/ROI/mask/window definition.
|
||||
7. **DTW recurrence/path equation** (`optional but recommended`): one compact recurrence or path-cost definition because DTW is uncommon in fluid mechanics; otherwise define rigorously in prose and supplement.
|
||||
|
||||
### Display/table ownership and caption ceilings
|
||||
|
||||
| Item | Owner and job | Status/source | Caption ceiling | Fallback |
|
||||
|---|---|---|---|---|
|
||||
| Two-configuration geometry/sensor schematic | §2.1; decode inlet/walls/outlet, fixed pinball/sensor geometry, case-dependent disturbance/target | `required`; redraw/production and exact geometry-source binding pending | Defines geometry and BCs; does not imply configuration equivalence, optimal sensing or solver-exact scale | Minimal line drawing plus configuration table |
|
||||
| DRL information/action/reward schematic | §2.3; distinguish observation, target, reward and independent evaluation | `required`; conceptual production item | No convergence, observability, optimality, causality or field equivalence | Compact flow diagram in prose/table form |
|
||||
| Configuration/method table | §2.1; compress physical/numerical ownership without history narrative | `required`; exact rows authority-bound | Defines separate configurations; no cross-configuration transfer | Split main compact table and supplement provenance matrix |
|
||||
| Case/metric semantics table | §2.4; separate periodic, event-relative, mean-only and steady estimands | `required`; definition-ready | No pooled score or result | Concise prose list |
|
||||
| Compact CFD qualification table | §2.5; qualify selected CelerisLab observables | `required`; current dossier/artifact bound | Solver-observable qualification only | One paragraph plus supplement pointer |
|
||||
|
||||
### Main text versus supplement/provenance
|
||||
|
||||
**Main text:** two physical configurations; geometry and BCs; article-wide Reynolds convention; essential LBM/MRT and curved-boundary method; physical actuation/observations; canonical normalization/PPO settings; target/reward concept; sensor rationale; DTW and spatial L2 division; case semantics; compact CelerisLab qualification; one minimal provenance statement.
|
||||
|
||||
**Supplement/provenance:** internal chain IDs; old versus current action maps, scaling, bias, normalization and DTW variants; software chronology; complete configs/hashes; full PPO tables; exact reward mappings/weights not needed in the argument; qualification matrices and sensitivities; mask/registration implementation; historical solver details; selected policy/normalizer/target binding.
|
||||
|
||||
**Explicit exclusions:** no performance values except the compact qualification table; no training diary; no claim that PPO converges; no `named task`; no reader-facing `Legacy`/`V5`; no older-solver validation; no target/control/zero role pooling; no force or DTW claim as field matching; no sensor optimality; no confinement-causes-control mechanism; no positive Illusion SR, causal CCD, OID result, steady transfer, or quantitative experiment claim.
|
||||
|
||||
## 6. Canonical method versus evidence provenance policy and rerun gate
|
||||
|
||||
### Main-text policy
|
||||
|
||||
Present the improved workflow as the **canonical method formulation**: physical action nondimensionalization, zero preset steady-derived bias, explicit normalization layers, matched policy/normalizer, current CelerisLab configuration and concise project-specific PPO settings. Do not write a Legacy-versus-V5 comparison in the scientific narrative.
|
||||
|
||||
### Compact provenance statement
|
||||
|
||||
Use one short neutral main-text sentence: “The deep analyses reported later use data generated with an earlier implementation under the parabolic-inflow/no-slip-wall physical configuration; implementation differences do not change the physical problem and are documented in the supplement.”
|
||||
|
||||
Detailed implementation differences and the fact that no complete canonical rerun was performed belong in supplement/provenance. Do not say that the earlier results were generated by the canonical pipeline or that equivalence, reproduction, or a matched rerun was demonstrated.
|
||||
|
||||
## 7. Terminology and notation freeze
|
||||
|
||||
- **Section title:** `Problem formulation, control framework and evaluation`.
|
||||
- **Reader-facing configurations:** `parabolic-inflow/no-slip-wall channel configuration`; `uniform-inflow/free-slip-wall channel configuration`.
|
||||
- **Provenance only:** internal `Legacy` and `V5` identifiers; never reader-facing adjectives.
|
||||
- **Cases:** use `periodic Kármán cloak case`, `Vortex case`, `Illusion case`, `Erase case`, or exact physical target/condition; never `named task` or bare `Re100`.
|
||||
- **Article-wide Reynolds policy:** $Re_D=U_{ref}D/\nu$. $U_{ref}$ is the uniform inlet speed in the uniform configuration and the bulk/mean inlet speed in the parabolic configuration; $U_{max}\simeq1.5U_{ref}$ for that profile. Where authoritative two-dimensional code labels used $2D$ as reference length, manuscript $Re_D$ is exactly one-half of the inherited code label. Equal labels never imply equal physical Reynolds numbers.
|
||||
- **Execution:** `wake-signature matching using forces and sparse downstream velocity observations`.
|
||||
- **Evaluation:** `field matching`; spatial L2 primary where supported; DTW complementary, signal-level and case-specific.
|
||||
- **Spatial notation:** distinguish ROI, geometry guard/solid mask, area weights, $E_{mean}$, $E_{phase}$ and any relative comparison. Spatial L2 is never `Level-2` or `Stage-2`.
|
||||
- **Roles:** target/reference, controlled, physical zero/uncontrolled, constant (only where acquired).
|
||||
- **Strength:** `defines`, `uses`, `compares`, `qualifies the selected observable`; not `fully validates`, `proves`, `guarantees`, `optimal`, `robust`, `equivalent`, or `mechanism` without dedicated authority.
|
||||
|
||||
## 8. Result, display and equation ownership inventory
|
||||
|
||||
| Item | Section 2 ownership | Later ownership / prohibition |
|
||||
|---|---|---|
|
||||
| Geometry, BCs, nondimensionalization, solver | Define once here | Later captions identify configuration without redefining method |
|
||||
| Actuation, observations, normalization, PPO settings | Canonical definitions here | SR owns its exact historical executable law; no action-map transfer |
|
||||
| Target/reference and online objective | Define roles and objective here | Capability/cross-case own outcomes under exact cases |
|
||||
| Sensor placement | Coordinates/radius/rationale and literature boundary here | No later optimality or observability inference |
|
||||
| DTW | Algorithm, aggregation, orientation and limits here | Performance/cross-case own values/windows |
|
||||
| Spatial L2/ROI | Reusable definitions here | Kármán performance owns detailed values; Field/CCD uses a different estimand and must not claim replication/agreement |
|
||||
| Mean/phase/event/steady semantics | Definition table here | Each result owner renews changed roles/window only |
|
||||
| CFD qualification | Compact selected-observable table here | No result section may promote it to controller/interpretation validation |
|
||||
| SR/CCD/OID/steady/experiment | Definitions only when needed for boundaries | Owned downstream under fixed scientific cautions |
|
||||
|
||||
**Near-collision freeze.** The standardized Kármán field metric and corrected CCD mean-field accounting remain **different estimands** because acquisitions, role sets, masks, weighting/normalization and windows differ. Periodic mean/phase, event-relative, steady profile, and native/normalized DTW also remain different estimands. SR action magnitude and steady spin are related contextual diagnostics only. No arithmetic pooling, replication, validation, agreement, closure or triangulation wording.
|
||||
|
||||
## 9. Exact literature/style identities and read levels
|
||||
|
||||
### Retained from `STYLE_02_CONTRACTS_METRICS.md`
|
||||
|
||||
1. G. Y. Cornejo Maceda, Y. Li, F. Lusseyran, M. Morzyński & B. R. Noack (2021), “Stabilization of the fluidic pinball with gradient-enriched machine learning control”, *Journal of Fluid Mechanics* **917**, A42. DOI `10.1017/jfm.2021.301`. `full-PDF`, targeted section-architecture read; metadata and relevant setup/sensor passages checked, not reproduced.
|
||||
2. J. Rabault, M. Kuchta, A. Jensen, U. Réglade & N. Cerardi (2019), “Artificial neural networks trained through deep reinforcement learning discover control strategies for active flow control”, *Journal of Fluid Mechanics* **865**, 281–302. DOI `10.1017/jfm.2019.62`. `full-PDF`, targeted section-architecture read; not equation-by-equation verified.
|
||||
3. R. Paris, S. Beneddine & J. Dandois (2021), “Robust flow control and optimal sensor placement using deep reinforcement learning”, *Journal of Fluid Mechanics* **913**, A25. DOI `10.1017/jfm.2020.1170`. `full-PDF`, targeted section-architecture read; not equation-by-equation verified.
|
||||
4. X. Mao, H. M. Blackburn & S. J. Sherwin (2015), “Nonlinear optimal suppression of vortex shedding from a circular cylinder”, *Journal of Fluid Mechanics* **775**, 241–265. DOI `10.1017/jfm.2015.304`. `full-PDF`, targeted section-architecture read; not equation-by-equation verified.
|
||||
|
||||
**Observed practice retained:** these papers vary in headings and equation density, but share dependency order from physical plant/baseline through objective/method and numerical qualification to results. Reusable definitions earn equations; geometry/sensors appear early; qualification caveats sit beside the observable they limit; tables compress settings or convergence rather than dumping provenance. This observation supports the architecture but establishes no project value or validity.
|
||||
|
||||
### Sensor-placement and DRL authorities
|
||||
|
||||
5. C. Raibaudo, P. Zhong, B. R. Noack & R. J. Martinuzzi (2020), “Machine learning strategies applied to the control of a fluidic pinball”, *Physics of Fluids* **32**, 015108. DOI `10.1063/1.5127202`. `full-text web/accepted-manuscript` targeted read. Direct precedent for three downstream hot-wire sensors at $x/D=10$ used to monitor/evaluate the wake; it does not establish the present transverse coordinates, radius, two-component measurement or optimality.
|
||||
6. S. Li, W. Li & B. R. Noack (2022), “Machine-learned control-oriented flow estimation for multi-actuator multi-sensor systems exemplified for the fluidic pinball”, *Journal of Fluid Mechanics* **952**, A36. DOI `10.1017/jfm.2022.908`. `full-report/full accessible manuscript` targeted sensor-placement read. Supports downstream multi-location velocity sensing for phase/state information and explicitly builds on Cornejo Maceda et al.; its nine-probe grid is not the present three-disk geometry.
|
||||
7. L. Marra et al. (2024), “Actuation manifold from snapshot data”, *Journal of Fluid Mechanics*. DOI `10.1017/jfm.2024.593`. Repository state `mapped/full record`; the full PDF was not read in this audit. Use only for low-dimensional controlled-flow plausibility and compact-sensing rationale, not exact placement, optimality or observability.
|
||||
8. J. Schulman, F. Wolski, P. Dhariwal, A. Radford & O. Klimov (2017), “Proximal Policy Optimization Algorithms”, arXiv:1707.06347. Read state inherited as exact original citation, `mapped/full-text record`; use only for the clipped-update principle unless directly reread during bibliography finalization.
|
||||
9. P. Garnier, J. Viquerat, J. Rabault, A. Larcher, A. Kuhnle & E. Hachem (2021), “A review on deep reinforcement learning for fluid mechanics”, *Computers & Fluids* **225**, 104973. DOI `10.1016/j.compfluid.2021.104973`. `full-text` state inherited from the Introduction freeze; optional broad DRL precedent.
|
||||
|
||||
## 10. Dependency-gated execution items
|
||||
|
||||
No unresolved author choice blocks the architecture. Before numerical drafting or production, bind exact per-result PPO settings, grid/time/precision values, selected policy/normalizer/target, qualification rows, and every displayed inherited Reynolds label. These are execution dependencies, not architecture decisions. Zero preset bias remains a design rationale rather than a tested superiority claim.
|
||||
|
||||
The article-wide convention uses bulk/mean inlet speed for the parabolic profile, with the ideal planar relation $U_{max}=1.5U_{ref}$. The current training authority verifies $Re_D=Re_{code}/2$ for its uniform-inflow code labels. An inspected historical parabolic repository note instead labels `U0` as inlet centreline velocity, so parabolic inherited labels must be converted only after source metadata identify centreline versus bulk/mean speed; no unsupported half-label conversion is inferred there.
|
||||
|
||||
## 11. Coverage ledger
|
||||
|
||||
`mapped` is not `read`; `read` is not independently verified; neither implies artifact audit.
|
||||
|
||||
| State | Source/input | Why in scope | Evidence boundary |
|
||||
|---|---|---|---|
|
||||
| `read` | Round-4 `README.md`, `00_MASTER_AGENT_PROMPT.md`, `00_GLOBAL_DECISION_LEDGER.md`, `00_DISPLAY_EQUATION_INVENTORY.md`, `00_QUALITY_GATE.md`, `00_STYLE_CROSSWALK.md`, `STYLE_02_CONTRACTS_METRICS.md`, `SECTION_01_INTRODUCTION_FREEZE.md`, `STYLE_01_INTRODUCTION.md` | Workflow, author decisions, style evidence, interfaces, deletion check | Full repository reads; not independent scientific verification or artifact audit |
|
||||
| `read` | Round-2 `05_REVISED_STRUCTURE_FREEZE.md`, `06_SECTION_LEVEL_JFM_STYLE_GUIDE.md`; Round-3 `03_CONTRACTS_METRICS.md`, `00_MODULE_INTERFACE_LEDGER.md`, `11_FINAL_CONSISTENCY_AUDIT.md` | Controlling architecture, method dossier, ownership and consistency | `full-report`; internal values/artifacts not recomputed |
|
||||
| `read` | `src/drl_pinball/train/TRAIN_PIPELINE.md`, `DESIGN_DECISIONS.md` | Current canonical training method, settings and explicit no-post-fix-rerun boundary | Current authority read directly; selected historical policies/results not independently rerun or artifact-audited |
|
||||
| `read` | Cornejo Maceda et al., Rabault et al., Paris et al., Mao et al. records retained from Style 02 | JFM organization and method-section practice | Declared `full-PDF` targeted reads; literature is style/context only |
|
||||
| `read` | Raibaudo et al. accessible accepted-manuscript/web text; Li et al. accessible full manuscript/repository copy | Sensor precedent and rationale | Targeted full-text reads; do not establish project geometry or optimality |
|
||||
| `read/full-report` | `CELERISLAB_VALIDATION_DOSSIER.md` | Current CelerisLab qualification architecture | Kan99b K2 is `partial pass` (selected frequency/drag/fluctuation observables pass; the mean-lift sign differs under the understood CFD coordinate/force-direction convention, not unresolved physics or wrong rotation; retain the dossier's aggregate label) and Sah04 S2 is `pass`; exact publication values remain artifact/manifest-bound; no older-solver validation |
|
||||
| `mapped` | Current solver/config/environment/evaluation code and manifests named by M03/training authorities | Exact implementation and artifact provenance | Located through read authorities; not exhaustively read or rerun |
|
||||
| `not-covered` | Raw policies/normalizers/targets, external fields, masks, hashes, full qualification matrices, new CFD/training/SR/CCD | Outside section-freeze scope | No rerun, equivalence, result or panel inferred |
|
||||
| `not-covered` | Full DTW sensitivity study, exhaustive PPO tutorial, software history and old implementation comparison | Supplement/provenance or unnecessary | Main architecture remains compact and physical |
|
||||
| `read` | `WAKE_L2_AND_SENSOR_PLACEMENT_RESEARCH_NOTES.md` | Exact sensor geometry/rationale, registered ROI, metric/phase semantics and bibliography boundaries | Working full-report authority reconciled with code on 2026-08-09; not an artifact audit |
|
||||
| `mapped/full record` | Marra et al. (2024), DOI `10.1017/jfm.2024.593` | Low-dimensional controlled-flow plausibility | Full PDF not read here; no exact placement, optimality or observability claim |
|
||||
| `external-inaccessible` | Hidden/external evidence payloads named by manifests | Artifact support outside the visible checkout | Authority-led deferral; no result inferred |
|
||||
|
||||
## 12. Quality gate and disposition
|
||||
|
||||
- [x] Frozen integrated section/style architecture, not manuscript prose.
|
||||
- [x] Reader-facing title and language are physical and JFM-compatible; `contracts`, `named task`, and internal chain labels are removed from reader-facing recommendations.
|
||||
- [x] Logical sequence makes Fig. 1 and later results decodable.
|
||||
- [x] Both inlet/wall configurations, confinement rationale, fixed sensor relationship and case-dependent disturbance/target geometry are represented.
|
||||
- [x] Bulk/mean-based article-wide $Re_D$ policy and centreline relation are explicit; bare `Re100` is prohibited.
|
||||
- [x] Main-text CFD, curved/moving boundary treatment, compact selected-observable qualification and older-solver exclusion are specified.
|
||||
- [x] Canonical no-bias/normalization/PPO method and historical-evidence provenance are separated; rerun completion/equivalence is not asserted.
|
||||
- [x] Sensor coordinates, radius, both velocity components, rationale, current ROI/metric authority, exact citation roles and non-optimality ceiling are retained.
|
||||
- [x] Target construction, reward roles and DRL flow schematic have distinct ownership.
|
||||
- [x] DTW receives algorithmic space and exact limitations; L2/ROI/mean/phase/case semantics remain independent.
|
||||
- [x] Displays/equations have owner, job, readiness, ceiling and fallback; inventory does not imply artifacts exist.
|
||||
- [x] Exact Style 02 identities/DOIs/read levels and observed practice are preserved.
|
||||
- [x] Canonical-main-text policy, compact earlier-implementation provenance, supplement detail, coverage states and fixed scientific cautions are explicit.
|
||||
- [x] `STYLE_02_CONTRACTS_METRICS.md` may be deleted because all unique necessary style evidence, identities/read levels, observed practice, recommendations, exclusions and ownership have been integrated here.
|
||||
- [x] `STYLE_01_INTRODUCTION.md` may be deleted because its unique necessary Introduction style evidence and exact identities/read states are preserved in `SECTION_01_INTRODUCTION_FREEZE.md`; no Section 1 architecture or caution depends solely on the style memo.
|
||||
|
||||
**Disposition:** `frozen` for Round-4 Section 2 architecture. No unresolved author choice remains. Run-specific numerical rows, source-specific parabolic Reynolds conversions, bibliography completion and display production are dependency-gated execution tasks and do not block the freeze.
|
||||
@@ -0,0 +1,252 @@
|
||||
# Round 4 discussion draft — Capability and physical progression
|
||||
|
||||
## 1. Status, purpose and bounded answer
|
||||
|
||||
**Disposition:** `discuss`. This is an author-discussable Round-4 architecture, not polished prose, a frozen section, or a publication-panel claim.
|
||||
|
||||
**Provisional reader-facing title:** **Capability across steady, periodic, transient and retargeted wakes**.
|
||||
|
||||
This is the first results-facing argument. It must establish, selectively rather than as a gallery, capability in four physically different settings; show why the periodic Kármán cloak has the strongest intersection of visual, signal and independent field evidence; and convert that hierarchy into the reason for analytical depth. Detailed Kármán evaluation, full cross-case metrics, steady interpretation and mechanisms remain owned later.
|
||||
|
||||
Call the display **the first results figure (likely Fig. 3)**, not Fig. 1. Section 2 owns preceding geometry/configuration and DRL-flow schematics; final numbering depends on whether those become one or two figures.
|
||||
|
||||
**Question.** What evidence shows progression from steady control to periodic Kármán cloaking, an event-relative Vortex case and a non-zero-target Illusion case without turning breadth into a pooled or general claim?
|
||||
|
||||
**Bounded answer.** Named displays support capability under distinct steady, periodic, event-relative and retargeted comparisons. The periodic Kármán cloak additionally has target–controlled–uncontrolled evidence in which complete-cycle mean and phase-resolved spatial L2 both improve strongly, with complementary signal alignment. That intersection warrants depth; it does not establish transfer, generalization, universal control, positive Illusion SR or causality.
|
||||
|
||||
Targeted Nowledge recall was used only to recover recent figure decisions; internal claims below were checked against repository files.
|
||||
|
||||
## 2. Exact argumentative arc and space budget
|
||||
|
||||
Use this continuous arc:
|
||||
|
||||
`viewing question → steady discovery seed → comparator-bearing Kármán headline → mean/phase quantitative result → action/sensor evidence → removal of periodicity in Vortex → non-zero-target Illusion → envelope and Kármán selection`
|
||||
|
||||
The reader should leave knowing: (i) four named capabilities are non-commensurate; (ii) Kármán is more than the most attractive image because controlled fields move closer to target in persistent-mean and phase-resolved senses; and (iii) only Kármán currently joins policy execution, independent fields and later action/field interpretation.
|
||||
|
||||
**Recommended budget:** 1,250–1,550 body words, eight functional paragraphs of about 140–190 words, plus one near-full-page two-column figure and caption. This is about 2.2–2.8 two-column text pages, or 3.5–4.5 pages of combined article real estate with the display. For an 11,500–13,500-word results/discussion body, this supports the provisional 11% share. Five paragraphs were too thin: separate jobs are needed for visual comparison, mean-versus-phase evidence, action/sensor evidence and the breadth boundary. If compression is required, merge P2/P3 and P6/P7; retain P4, P5 and P8.
|
||||
|
||||
Use at most two printed subheadings: **Periodic Kármán cloaking as the reference case** and **Extensions in temporal and target definition**. The same paragraph order works unheaded.
|
||||
|
||||
## 3. Paragraph-by-paragraph architecture
|
||||
|
||||
### P1 — Viewing contract and figure placement
|
||||
|
||||
**Says.** Place the figure with the opening. All blocks compare a controlled state with a declared target/reference and uncontrolled state, but target meaning and temporal coordinate change. Renew only Section 2's target/controlled/uncontrolled, spatial-L2 and DTW vocabulary. State explicitly that layout does not create a common ranking.
|
||||
|
||||
**Evidence.** Section 2 definitions; Round-2 progression; Round-3 capability/audit; reproduction role schema.
|
||||
|
||||
**Local caveat.** Reader prose uses physical configuration names; internal Legacy/V5 identifiers remain provenance-only. Similar graphics do not make windows or metrics equivalent.
|
||||
|
||||
**Transition.** Explain why the progression begins with a steady result.
|
||||
|
||||
### P2 — Steady discovery seed
|
||||
|
||||
**Says.** Approximately steady rear-cylinder rotation produced the first useful low-deficit state. Compare uniform reference, accepted controlled endpoint and stationary/uncontrolled pinball in the uniform-inflow/free-slip-wall configuration at manuscript $Re_D=50$. A late streamline or disturbance view supplies genuine empirical content without opening the full steady analysis.
|
||||
|
||||
**Evidence.** Current steady endpoint/display authorities and Round-3 steady synthesis.
|
||||
|
||||
**Local caveat.** This is a late/settled endpoint, not a periodic mean or phase view; it is not an optimum, stability result, mechanism or evidence for parabolic/no-slip SR/CCD.
|
||||
|
||||
**Transition.** Periodic shedding tests whether controlled evolution follows a time-dependent target.
|
||||
|
||||
### P3 — Kármán comparator-bearing headline
|
||||
|
||||
**Says.** For the periodic Kármán cloak in the parabolic-inflow/no-slip-wall configuration, read three phase-0 vorticity panels as **target/reference → controlled → uncontrolled**. The controlled wake approaches the target more closely in the registered downstream region. Discuss target proximity and wake organization, not colours or whole-domain identity.
|
||||
|
||||
**Evidence.** Current `karman_re100` role authorities, reproduction manifest and registered evaluation.
|
||||
|
||||
**Local caveat.** Phase 0 is one deterministic phase-aligned snapshot, not a mean, ensemble or uncertainty statement. Final $Re_D$ follows Section 2 policy; no bare inherited `Re100` appears.
|
||||
|
||||
**Transition.** One phase view is persuasive but cannot carry the field claim.
|
||||
|
||||
### P4 — Mean and phase quantitative headline
|
||||
|
||||
**Says.** Give one substantive callout: controlled registered target-field error is about **82% lower than uncontrolled** for both complete-cycle mean and separate eight-slot phase diagnostics. The badge may show rounded source values: $E_{mean}=0.086$ versus $0.475$ and $E_{phase}=0.113$ versus $0.631$ (controlled versus uncontrolled). The mean tests persistent organization; phase-resolved L2 tests evolving periodic organization at matched slots.
|
||||
|
||||
**Evidence.** Exact authority rows `0.085771/0.475206`, `0.112680/0.630878`, relative reductions `0.819508/0.821392`.
|
||||
|
||||
**Local caveat.** This is a preview. Kármán performance owns ROI, geometry guard/mask, normalization, cycles, phase selection, precision and interpretation. Never add the numerically close corrected four-role CCD errors: they are a different estimand.
|
||||
|
||||
**Transition.** Ask what action and sparse signals add.
|
||||
|
||||
### P5 — Action/sensor evidence and depth selection
|
||||
|
||||
**Says.** Show three controlled rotation commands over roughly three target cycles beside one legible sensor phase portrait (or, only if readable, a 2×3 six-channel microgrid). Legend order remains target → controlled → uncontrolled. The target has no action and must not receive a fictitious zero-action trace. State that controlled sparse signals align more closely with target; optionally report target-normalized DTW similarity `0.933` versus `0.625` in prose/caption. Repeated action offset plus modulation motivates later inquiry but is not yet a control-skeleton conclusion.
|
||||
|
||||
**Evidence.** `timeseries.csv` schema, reproduction phase portraits, Kármán DTW summaries and plotting authority.
|
||||
|
||||
**Local caveat.** DTW is higher-is-better signal similarity under a declared window—not spatial error, delay or causality. SR formula/deletion belongs later. Training reward is omitted.
|
||||
|
||||
**Transition.** Change one physical condition at a time.
|
||||
|
||||
### P6 — Vortex: removing periodicity
|
||||
|
||||
**Says.** Use one centred incoming Lamb-vortex case. Show **target/reference → controlled → uncontrolled** vorticity at one declared event-relative offset, recommended `o=0` subject to source/legibility audit. A tiny controlled-action sparkline is optional; do not add another phase portrait.
|
||||
|
||||
**Evidence.** `vortex_lamb_y000`, `event_fields.npz`, offsets `[-10,-5,0,5,10]`, Vortex timeseries authority.
|
||||
|
||||
**Local caveat.** This is event-relative, never phase-resolved. Quantitative event L2 is blocked pending exact evaluator/role binding, so no Vortex L2 appears. It supports named comparable-scale transient capability, not arbitrary-flow control, transfer or generalization.
|
||||
|
||||
**Transition.** Vortex changes temporal coordinate; Illusion changes the desired signature.
|
||||
|
||||
### P7 — Illusion: non-zero retargeting
|
||||
|
||||
**Says.** Show **non-zero target → controlled pinball → uncontrolled pinball** for the retained 0.75 target-radius Illusion case at periodic phase 0. The target is a separately constructed body wake/signature, not physical-zero flow. The question is retargeting rather than suppression.
|
||||
|
||||
**Evidence.** `ill_075L`, seed 43, target/controlled/physical-zero phase roles and reproduction manifest; physical target radius ratio 0.75.
|
||||
|
||||
**Local caveat.** PPO capability only. Print no field L2 unless exact current roles/evaluator are production-bound. This is not full-state equivalence, pooled efficacy or positive Illusion SR. Mention historical negative Illusion SR in one clause, not as another panel.
|
||||
|
||||
**Transition.** Bound breadth and narrow the paper.
|
||||
|
||||
### P8 — Envelope, Re400 decision and bridge
|
||||
|
||||
**Says.** Configuration, temporal coordinate and target definition change, so no pooled ranking follows. Kármán is selected because it uniquely combines comparator-bearing field evidence, periodic action/sensor records and later executable-law/field-decomposition evidence. Vortex/Illusion return under case-specific metrics; steady returns for separate context.
|
||||
|
||||
**Recommendation:** **omit the Re400 marker from this figure.** It meant a small annotation for a uniform/free-slip condition where mean error improved but phase error degraded. It is scientifically useful as an estimand-divergence warning, but a four-case figure with traces and a quantitative badge is already dense. It would introduce another configuration/Re condition and compete with the progression. Consequence: the opening loses a visible cross-Re limit; compensate with one prose sentence here or place exact values with the performance/envelope owner. The ladder is unchanged.
|
||||
|
||||
**Bridge.** End with the unresolved inference: sparse-signature alignment plus one field view establish capability, but independent mean and phase evaluation determines how far the selected periodic wake approaches target.
|
||||
|
||||
## 4. Concrete large-figure storyboard
|
||||
|
||||
### Format and macro-grid
|
||||
|
||||
- **Provisional label:** first results figure, likely Fig. 3; final numbering follows Section-2 display production.
|
||||
- **Size:** full two-column width, about **178 mm × 190–210 mm** before caption; 0.8–1.0 journal page. Never single-column.
|
||||
- **Area:** Kármán **58–62%**; steady **11–13%**; Vortex **13–15%**; Illusion **13–15%**.
|
||||
- **Rows 1–2:** contiguous Kármán block across full width.
|
||||
- **Row 3:** steady (0.9 width unit), Vortex (1.05), Illusion (1.05), in physical order.
|
||||
- Separate blocks by whitespace/rules, not arrows implying numerical continuation. No Re400 or experiment block.
|
||||
|
||||
### Kármán block, provisional (a–f)
|
||||
|
||||
**Field row (34–38% total figure height):**
|
||||
- (a) target/reference phase-0 vorticity;
|
||||
- (b) controlled phase-0 vorticity;
|
||||
- (c) uncontrolled phase-0 vorticity.
|
||||
|
||||
Equal widths, same crop/registration and one common symmetric scale. Print role headings above panels and physical configuration/$Re_D$ once in the block header.
|
||||
|
||||
**Evidence strip (18–22% total height):**
|
||||
- (d) three controlled action traces over about three target cycles; no target action;
|
||||
- (e) one sensor phase portrait, stable order target → controlled → uncontrolled; prefer one legible portrait over six tiny ones;
|
||||
- (f) compact text/mini-dumbbell badge: `mean L2 0.086 | 0.475`, `phase L2 0.113 | 0.631`, labelled `controlled | uncontrolled`, with `≈82% lower`. Safe default: keep DTW in prose/caption so field evidence remains visually primary.
|
||||
|
||||
### Subordinate blocks
|
||||
|
||||
**(g) Steady.** Three-column microtriplet of late total-velocity streamlines: `uniform reference → accepted controlled endpoint → stationary/uncontrolled`. Label `late/settled`. Do not add profile, reverse branch, spin sweep or theory.
|
||||
|
||||
**(h) Vortex.** Three-column event-vorticity triplet: `target/reference → controlled → uncontrolled`, centred Lamb vortex, one literal offset (default `o=0`, audit before production). Optional thin action sparkline. No event-L2 badge.
|
||||
|
||||
**(i) Illusion.** Three-column phase-0 vorticity triplet: `non-zero target → controlled → uncontrolled`, target-radius ratio 0.75, seed 43. Subtitle `PPO capability`. No SR field/curve or efficacy bar.
|
||||
|
||||
### Visual grammar
|
||||
|
||||
1. Invariant role order: target/reference left, controlled centre, uncontrolled right.
|
||||
2. Field colour encodes physics, not role. Use labels/borders: target black, controlled blue, uncontrolled dark grey/orange.
|
||||
3. Physical headers only: `steady reference`, `periodic Kármán cloak`, `incoming Lamb vortex`, `0.75-radius Illusion target`. Internal IDs stay in provenance.
|
||||
4. Temporal badges: `late/settled`, `periodic phase 0`, `event offset o=...`, `periodic phase 0`.
|
||||
5. Common normalized crop within each triplet where geometry permits; do not fake identical registration across configurations.
|
||||
6. One role legend, one portrait legend and one colourbar per compatible family—not per micro-panel.
|
||||
7. Minimum final typography: headers 8–9 pt, roles 7–8 pt, axes/colourbars 6.5–7 pt. Drop optional sparkline/DTW before shrinking text.
|
||||
|
||||
### Colour-scale policy
|
||||
|
||||
Use one common symmetric vorticity scale within every target–controlled–uncontrolled triplet. Across cases/configurations, use case-specific nondimensional limits unless production audit confirms identical normalization and a common scale that preserves structure. A universal scale can hide low-amplitude structure and imply commensurability. Kármán/Illusion may share only if units and normalization are identical and neither saturates. Steady streamlines use their own speed legend. Thus: **common scales within declared comparisons; case-specific scales across non-equivalent comparisons**.
|
||||
|
||||
### Mean, phase and event decoding
|
||||
|
||||
- **Mean:** average over declared complete cycles/window; persistent organization. Here represented by $E_{mean}$, not another map.
|
||||
- **Phase 0:** one independently aligned periodic snapshot; not mean, repeated-cycle ensemble or uncertainty.
|
||||
- **Event-relative:** one offset from a canonical transient event; never periodic phase.
|
||||
- **Late/settled:** endpoint/late field under declared criterion; not phase or mean unless explicitly averaged.
|
||||
|
||||
Pairing phase-0 Kármán visuals with mean and phase scalar summaries lets the eye see one periodic state without asking that snapshot to carry the whole field claim.
|
||||
|
||||
## 5. Output disposition
|
||||
|
||||
| Output | Headline figure | Later owner / reason |
|
||||
|---|---|---|
|
||||
| Action traces | Include once for Kármán; Vortex sparkline optional | SR owns full law/deletion; cross-case owns Vortex detail |
|
||||
| Raw force/sensor time series | Omit | Performance/supplement; too dense |
|
||||
| Force phase portraits | Omit | Sensor portrait is closer to wake-signature objective; both duplicate grammar |
|
||||
| Sensor phase portraits | Include once for Kármán | Full six-channel evidence later/supplement |
|
||||
| Mean fields | No map; Kármán mean-L2 badge only | Performance/field owner; adding mean triplets doubles fields |
|
||||
| Phase-0 vorticity | Include Kármán and Illusion | Eight-phase sequences later/supplement |
|
||||
| Event vorticity | Include one Vortex offset | Multi-offset sequence cross-case/supplement |
|
||||
| Streamlines | Include only steady | Legible endpoint distinction; detail later |
|
||||
| Relative-error maps | Omit | Performance owner; need mask/scale explanation |
|
||||
| Training reward curves | Omit | Retention/performance or supplement; not physical capability |
|
||||
| DTW | One Kármán prose/caption callout | Exact formula/window later; no pooled bars |
|
||||
| Spatial L2 | Kármán mean/phase badge | Detailed values/ROI/window next section; no common cross-case row |
|
||||
| Experiment | Exclude | Author-owned late qualitative section |
|
||||
| Re400 divergence | Omit; prose or later inset | Avoid fifth condition/configuration clutter |
|
||||
|
||||
## 6. Caption skeleton
|
||||
|
||||
> **[Likely Fig. 3; final number depends on preceding displays.] Capability and physical progression under distinct comparison definitions.** (a–c) Periodic Kármán cloak in the parabolic-inflow/no-slip-wall configuration at manuscript-defined $Re_D$: target/reference, controlled and uncontrolled phase-0 vorticity, in that order, with common symmetric scale and matched crop. (d) Controlled cylinder rotations over [window]; the target has no action trajectory. (e) Sparse downstream sensor phase portrait, target, controlled and uncontrolled. (f) Registered field-error headline: complete-cycle mean $E_{mean}$ and separately phased $E_{phase}$, controlled versus uncontrolled; DTW, if stated, is higher-is-better sparse-signal similarity, not a field norm. (g) Steady seed in the uniform-inflow/free-slip-wall configuration: uniform reference, accepted controlled late/settled endpoint and stationary/uncontrolled pinball, shown as streamlines. (h) Centred Lamb-vortex target, controlled and uncontrolled vorticity at event offset $o=[...]$; the offset is not phase and no event L2 is asserted. (i) 0.75 target-radius Illusion, seed 43: non-zero target, controlled and uncontrolled phase-0 vorticity; PPO capability only, with no positive Illusion-SR claim. Scales are common within directly compared role triplets and case-specific across non-equivalent cases unless audited normalization permits sharing. No pooled efficacy, transfer/generalization or experiment is implied.
|
||||
|
||||
Caption must retain physical configuration/$Re_D$, target meaning, role order, variable/scale, temporal coordinate, Kármán mean/phase distinction, DTW orientation if shown, absent Vortex event L2, Illusion PPO-only boundary and no pooling. Keep internal IDs, policy names, full ROI/mask/cycle details and package paths out of the reader caption.
|
||||
|
||||
## 7. Ownership, interfaces and deletions
|
||||
|
||||
**Imports:** Section 2 supplies configurations, $Re_D$, role/action/sensor definitions, DTW/L2 and temporal semantics. Kármán performance supplies one rounded headline comparison. Cross-case supplies Vortex offset and Illusion evidence status. Steady supplies chronology and one endpoint view.
|
||||
|
||||
**Exports:** bounded breadth, Kármán as sole deep case, and the unresolved independent-field question. No mechanism, transferable law, pooled score or experiment.
|
||||
|
||||
**Deletion ownership:**
|
||||
- This section deletes full Kármán ROI/window/mask/cycle/retention detail; performance owns it.
|
||||
- Performance refers back without repeating progression.
|
||||
- Cross-case owns Vortex matrix/DTW, Illusion negative-SR numbers and Erase; delete them here.
|
||||
- Steady owns sign, spin sweep, reverse branch, profiles, theory and interpretation; delete them here.
|
||||
- Globally omit from this figure reward curves, force portraits, mean/error-map triplets, Re400 and experiment.
|
||||
|
||||
## 8. Claim ceiling and near-collision
|
||||
|
||||
**Green:** four named capability blocks under distinct comparisons; Kármán controlled error lower than uncontrolled in declared mean and phase L2; Kármán controlled DTW similarity higher under its own contract; Kármán selected because evidence types coexist.
|
||||
|
||||
**Amber:** breadth across named dynamics/objectives, not generality; steady is context; Vortex supports a named comparable-scale transient reading; Illusion shows PPO objective retargeting.
|
||||
|
||||
**Red:** pooled efficacy/ranking; transfer/generalization/robustness/universality; positive Illusion SR; causal CCD/force-to-field mechanism; DTW as delay/field equivalence; visual similarity as full-state proof; experiment-as-validation.
|
||||
|
||||
**Near-collision:** Kármán mean and phase L2 are related diagnostics, not pooled. Performance L2 and corrected four-role CCD means are different estimands. Steady profiles, periodic L2, event views and Illusion views are different estimands. DTW and L2 are related diagnostics, not replication, agreement or triangulation.
|
||||
|
||||
## 9. Readiness and author decisions
|
||||
|
||||
| Component | Status | Dependency/fallback |
|
||||
|---|---|---|
|
||||
| Composite | `dependency-gated` | Bind source/crop/scale/render provenance; fallback to source-bound panes |
|
||||
| Kármán triplet/trace/portrait | `dependency-gated` | Bind raw fields, phase, indices, action sign/order; portrait may move later |
|
||||
| Kármán L2 badge | `required` for recommendation | Values text-ready; fallback is P4 sentence |
|
||||
| Steady triplet | `dependency-gated` | Bind uniform/accepted/stationary arrays; fallback prose/two panes |
|
||||
| Vortex triplet | visual `dependency-gated`; event L2 `blocked` | Confirm roles/offset; no number |
|
||||
| Illusion triplet | `dependency-gated` | Bind target/controlled/uncontrolled, seed and phase; fallback target/controlled pair |
|
||||
| Re400 | `optional`, recommend omit | Prose/later inset |
|
||||
| Experiment | excluded by author | Later section only |
|
||||
|
||||
No component is called publication-ready.
|
||||
|
||||
**Decisions needing discussion:**
|
||||
1. Approve one near-full-page figure with Kármán about 60% (recommended) versus a two-page display.
|
||||
2. Choose steady streamlines (recommended for small-panel legibility) versus disturbance magnitude; consequence is variable non-uniformity that the caption bounds.
|
||||
3. Confirm centred Lamb vortex and exact event offset; default `o=0` unless an adjacent offset is more legible.
|
||||
4. Choose one Kármán sensor portrait (recommended) versus a 2×3 six-channel microgrid.
|
||||
5. Keep DTW out of the badge and in prose/caption (recommended), preserving L2 primacy.
|
||||
6. Omit Re400 (recommended); consequence is no visible cross-Re failure marker in the opening.
|
||||
7. Resolve final number after Section-2 figure production; working label likely Fig. 3.
|
||||
|
||||
## 10. Compact evidence ledger and disposition
|
||||
|
||||
| State | Inputs | Boundary |
|
||||
|---|---|---|
|
||||
| `read` | Round-2 freeze/style; required Round-3 capability, contracts, performance, cross-case, steady and final audit; Round-4 controls, Style 03 and Section 2 | Full-report architecture/evidence reads; not raw-artifact verification |
|
||||
| `read` | `DATA_DICTIONARY.md`, reproduction manifest, SR reproduction/plot-package READMEs, steady figure README, Round-1 case-metric/Confirmation navigation | Role/order/schema/material availability and visual contracts; no new result |
|
||||
| `read` | `docs/My_Confirmation/Thesis_template.pdf` through available PDF text | Context/material navigation only; no current source-array contract or number inferred |
|
||||
| `mapped` | External phase/event/late role payloads and rendered reproductions | Candidate production sources, not panel approval |
|
||||
| `not-covered` | New CFD/training/evaluation, event-L2 code, rendering, pixel-level PDF audit | Outside architecture; no inference |
|
||||
| `external-inaccessible` | External Optane payloads/model-normalizer bundles where unavailable | Defers publication-source audit; does not block discussion text |
|
||||
|
||||
The old “Fig. 1” numbering assumption is superseded because Section 2 now owns preceding displays; scientific role is unchanged. The eight-paragraph plan and exact storyboard repair the earlier thin allocation and unspecified figure. Exact crops, scales, hashes, selected offset and trace indices remain production dependencies. Quantitative Vortex event L2 remains blocked; no recommended panel needs it.
|
||||
|
||||
**Disposition:** `discuss`, not frozen. Safe recommendation: approve the one-page Kármán-dominant composition; include one Kármán action/sensor group and only Kármán mean/phase L2 as the quantitative badge; use steady streamlines and Vortex/Illusion vorticity microtriplets; omit Re400, reward curves, error maps and experiment.
|
||||
@@ -0,0 +1,262 @@
|
||||
# Round 4 style research — capability opening and headline figure
|
||||
|
||||
**Lane:** section-style research for the capability/progression opening and its headline multi-panel figure.
|
||||
**Status:** `researching` complete; style recommendation ready for coordinator use, not a section freeze, manuscript draft, scientific authority, or panel-readiness declaration.
|
||||
|
||||
## 1. Scope, question, and bounded recommendation
|
||||
|
||||
### Scope and question
|
||||
|
||||
This memo studies how four close *Journal of Fluid Mechanics* papers open a capability/results sequence and use headline figures to move from a declared comparator to evidence, limitation, and deeper analysis. The coded features are breadth versus depth, comparator introduction, panel ordering, captions, failure/limit placement, visual hierarchy, and bridges. External papers provide rhetorical precedent only; they establish no internal result, artifact validity, priority, or scientific claim.
|
||||
|
||||
### Bounded recommendation
|
||||
|
||||
Use the opening as a selective argument rather than a case gallery: establish the viewing contract and comparator first; give the dominant reference case most visual and textual weight; arrange subordinate panels as a readable progression in changed task conditions; place one consequential limitation beside the capability it bounds; and end by converting breadth into the reason for selecting one deep case. The caption should identify roles and comparison semantics while leaving inference, selection logic, and the bridge to prose.
|
||||
|
||||
The recommended rhetorical sequence is:
|
||||
|
||||
`question and viewing contract → comparator-bearing headline → bounded progression → local limit → selection of depth`
|
||||
|
||||
This is a style contract only. It neither selects source arrays nor licenses any quantitative annotation.
|
||||
|
||||
## 2. Coverage ledger
|
||||
|
||||
| Status | Source/input | Why in scope | Evidence state and reason |
|
||||
|---|---|---|---|
|
||||
| `read` | `ROUND4_TOP_DOWN/README.md`, `00_MASTER_AGENT_PROMPT.md`, `00_QUALITY_GATE.md`, `00_GLOBAL_DECISION_LEDGER.md`, `00_DISPLAY_EQUATION_INVENTORY.md` | Round-4 role, status, claim, ownership, display-readiness, and delivery controls | Read directly as repository controls; not independently verified and not artifact-audited |
|
||||
| `read` | `ROUND2_TOP_DOWN/05_REVISED_STRUCTURE_FREEZE.md` and `06_SECTION_LEVEL_JFM_STYLE_GUIDE.md` | Controlling architecture plus prior JFM style synthesis | Read directly as full reports; not independently verified and not artifact-audited |
|
||||
| `read` | `ROUND3_BOTTOM_UP/02_CAPABILITY_PROGRESSION.md`, `00_QUALITY_GATE.md`, and `11_FINAL_CONSISTENCY_AUDIT.md` | Capability opening, figure contract, later readiness findings, and duplication boundaries | Read directly as full reports; not independently verified and not artifact-audited |
|
||||
| `read` | Rabault, Kuchta, Jensen, Réglade & Cerardi (2019), “Artificial neural networks trained through deep reinforcement learning discover control strategies for active flow control,” *Journal of Fluid Mechanics* **865**, 281–302 | Close DRL capability opening; performance-to-flow bridge and local limits | `full-PDF`; figure/section-focused reading; DOI and PDF availability checked against the workspace record; no equation-by-equation verification or source-data audit |
|
||||
| `read` | Paris, Beneddine & Dandois (2021), “Robust flow control and optimal sensor placement using deep reinforcement learning,” *Journal of Fluid Mechanics* **913**, A25 | Reference-case depth followed by physics, robustness, and sensing breadth | `full-PDF`; figure/section-focused reading; DOI and PDF availability checked against the workspace record; no source-data audit |
|
||||
| `read` | Xia, Zhang, Kerrigan & Rigas (2024), “Active flow control for bluff body drag reduction using reinforcement learning with partial measurements,” *Journal of Fluid Mechanics* **981**, A17 | Capability–failure–repair opening and parametric failure envelope | `full-PDF`; figure/section-focused reading; DOI and PDF availability checked against the workspace record; no source-data audit |
|
||||
| `read` | Wang, Yan, Hu, Chen, Rabault & Noack (2024), “Dynamic feature-based deep reinforcement learning for flow control of circular cylinder with sparse surface pressure sensing,” *Journal of Fluid Mechanics* **988**, A4 | Baseline comparison, harder-case progression, caption hierarchy, and later physical reading | `full-PDF`; figure/section-focused reading; DOI and PDF availability checked against the workspace record; no source-data audit |
|
||||
| `mapped` | Remaining JFM identities in the Round-2 style guide and the workspace `JFM/` and `JFM References/` indexes | Broader context and candidate alternatives | Located through the controlling guide/indexes only; not re-read for this lane and not used as direct evidence here |
|
||||
| `not-covered` | Exhaustive sentence coding, all figures in the four papers, unrelated JFM control papers, new literature search, and bibliographic expansion | Not necessary for the bounded comparison of capability openings and headline figures | No practice claim inferred beyond the four-paper sample |
|
||||
| `not-covered` | Internal raw fields, arrays, rendering scripts, figure generation, CFD, training, SR, CCD, and experiment analysis | Outside style-research scope | No panel existence, result, or readiness inferred |
|
||||
| `external-inaccessible` | None essential to this style lane | All four selected records had readable PDFs in the literature workspace | No external-access blocker; full-PDF reading is still not independent scientific verification or artifact audit |
|
||||
|
||||
## 3. Exact literature identities and read levels
|
||||
|
||||
1. Jean Rabault, Miroslav Kuchta, Atle Jensen, Ulysse Réglade & Nicolas Cerardi, “Artificial neural networks trained through deep reinforcement learning discover control strategies for active flow control,” *Journal of Fluid Mechanics* **865** (2019), 281–302. DOI: `10.1017/jfm.2019.62`. Read level: `full-PDF`, targeted to §§2.1, 3.1–3.2, Figs. 2–5, and appendices carrying comparator/limit evidence.
|
||||
2. Romain Paris, Samir Beneddine & Julien Dandois, “Robust flow control and optimal sensor placement using deep reinforcement learning,” *Journal of Fluid Mechanics* **913** (2021), A25. DOI: `10.1017/jfm.2020.1170`. Read level: `full-PDF`, targeted to the uncontrolled/reference definitions and §4 results sequence, especially Fig. 9 and the robustness/sensor-limit figures.
|
||||
3. Chao Xia, Jiaqi Zhang, Eric C. Kerrigan & Georgios Rigas, “Active flow control for bluff body drag reduction using reinforcement learning with partial measurements,” *Journal of Fluid Mechanics* **981** (2024), A17. DOI: `10.1017/jfm.2024.69`. Read level: `full-PDF`, targeted to §§2.4 and 3, especially Figs. 2–5 and the history, sensor, and Reynolds-number limit figures.
|
||||
4. Qiulei Wang, Dong Yan, Guodong Hu, Gang Chen, Jean Rabault & Bernd R. Noack, “Dynamic feature-based deep reinforcement learning for flow control of circular cylinder with sparse surface pressure sensing,” *Journal of Fluid Mechanics* **988** (2024), A4. DOI: `10.1017/jfm.2024.333`. Read level: `full-PDF`, targeted to §§3.2 and 4, especially Figs. 2, 5–7, 9, 13–14, and 16.
|
||||
|
||||
Bibliographic year follows the final JFM volume citation, rather than the earlier year displayed by some discovery records.
|
||||
|
||||
## 4. Observed JFM practice
|
||||
|
||||
The following statements are observations from this four-paper sample, not universal JFM rules.
|
||||
|
||||
### 4.1 Capability openings are comparator-led, but the comparator is often prepared earlier
|
||||
|
||||
- **Rabault et al.** prepare the uncontrolled case during numerical qualification, then reuse it visibly in the results. The opening results sequence moves from multi-run learning behaviour (Fig. 2) to controlled-versus-baseline temporal performance (Fig. 3), and only later to flow morphology (Fig. 5). The comparator therefore exists before interpretation begins.
|
||||
- **Paris et al.** define both the uncontrolled flow and the unstable base flow before the main result. Fig. 9 places drag above action/lift and includes the reference levels directly, so the reader can interpret the transient and maintained regimes without reconstructing the comparison from prose.
|
||||
- **Xia et al.** define three controller/observation configurations schematically before comparing them. The results then preserve the same logical order: full-measurement reference, partial-measurement degradation, and history-enabled repair. Failure is not an appendix; it is the hinge of the argument.
|
||||
- **Wang et al.** first contrast vanilla and dynamic-feature architectures, then use the baseline case to compare sparse variants against a stronger reference before extending to harder conditions.
|
||||
|
||||
**Observed pattern:** the opening does not ask an unlabelled field image to prove capability. A reference, role set, or controller comparison is named before the principal visual inference.
|
||||
|
||||
### 4.2 Breadth follows a reference-case depth anchor
|
||||
|
||||
- **Rabault et al.** use breadth first to establish repeatability across trainings, but devote interpretive depth to one controlled trajectory and its wake morphology.
|
||||
- **Paris et al.** establish a detailed reference case before moving to physics analysis, Reynolds-number variation, noise, and sensor placement. Breadth is earned through dedicated tests, not implied by the first figure.
|
||||
- **Xia et al.** keep the opening comparison narrow and causal in narrative structure—success, degradation, repair—then examine history length, sensor count, Reynolds number, and algorithm choice separately.
|
||||
- **Wang et al.** establish the baseline capability before escalating flow difficulty and dimensionality; sensor quantity and position studies supply intermediate depth.
|
||||
|
||||
**Observed pattern:** breadth is a sequence of changed questions. It is not a montage treated as a common performance scale. One reference case supplies the visual and argumentative anchor.
|
||||
|
||||
### 4.3 Panel order carries the argument
|
||||
|
||||
- **Outcome before actuation:** Rabault Fig. 3 and Paris Fig. 9 foreground drag/performance, with action placed in an inset or lower panel. The hierarchy answers “what changed?” before “what action accompanied it?”
|
||||
- **Reference before degraded/repaired variant:** Xia preserves full-measurement → partial-static → partial-dynamic ordering across learning, temporal, and field figures. Repetition of the same order reduces decoding cost.
|
||||
- **Primary metric before auxiliary evidence:** Wang Fig. 5 orders mean drag, reward, and lift fluctuation, maintaining the distinction between the physical headline, training objective, and secondary fluctuation evidence.
|
||||
- **Baseline and controlled fields use matched positions:** Rabault Fig. 5 stacks uncontrolled and controlled views within repeated variable/statistic groupings, enabling direct eye movement rather than cross-page comparison.
|
||||
|
||||
**Observed pattern:** panel letters are not neutral labels. The order reproduces the logical comparison, and spatial alignment is used to make the intended pairwise reading immediate.
|
||||
|
||||
### 4.4 Captions identify the reading path but do not carry the whole argument
|
||||
|
||||
- Captions name variables, cases, panel letters, and statistical conventions. Wang explicitly notes multi-run averaging and shaded variation; Paris identifies top/bottom quantities and insets; Xia maintains controller labels across figures.
|
||||
- Captions can include a restrained interpretive cue, as in Rabault’s morphology comparison, but the physical explanation and consequence are developed in the text.
|
||||
- The papers do not rely on captions to explain why one case deserves later depth or what broader claim is disallowed. Those are prose jobs.
|
||||
|
||||
**Observed pattern:** effective captions are locally self-decoding—what, compared with what, and how displayed—while selection logic, inference, and bridge remain in the body.
|
||||
|
||||
### 4.5 Failures and limits are placed where they alter interpretation
|
||||
|
||||
- Rabault places exploration-noise and unstable-maintenance limits beside performance; reduced-observation sensitivity is treated separately.
|
||||
- Paris reports failure of simple action-pattern reductions in the physical-analysis sequence and shows limits at harder Reynolds numbers and extreme sensor sparsity in the corresponding extension sections.
|
||||
- Xia makes partial-observation failure the central comparison, then places excessive-history, single-sensor, and cross-Reynolds degradation beside each parameter study.
|
||||
- Wang places vanilla sparse-sensing failure in the baseline comparison and discusses diminishing returns/redundancy in the sensor study.
|
||||
|
||||
**Observed pattern:** a limit is most effective when it changes the inference from the adjacent evidence. A remote limitations paragraph does not substitute for a local boundary.
|
||||
|
||||
### 4.6 Bridges transform capability into the next question
|
||||
|
||||
- Rabault moves from performance to altered recirculation and pressure morphology.
|
||||
- Paris moves from the reference result to post-hoc physical analysis, then to robustness and sensing questions.
|
||||
- Xia turns partial-observation degradation into the need for temporal history, and later turns parameter sensitivity into an envelope of applicability.
|
||||
- Wang turns sparse-sensor comparison into sensor-number and placement questions, then returns to flow fields for physical interpretation.
|
||||
|
||||
**Observed pattern:** the bridge does not announce “the next section.” It states what the current display leaves unresolved and therefore what evidence must follow.
|
||||
|
||||
## 5. Recommendation for the capability/progression opening
|
||||
|
||||
### 5.1 Breadth versus depth
|
||||
|
||||
1. Make the figure answer one capability question, not summarize the manuscript.
|
||||
2. Give the designated deep case the largest contiguous visual region and the clearest comparator-bearing evidence.
|
||||
3. Keep breadth panels subordinate and order them by the intended progression in changed temporal or objective contract.
|
||||
4. Do not encode breadth as generality. Each extension should be read as a named case under its own comparison semantics.
|
||||
5. Reserve mechanism or control-skeleton interpretation for later evidence. The opening should explain why depth is selected, not pre-announce its result.
|
||||
|
||||
**Trade-off:** a broad composition can orient the reader quickly but becomes a gallery if every case receives equal weight. A dominant anchor plus smaller progression panels sacrifices apparent symmetry to gain argumentative hierarchy.
|
||||
|
||||
### 5.2 Comparator introduction
|
||||
|
||||
- The first paragraph should state the figure’s question and the comparison logic before any panel-specific result.
|
||||
- Every retained panel needs a visibly recoverable target/reference and role. Where a quantitative comparator is not ready, label the panel as visual capability rather than allowing proximity or aesthetics to imply measured matching.
|
||||
- Reuse definitions from the contracts section briefly; do not rebuild the methods or metric equations in this opening.
|
||||
- Never use a common bar, colour scale, or ranking device to imply commensurability across distinct estimands.
|
||||
|
||||
**Trade-off:** dense role labels protect interpretation but can overload the image. Put the shortest role/chain identifiers in-panel and the exact comparison semantics in the caption; retain detailed contracts in the preceding section or a compact table.
|
||||
|
||||
### 5.3 Panel ordering and visual hierarchy
|
||||
|
||||
Recommended reading order:
|
||||
|
||||
1. a compact historical seed;
|
||||
2. the dominant comparator-bearing headline case;
|
||||
3. the temporal-contract extension;
|
||||
4. the objective-contract extension;
|
||||
5. at most one small, attached limit marker if it materially changes the headline.
|
||||
|
||||
Use consistent panel grammar: case/chain identifier, target/reference, controlled role, and temporal coordinate in stable locations. Keep matched views aligned. Use area, position, and annotation density—not decorative colour—to establish hierarchy. Any degradation marker should attach to the claim it limits rather than become an equal fifth gallery panel.
|
||||
|
||||
### 5.4 Caption contract
|
||||
|
||||
The caption should, in this order:
|
||||
|
||||
1. state that the composition shows a progression under distinct contracts, not a common efficacy comparison;
|
||||
2. decode each panel’s case/chain, target or reference, controlled/comparator role, and temporal coordinate;
|
||||
3. identify whether evidence is quantitative, complementary, visual-only, or unavailable;
|
||||
4. state the decisive local limits, including non-equivalent metrics and any historical/negative boundary;
|
||||
5. avoid mechanism, generalization, validation, robustness, or cross-chain continuity wording.
|
||||
|
||||
Keep the caption descriptive enough to stand beside the figure, but do not place the section’s selection argument, literature comparison, or full caveat ledger there.
|
||||
|
||||
### 5.5 Failures and limits
|
||||
|
||||
Retain one visible boundary only if it changes how the headline is read and has an exact source. Other blockers belong in local prose, later dedicated sections, or omission fallbacks. A missing quantitative panel should remain absent or explicitly visual-only; it must not be repaired with a suggestive image, filename, or cross-case surrogate.
|
||||
|
||||
The opening’s local-limit sentence should perform three jobs: name the affected claim, state the bounded scope, and explain why the deep case is selected without implying that unselected cases failed.
|
||||
|
||||
### 5.6 Paragraph jobs and figure-to-text cadence
|
||||
|
||||
1. **Viewing contract and placement.** Ask the capability question; name the comparator logic; place the headline figure; state that panels are non-pooled.
|
||||
2. **Historical seed.** Establish chronology and visible capability only; defer separate-chain physics and theory.
|
||||
3. **Headline case.** Read the dominant comparison and state why its evidence intersection warrants depth; detailed result values remain with the performance owner.
|
||||
4. **Progression.** Change one contract at a time and state the evidence level for each subordinate case.
|
||||
5. **Boundary and bridge.** Place the strongest local limitation, then convert breadth into the next question: what does independent field evidence establish in the selected deep case?
|
||||
|
||||
Use one principal figure claim per paragraph. The prose should read the comparison and consequence, not narrate every panel feature.
|
||||
|
||||
## 6. Ownership, interfaces, and bridge
|
||||
|
||||
### Result ownership
|
||||
|
||||
- **Capability/progression owns:** headline composition, historical/progression logic, short comparator-bearing preview, evidence-level labels, one local boundary, and selection of the sole deep case.
|
||||
- **Contracts owns:** reusable plant, chain, action, role, comparator, metric, ROI/window, and compact qualification definitions.
|
||||
- **Performance owns:** detailed field and signal results, values, windows, masks, retained-policy status, and limitations of the deep-case evaluation.
|
||||
- **Cross-case owns:** detailed transient/objective-extension evidence and failures under their case-specific estimands.
|
||||
- **Steady/theory owns:** full separate-chain endpoint, sign, physical context, and theory/failure boundary.
|
||||
|
||||
### Imports
|
||||
|
||||
Import only the minimum viewing definitions from contracts, the panel/readiness boundaries from the display inventory, and the capability evidence levels from the audited Round-3 module. Literature contributes style precedent only.
|
||||
|
||||
### Exports
|
||||
|
||||
Export a bounded breadth statement, the identity of the selected deep case, and the unresolved question that justifies independent field evaluation. Export no pooled score, mechanism, cross-case interpretation, or panel-readiness claim.
|
||||
|
||||
### Opening dependency and closing bridge
|
||||
|
||||
**Opening dependency:** the reader must already know enough of the role, comparator, chain, and metric vocabulary to decode the figure without a methods digression.
|
||||
|
||||
**Closing bridge function:** transform “named capability under distinct contracts” into “why the selected case requires and permits deeper independent field evaluation.” The bridge should name the missing inference, not a section number.
|
||||
|
||||
### Duplication deletions
|
||||
|
||||
- Delete detailed deep-case values and metric-window prose from the capability opening; retain only the headline comparison needed to read the figure.
|
||||
- Delete full steady/theory interpretation from the historical seed.
|
||||
- Delete detailed transient and objective-extension result prose from the progression preview.
|
||||
- Delete the global claim ladder, package histories, and artifact-status inventory from narrative paragraphs.
|
||||
- Do not duplicate caption-level role decoding in adjacent prose unless it changes the inference.
|
||||
|
||||
## 7. Claim ceiling and near-collision findings
|
||||
|
||||
### Green
|
||||
|
||||
- The opening may state named capability/progression under explicitly distinct contracts.
|
||||
- It may identify one case as the sole deep case because the required evidence types coexist there.
|
||||
- It may describe a panel as quantitative, complementary, visual-only, historical/negative, dependency-gated, or blocked when the controlling authority supports that status.
|
||||
|
||||
### Amber
|
||||
|
||||
- The composition may be described as breadth across named temporal dynamics or objectives only when immediately bounded as non-pooled and non-general.
|
||||
- A visual progression may motivate deeper examination; it cannot establish mechanism, transfer, robustness, or full-state equivalence.
|
||||
|
||||
### Prohibited red
|
||||
|
||||
No pooled efficacy, universal/general capability, robustness, stability, optimality, transfer, mechanism, causality, positive Illusion SR, experiment-as-validation, Legacy/V5 continuation, DTW-as-field-equivalence, or visual-match-as-quantitative-proof wording.
|
||||
|
||||
### Near-collision findings
|
||||
|
||||
The figure juxtaposes **different estimands**, not repeated measurements of one quantity. Periodic mean/phase fields, transient event-relative views, steady profiles, and target-retargeting visuals differ in acquisition, role set, chain, domain/ROI, mask, weighting/normalization, and window/coordinate. They must not be combined arithmetically or described as replication, validation, agreement, or triangulation. Closely valued results owned by the deep-performance and later field/CCD sections remain different estimands under their distinct role acquisition, masks, weighting, and windows; the capability figure should not place them together.
|
||||
|
||||
## 8. Display contract and readiness
|
||||
|
||||
| Display | Owner | Argumentative job | Source/roles required | Caption ceiling | Classification and readiness | Fallback |
|
||||
|---|---|---|---|---|---|---|
|
||||
| Headline capability/progression composition | Capability/progression | Establish a selective progression and make the deep-case choice legible | Exact authority for every retained panel; chain/solver/geometry; case/seed/role; comparator; temporal coordinate; metric status; rendering provenance | Named capability under distinct contracts; no pooled efficacy, generality, mechanism, or cross-chain continuity | `dependency-gated`; text contract is usable, but no panel is publication-ready merely from this memo | Retain only source-bound panels; omit unsupported annotations |
|
||||
| Dominant comparator-bearing block | Capability/progression, with detailed values owned by performance | Carry the principal visual comparison and justify depth selection | Exact target, controlled, and comparator roles under one bound chain and evaluation contract | Headline comparison only; detailed values, ROI/mask/window, and interpretation remain with performance | `dependency-gated` pending source arrays and rendering audit | Compact source-bound comparison or textual preview |
|
||||
| Subordinate progression panels | Capability/progression; detail later owned by cross-case and steady sections | Mark chronology and one-at-a-time contract changes | Exact case/chain/role and evidence-level identity for each panel | Visual/named capability only where quantitative roles are absent | Mixed `dependency-gated`/`blocked` according to the global inventory | Omit blocked quantitative content; preserve bounded prose progression |
|
||||
| Optional degradation/limit marker | Capability/progression | Prevent the headline from being read as uniform breadth or generality | Exact authority and same-case comparator for the displayed diagnostic | Named diagnostic only; no failure cause or cross-condition law | `optional`, drop-first | One local limit sentence in prose |
|
||||
|
||||
No new table or equation is recommended for this opening. Reusable definitions belong to contracts; detailed quantitative tables belong to the owning result sections. Text readiness remains separate from panel/artifact readiness.
|
||||
|
||||
## 9. Anti-patterns
|
||||
|
||||
1. **Equal-weight gallery:** all cases receive the same area, encouraging an unsupported common ranking.
|
||||
2. **Comparator afterthought:** target, baseline, or role appears only in the caption after the reader has already interpreted the image.
|
||||
3. **Caption as manuscript:** the caption carries selection logic, mechanism, caveats, and method detail that the prose should argue.
|
||||
4. **Metric wallpaper:** several incompatible scores are printed to make the composition look quantitative.
|
||||
5. **Aesthetic equivalence:** similar colours or wake shapes are treated as measured matching.
|
||||
6. **Failure quarantine:** degradation is delayed to Discussion even though it changes the headline claim.
|
||||
7. **Failure spectacle:** a weak or blocked case is promoted to an equal panel merely to signal honesty, obscuring the principal progression.
|
||||
8. **Method-first opening:** architecture, training chronology, or package history precedes the comparator-bearing result.
|
||||
9. **Mechanism pre-announcement:** the opening names the later control-skeleton interpretation before executable and field evidence.
|
||||
10. **Breadth-to-generality leap:** several named cases become “robust,” “general,” or “arbitrary-flow” capability without dedicated tests.
|
||||
11. **Chain continuity by layout:** Legacy and V5 panels share visual grammar without explicit lane labels and are read as one numerical sequence.
|
||||
12. **Bridge by section number:** the closing sentence announces the next section instead of identifying the unresolved scientific question.
|
||||
|
||||
## 10. Conflicts, deferrals, blockers, and disposition
|
||||
|
||||
### Conflicts
|
||||
|
||||
No conflict was found between the four-paper style sample and the controlling recommendation that a comparator-bearing reference case precede extensions and interpretation. The literature sample does not resolve internal panel provenance or scientific claim questions and is not used to do so.
|
||||
|
||||
### Deferrals and blockers
|
||||
|
||||
Panel provenance, raw arrays, masks, exact renderings, and publication production remain outside this lane. Quantitative transient content, unsupported objective-extension roles, and any panel listed as blocked in the global inventory remain blocked; this memo does not lower those gates. Optional missing panels do not block the style contract when omission preserves the argument.
|
||||
|
||||
### Author/coordinator decisions
|
||||
|
||||
- Decide the final retained panel set under the existing safe default: preserve the dominant progression and omit optional or blocked content first.
|
||||
- Decide whether one source-bound degradation marker materially improves the opening; safe default is prose-only limitation if the marker crowds the hierarchy.
|
||||
- Preserve the existing exclusion of experiment from the headline composition.
|
||||
|
||||
### Disposition
|
||||
|
||||
`discuss`: the style research is complete and passes the Round-4 style-memo criteria, but panel selection remains a coordinator/author decision and panel readiness remains dependency-gated. This first-pass memo does not self-freeze the capability section.
|
||||
@@ -0,0 +1,230 @@
|
||||
# Round 4 style research — Kármán performance
|
||||
|
||||
**Lane:** section-style research only; not manuscript prose, internal-result authority, panel production, or a section freeze.
|
||||
**First-pass disposition:** `revise` pending coordinator second audit.
|
||||
|
||||
## 1. Scope, question and bounded recommendation
|
||||
|
||||
**Question.** How do the closest JFM reference-case performance sections order results, establish baselines and comparators, separate field evidence from signal/global metrics, pace figures and tables, place exact values and local limitations, and turn a performance result into the next interpretive question?
|
||||
|
||||
**Bounded recommendation.** Give this section one compact reference-case argument: declare the exact evaluation/comparator contract; state the primary field result; place complementary signal evidence beside rather than above it; expose one local evidential boundary; then hand the established performance to the action-law question. This memo recommends rhetorical form only. It supplies no project prose, no project number, no new internal claim, and no inference from external literature to internal evidence.
|
||||
|
||||
The closest precedents do not impose one universal order. Paris et al., Rabault et al. and Cornejo Maceda et al. establish global or temporal performance before spatial interpretation, whereas Mao et al. alternates sensitivity, forcing distribution, global suppression, fields and stability because each representation tests a different step. The transferable practice is **comparator-led evidence escalation**, not mechanical “signals first” imitation. For this paper, the controlling architecture independently requires spatial L2 to carry the field claim and DTW to remain complementary; that project-specific choice overrides the references' drag-centred ordering.
|
||||
|
||||
## 2. Coverage ledger
|
||||
|
||||
| Status | Source/input | Why in scope | Evidence state/reason |
|
||||
|---|---|---|---|
|
||||
| `read` | `ROUND4_TOP_DOWN/README.md` | Round-4 purpose, phases, statuses and Round-5 handoff | Read directly; not independent scientific verification or artifact audit |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_MASTER_AGENT_PROMPT.md` | Required deliverables, fixed ladder, style method and first-pass rule | Read directly; controlling workflow |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_QUALITY_GATE.md` | Style gate and coverage-state semantics | Read directly; controlling workflow |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_GLOBAL_DECISION_LEDGER.md` | Performance ownership, M04/M06 boundary and choices | Read directly; decision authority, not evidence |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_DISPLAY_EQUATION_INVENTORY.md` | Display/equation classification and fallback | Read directly; inventory is not artifact readiness |
|
||||
| `read` | `ROUND2_TOP_DOWN/05_REVISED_STRUCTURE_FREEZE.md` | Frozen section role, space and metric hierarchy | `full-report`; not independently verified or artifact-audited |
|
||||
| `read` | `ROUND2_TOP_DOWN/06_SECTION_LEVEL_JFM_STYLE_GUIDE.md` | R2 observed practice and performance contract | `full-report`; four closest PDFs were re-read below |
|
||||
| `read` | `ROUND3_BOTTOM_UP/04_KARMAN_PERFORMANCE.md` | Owning R3 dossier, interfaces, limits and display needs | `full-report`; internal values intentionally not reproduced; raw artifacts not audited |
|
||||
| `read` | `ROUND3_BOTTOM_UP/11_FINAL_CONSISTENCY_AUDIT.md` | Ownership/deletion and near-collision ruling | `full-report`; no recomputation or artifact audit |
|
||||
| `read` | Paris, Beneddine & Dandois, “Robust flow control and optimal sensor placement using deep reinforcement learning”, *Journal of Fluid Mechanics* 913, A25 (2021), DOI `10.1017/jfm.2020.1170` | DRL reference-case progression from performance to fields and robustness | `full-PDF`; targeted reread of §4/displays; metadata/DOI checked; no numerical reproduction |
|
||||
| `read` | Mao, Blackburn & Sherwin, “Nonlinear optimal suppression of vortex shedding from a circular cylinder”, *Journal of Fluid Mechanics* 775, 241–265 (2015), DOI `10.1017/jfm.2015.304` | Global performance, fields, stability, interpretation and realizability | `full-PDF`; targeted reread of §4; metadata/DOI checked; no numerical reproduction |
|
||||
| `read` | Rabault, Kuchta, Jensen, Réglade & Cerardi, “Artificial neural networks trained through deep reinforcement learning discover control strategies for active flow control”, *Journal of Fluid Mechanics* 865, 281–302 (2019), DOI `10.1017/jfm.2019.62` | Compact DRL reference case with spread, traces, fields and limits | `full-PDF`; targeted reread of §3/relevant appendix; metadata/DOI checked; no numerical reproduction |
|
||||
| `read` | Cornejo Maceda, Li, Lusseyran, Morzyński & Noack, “Stabilization of the fluidic pinball with gradient-enriched machine learning control”, *Journal of Fluid Mechanics* 917, A42 (2021), DOI `10.1017/jfm.2021.301` | Closest plant and comparator ladder | `full-PDF`; targeted reread of §4/summary displays; metadata/DOI checked; no numerical reproduction |
|
||||
| `mapped` | Remaining six exact JFM papers in the R2 style guide | Broader precedent not needed for this bounded reread | Identities and prior lessons mapped through R2; PDFs not re-read in this lane |
|
||||
| `mapped` | R3 controls/modules other than M04/M11 | Cross-section interfaces and terminology | Located through ledgers/audit; not read as evidence for this question |
|
||||
| `not-covered` | New search, exhaustive sentence coding, unrelated JFM papers and citation-network expansion | Four closest PDFs plus controls answer the bounded question | No practice/scientific claim inferred beyond this sample |
|
||||
| `not-covered` | Raw fields/arrays, model bundles, evaluator reruns, CFD, training, SR/CCD analysis and figure generation | Explicitly outside style research | No internal quantity independently verified; no panel called ready |
|
||||
| `external-inaccessible` | None essential to the style question | All four PDFs were available | Hidden/external project artifacts remain out of scope and artifact-audit deferred |
|
||||
|
||||
**State boundary.** `read` means the named report or targeted full PDF was inspected. It does not mean an internal result was independently verified, a bibliography exhaustively audited, or a project artifact audited. Targeted memory recall only located controlling decisions; repository controls and full PDFs supplied the memo's substance.
|
||||
|
||||
## 3. Observed practice in the four closest JFM papers
|
||||
|
||||
### 3.1 Paris, Beneddine & Dandois (2021)
|
||||
|
||||
**Result order.** Section 4 moves through control performance/efficiency (§4.1), controlled-flow analysis (§4.2), Reynolds and noise robustness (§4.3), sensor-number/location effects (§4.4), and sensor optimization (§4.5). The reference case is established before robustness and design extensions.
|
||||
|
||||
**Baseline/comparator.** The uncontrolled baseline and steady base-flow reference anchor the opening. A literature table situates the headline; robustness later uses policies trained at individual Reynolds numbers as stricter internal comparators. Comparator changes are explicit.
|
||||
|
||||
**Signals and fields.** Drag, lift and actuation histories precede vorticity and recirculation fields; spectra connect temporal behaviour to the physical reading. This order reflects a drag-centred objective, not a rule that reward-adjacent signals outrank independent fields.
|
||||
|
||||
**Cadence/numbers.** Time traces, efficiency, fields, spectra, robustness and sensing each receive nearby displays. A few inference-changing values appear in prose (baseline and headline reduction); multi-study/noise comparisons move to tables. Captions carry parameters and panel definitions.
|
||||
|
||||
**Limits/transitions.** Limits sit beside claims: Reynolds asymmetry is related to sensor coverage; absence of a simple action pattern appears during flow analysis; reference-Re-only optimization limits broader comparison. The transition is `global change → spatial/spectral correlate → bounded explanation → robustness test`.
|
||||
|
||||
### 3.2 Mao, Blackburn & Sherwin (2015)
|
||||
|
||||
**Result order.** Section 4 proceeds from sensitivity (§4.1), to nonlinear control distribution (§4.2), suppression effects (§4.3), mechanism synthesis (§4.4), and controlled drag (§4.5). Representations alternate because each answers the next deductive question.
|
||||
|
||||
**Baseline/comparator.** Linear sensitivity is the recurring simple comparator for nonlinear solutions; uncontrolled and two-/three-dimensional states enter where they test suppression and admissibility.
|
||||
|
||||
**Signals and fields.** Frequency response and global unsteadiness establish forcing choice; boundary distributions expose linear/nonlinear divergence; vorticity shows suppression; stability distinguishes a lasting state from a transient image; drag closes the consequence.
|
||||
|
||||
**Cadence/numbers.** Results use a figure chain and no results table. Prose retains decisive selected frequencies, approximate drag-reduction ranges and effectiveness; parameters/contours remain in plots and captions.
|
||||
|
||||
**Limits/transitions.** The shrinking linear regime, failure of non-zero-frequency forcing and actuator-efficiency omission are local. The key move is `field suppression → stability test → physical schematic`; synthesis is earned by prior evidence.
|
||||
|
||||
### 3.3 Rabault et al. (2019)
|
||||
|
||||
**Result order.** Section 3 first establishes drag reduction (§3.1), including repeat learning and controlled histories, then analyses the strategy (§3.2) through wake and pressure fields. “What/how much” and “what changed” remain adjacent.
|
||||
|
||||
**Baseline/comparator.** The uncontrolled flow is qualified before results and remains primary. A separately constructed steady symmetry-constrained reference is a limiting comparator, not the uncontrolled baseline.
|
||||
|
||||
**Signals and fields.** Multi-run learning statistics and drag/control histories precede velocity and pressure fields. Fields explain a scalar objective; they are not an independent field-matching metric.
|
||||
|
||||
**Cadence/numbers.** Learning spread, definitive histories, then velocity/pressure fields form a compact sequence. Prose repeats headline magnitudes that change interpretation; figures carry temporal/spatial detail.
|
||||
|
||||
**Limits/transitions.** Network internals are treated as a black box; probe sensitivity is outside scope; training cost is contextualized. Field differences motivate a fluid-mechanical analogy. That causal register is evidence-specific and should not be copied under a descriptive ceiling.
|
||||
|
||||
### 3.4 Cornejo Maceda et al. (2021)
|
||||
|
||||
**Result order.** Control-space richness increases: symmetric steady (§4.1), general non-symmetric steady (§4.2), feedback (§4.3), after the unforced flow is established. Each residual failure motivates the next space.
|
||||
|
||||
**Baseline/comparator.** A common normalized cost keeps unforced, simple-steady, richer-steady and feedback cases legible. Each comparator asks whether greater freedom buys a specific increment.
|
||||
|
||||
**Signals and fields.** Within each stage, lift/cost histories and spectra establish performance before phase/mean vorticity fields interpret it. Summary graphics later compress the hierarchy.
|
||||
|
||||
**Cadence/numbers.** A repeated metric/history→field→local-conclusion rhythm creates continuity. Selected costs/control vectors appear in prose/equations; detailed law weights move to a table; summary graphics collect comparative values.
|
||||
|
||||
**Limits/transitions.** Failure of the best symmetric forcing to stabilize motivates richer search; incomplete feedback authority stays beside the final comparison. The bridge is `best current result → residual failure → justified increase in control freedom`.
|
||||
|
||||
## 4. Observed practice versus project recommendation
|
||||
|
||||
### 4.1 Result order
|
||||
|
||||
**Observed.** All four establish a named reference case before extensions. Three move mainly from global/time evidence to fields; Mao alternates representations. Broad physical synthesis follows the comparator and principal outcome.
|
||||
|
||||
**Recommended four-paragraph order.** (1) Exact chain/case/Re/roles/comparator/window and selection contract, briefly imported from Contracts. (2) Primary mean and phase spatial L2 as separate estimands against the declared physical-zero comparator, with the principal display and local ROI/mask/window limits. (3) Complementary DTW orientation/result and a direction-only consistency check under its separate signal contract. (4) Phase/mask/seed/retention boundary and the action-law-complexity handoff. If V5 remains, label it secondary and separate.
|
||||
|
||||
**Trade-off.** This reverses the predominant drag-centred JFM sequence but follows the frozen evidentiary hierarchy. DTW-first would be familiar to DRL readers yet make independent fields look confirmatory; L2-first prevents optimized signals from appearing self-validating.
|
||||
|
||||
### 4.2 Baseline and comparator
|
||||
|
||||
**Observed.** Papers name a physical baseline first, then add simple, limiting or separately trained comparators only to sharpen a question. Shared metrics are used only when contracts permit.
|
||||
|
||||
**Recommended.** Keep the principal comparison triadic and chain-local: declared target, controlled role and physical zero. Import definitions, do not retell acquisition. Put V5 retention in a separate sentence/inset, never another Legacy field row. Do not manufacture a theoretical-best comparator because a reference paper uses one.
|
||||
|
||||
### 4.3 Field versus signal metrics
|
||||
|
||||
**Observed.** Global/time signals usually establish efficacy and fields explain it because the references optimize drag/lift/fluctuations. Spectra often bridge layers. None licenses signal alignment as spatial equivalence.
|
||||
|
||||
**Recommended.** Name the estimand in every result sentence: `complete-cycle mean-field error`, `phase-resolved field diagnostic`, or `signal-level DTW similarity`. Define DTW orientation once. If diagnostics point the same way, call them consistent under separate contracts—not replicated, independently validating, or the same evidence.
|
||||
|
||||
### 4.4 Figure/table cadence
|
||||
|
||||
**Observed.** Displays arrive when the argument changes representation. Paris is figure/table dense; Mao uses a figure chain; Rabault uses learning→signal→field; Cornejo Maceda repeats metric→field and later compresses the hierarchy. Tables suit many comparable rows; figures suit temporal/spatial relationships.
|
||||
|
||||
**Recommended.** Use one compact L2-first performance display with subordinate DTW. A compact result table is fallback if field rendering provenance is unavailable. A V5 seed/retention inset is optional/drop-first. Keep one principal display claim per paragraph.
|
||||
|
||||
### 4.5 Exact numbers
|
||||
|
||||
**Observed.** Numbers enter prose when they define a baseline, headline, selected control or consequential limit. Multi-condition values go to tables/plots; contour/setup values to captions.
|
||||
|
||||
**Recommended.** Retain the minimum exact set needed for controlled-versus-zero comparison per owned estimand. Move provenance precision/secondary variants to table or supplement. This memo intentionally reproduces no project number. Never cite a number without chain, comparator, metric orientation and window.
|
||||
|
||||
### 4.6 Local limitations
|
||||
|
||||
**Observed.** Strong references put limitations beside affected results: incomplete authority, shrinking linear validity, sensor coverage, absent actuator efficiency, black-box interpretation or restricted optimization. Failure often motivates the next test.
|
||||
|
||||
**Recommended.** Attach field→mask/phase-sampling limit; DTW→signal/window/normalization limit; V5 retention→separate-chain/best-versus-retained/no-robustness limit; display→artifact-readiness limit. End with one aggregate ceiling: named-contract field matching does not identify an action law or mechanism.
|
||||
|
||||
### 4.7 Result-to-interpretation transitions
|
||||
|
||||
**Observed.** Effective bridges transform results into tests: performance→controlled-flow analysis; suppression→stability; simple-forcing failure→richer control; global change→spatial/spectral correlate.
|
||||
|
||||
**Recommended.** Use two steps: the named field comparison is not solely a reward-proxy statement; it still does not determine whether learned action is complex or compressible. SR may test an executable surrogate against the declared contract, not treat performance as proof of a persistent-action mechanism.
|
||||
|
||||
## 5. Paragraph jobs, ownership and interfaces
|
||||
|
||||
| Job | Import | Owned function | Local limit | Export |
|
||||
|---|---|---|---|---|
|
||||
| P1 exact contract | Contracts: chain, roles, Re, metrics, ROI/window | Establish one comparison and selection status | Do not repeat Methods/capability | Legible comparator |
|
||||
| P2 L2-first result | M04 detailed Legacy values/contracts | Centre independent mean/phase field evidence | Separate estimands; local mask/window limits | Named field matching only |
|
||||
| P3 signal companion | DTW definition/orientation once | Compare direction without pooling | Reward-adjacent, signal-only; no delay/equivalence | Diagnostic compatibility only |
|
||||
| P4 boundary/handoff | M04 seed/retention boundary | Define ceiling and action-law question | Legacy/V5 separate; no robustness | Performance arbiter, not mechanism |
|
||||
|
||||
**Ownership.** Performance alone owns detailed Legacy L2/DTW values, roles, ROI/windows and V5 seed/retention boundary. Capability keeps a headline. Contracts owns reusable equations. Field/CCD owns corrected four-role mean/residual estimands and cannot use M04 as replication/closure. SR owns its law and same-protocol arbitration.
|
||||
|
||||
**Opening dependency.** Contracts must freeze role vocabulary, Legacy/V5, Reynolds conventions, metric orientation and ROI/window notation.
|
||||
|
||||
**Closing bridge.** Convert “what the policy achieved” into “how complex an executable law must be,” withholding mechanism, necessity and uniqueness.
|
||||
|
||||
**Duplication deletions.** Delete detailed values from capability; repeated metric equations from performance; any second full M04/M06 collision explanation outside its owner/interface; training chronology/artifact inventories from narrative; duplicate V5 support.
|
||||
|
||||
## 6. Claim ceiling and near-collision audit
|
||||
|
||||
**Green:** `is closer to the declared target than physical zero under the named spatial estimand`; `mean and phase diagnostics are separate`; `signal similarity is directionally consistent under a separate contract`; `retention is secondary and chain-specific`.
|
||||
|
||||
**Amber:** `supports named-condition field matching in the registered region`; `is not solely a reward-adjacent signal statement`; `motivates testing a reduced executable law`.
|
||||
|
||||
**Prohibited red:** `restores the wake`, `full-field identity`, `mechanism`, `causes`, `proves`, `robust`, `general`, `stable`, `optimal`, `universal`, `independently validates`, `replicates`, `triangulates`; DTW as spatial error/delay/causality/equivalence; mean/phase agreement as uncertainty; retention as convergence; Legacy/V5 continuation; literature as internal authorization.
|
||||
|
||||
| Pair | Contract check | Class | Required treatment |
|
||||
|---|---|---|---|
|
||||
| M04 mean L2 vs phase L2 | Same family/direction; different averaging/phase construction/window | `related diagnostic` | Separate; no pooling/uncertainty claim |
|
||||
| M04 L2 vs DTW | Fields/ROI versus sparse histories; different normalization/geometry/window | `different estimand` | L2 field claim; DTW complementary |
|
||||
| M04 PPO mean vs M06 corrected four-role mean | Different acquisitions, roles, mask, weighting/normalization and windows despite proximity | `different estimand` | Never combine, cross-normalize, replicate, validate, agree, close or triangulate |
|
||||
| Legacy performance vs V5 retention | Different chain, solver/geometry, selection and role | `different estimand` | Secondary context; no pooling/continuation |
|
||||
|
||||
## 7. Figure, table and equation contracts
|
||||
|
||||
### Kármán performance display
|
||||
- **Owner/class:** performance; `dependency-gated`.
|
||||
- **Job/source/roles:** named Legacy target/controlled/physical-zero comparison from exact R3-bound authorities; L2 primary, DTW subordinate; mean/phase distinct.
|
||||
- **Caption ceiling:** chain, case/Re, roles, ROI/mask/window, normalization and phase semantics; no causal, robust, universal, cross-chain or M06-validation wording.
|
||||
- **Readiness:** text/numbers bound in R3; this pass does not establish raw-role/manifest/render/orientation readiness.
|
||||
- **Fallback:** compact result table; omit unsourced redraws.
|
||||
|
||||
### Optional V5 seed/retention inset
|
||||
- **Owner/class:** performance; `dependency-gated`, optional/drop-first.
|
||||
- **Job:** distinguish discovery from late retention without strengthening Legacy.
|
||||
- **Ceiling/readiness:** separate-chain diagnostic; no robustness/convergence or selected-policy reproducibility without exact model/normalizer. Text bounded; redraw dependent.
|
||||
- **Fallback:** one sentence or omission.
|
||||
|
||||
### Compact performance table
|
||||
- **Owner/class:** performance; `optional` fallback.
|
||||
- **Job/source:** controlled and zero side by side for each separate Legacy estimand from M04; DTW orientation explicit; no M06/V5 rows in Legacy block.
|
||||
- **Ceiling/readiness:** no universal score/uncertainty/cross-chain synthesis; final notation, rounding and source audit remain production work.
|
||||
- **Fallback:** only inference-changing values in prose.
|
||||
|
||||
### Metric equations
|
||||
- **Owner/class:** Contracts; manuscript-wide `required`, reused rather than re-derived here.
|
||||
- **Job/ceiling:** define mean L2, phase L2, zero-relative comparison and DTW orientation once; spatial L2 is not a pipeline gate and DTW not delay/field equivalence.
|
||||
- **Readiness/fallback:** definition-ready; notation freeze pending; use inline reminder/pointer if no repeated display.
|
||||
|
||||
## 8. Anti-patterns
|
||||
|
||||
1. Training diary before comparator.
|
||||
2. Reward-first self-validation.
|
||||
3. Flattening mean L2, phase L2, DTW and retention into “performance”.
|
||||
4. Unannounced comparator drift.
|
||||
5. Precision/seed dumping in prose.
|
||||
6. Caption-like panel narration.
|
||||
7. Delaying mask, phase, signal-only or retention limits to Discussion.
|
||||
8. Pre-empting SR/CCD interpretation.
|
||||
9. Calling M06 replication/agreement/closure or DTW independent field validation.
|
||||
10. Using V5 to imply Legacy robustness/reproducibility.
|
||||
11. Borrowing causal language from papers with different dedicated tests.
|
||||
12. Package/path/hash/report voice in manuscript narrative.
|
||||
13. Empty “next section presents SR” transition.
|
||||
|
||||
## 9. Decisions, conflicts, deferrals and blockers
|
||||
|
||||
**Evidence-governed:** M04 owns detailed Legacy L2/DTW; L2 is primary and DTW complementary; mean/phase stay separate; M04/M06 differ; Legacy/V5 stay separate; the close asks action-law complexity, not mechanism.
|
||||
|
||||
**Author/coordinator choices (`discuss`).** (1) Recommend one L2-first display with subordinate DTW; safe fallback compact table. (2) Retain V5 inset only if it does not displace Legacy; safe default omit/drop-first. (3) Put only headline comparator values in prose and detailed precision in table/supplement; exact density/rounding awaits audit. None is frozen here.
|
||||
|
||||
**Conflict check.** No control/R2/R3 conflict was found. The cross-paper tendency toward scalar-first order reflects drag/stabilization objectives; it does not override the frozen project hierarchy that makes independent spatial field matching primary.
|
||||
|
||||
**Deferrals/blockers.** No style-content blocker. Panel audit/render provenance/final rounding are deferred. Same-selection V5 model/normalizer is required only if a selected-policy reproduction panel is retained. First pass cannot self-freeze; coordinator must audit controls, identities/read levels, ownership, near-collision wording, display status and choices.
|
||||
|
||||
## 10. Validation and disposition
|
||||
|
||||
- [x] Only the requested style-research file created.
|
||||
- [x] All Round-4 controls/README, R2 freeze/style, and R3 M04/final audit read.
|
||||
- [x] Four closest JFM papers re-read at `full-PDF` level with exact identities/DOIs.
|
||||
- [x] Observation separated from recommendation.
|
||||
- [x] Result order, comparator, field/signal metrics, cadence, numbers, limitations and transitions covered.
|
||||
- [x] Anti-patterns and coverage/evidence states included.
|
||||
- [x] No project prose, project number or new internal claim reproduced.
|
||||
- [x] Text readiness separated from artifact readiness; Legacy/V5 and M04/M06 boundaries preserved.
|
||||
|
||||
**Disposition:** `revise` — style research is complete with no content blocker, but Round-4 rules prohibit first-pass self-freezing. Required next action is coordinator second audit; display choices remain `discuss`.
|
||||
@@ -0,0 +1,306 @@
|
||||
# Round 4 section-style research — explicit learned-law/control-law section
|
||||
|
||||
**Lane:** Section-style research for the SR-law unit.
|
||||
**Status:** `researching` (style recommendation complete; this first-pass memo cannot freeze a section contract).
|
||||
**Function:** Decision support for a later section freeze and Round-5 drafting. This is not manuscript prose, internal evidence, or a claim authority.
|
||||
|
||||
## 1. Scope, question and bounded recommendation
|
||||
|
||||
### Scope and question
|
||||
|
||||
This memo studies how close *Journal of Fluid Mechanics* papers present an explicit learned or reduced law: how an equation enters the argument; how notation, imposed structure and physical translation are handled; what evidence arbitrates an offline or learned representation; where deletion-like tests are limited; how equations and figures alternate; and what claim ceiling the section itself makes visible.
|
||||
|
||||
External literature is used only to establish rhetorical and methodological practice. It establishes no result, priority, validity or artifact readiness for the present paper. The Round-3 SR dossier is used only to identify the later section's required jobs and scientific boundaries; its project-specific prose and numbers are not reproduced here.
|
||||
|
||||
### Bounded recommendation
|
||||
|
||||
Build the later SR section as a five-move argument:
|
||||
|
||||
1. define the surrogate estimand and its boundary;
|
||||
2. state the recovery/deployment contract, including imposed structure;
|
||||
3. display the executable law once and decode only notation, sign, action order and topology;
|
||||
4. let matched closed-loop execution arbitrate the law, with offline fit remaining diagnostic;
|
||||
5. present deletion as a bounded parent-relative perturbation and end on the field-level question exported to the next section.
|
||||
|
||||
Use one displayed symmetry/deployment map and one displayed canonical law. Place each before the evidence that depends on it. Prefer a compact law/symmetry display followed by a closed-loop comparison and then a deletion display or table; omit a direct PPO/SR/zero panel unless its acquisition roles, comparator and window are genuinely matched. The section may support a bounded control-skeleton interpretation, but not a governing law, mechanism, causality, necessity, sufficiency, uniqueness, optimality, robustness, stability, universality or transfer.
|
||||
|
||||
## 2. Coverage ledger
|
||||
|
||||
Evidence states are intentionally separate: `read` does not mean independently verified or artifact-audited; `mapped` does not mean read.
|
||||
|
||||
| Status | Source/input | Why in scope | Read level and evidence state |
|
||||
|---|---|---|---|
|
||||
| `read` | `ROUND4_TOP_DOWN/README.md` | Round-4 purpose, phase and status controls | Full repository document; not independently verified; no artifact audit applicable |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_MASTER_AGENT_PROMPT.md` | Required method, fixed ladder and deliverables | Full repository document; not independently verified; no artifact audit applicable |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_QUALITY_GATE.md` | Style-memo pass/revise/block criteria | Full repository document; not independently verified; no artifact audit applicable |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_GLOBAL_DECISION_LEDGER.md` | Ownership, claim ceiling, duplication and unresolved choices | Full repository document; not independently verified; no artifact audit applicable |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_DISPLAY_EQUATION_INVENTORY.md` | SR equation/display ownership and readiness boundary | Full repository document; not independently verified; no artifact audit applicable |
|
||||
| `read` | `ROUND2_TOP_DOWN/05_REVISED_STRUCTURE_FREEZE.md` | Frozen section ownership, Kármán/cloak-only positive SR scope and figure logic | Full repository report read directly; not independently verified; not artifact-audited |
|
||||
| `read` | `ROUND2_TOP_DOWN/06_SECTION_LEVEL_JFM_STYLE_GUIDE.md` | Prior JFM style synthesis and SR cadence | Full repository report; not independently verified; bibliography checked against the records below where relevant |
|
||||
| `read` | `ROUND3_BOTTOM_UP/00_MODULE_INTERFACE_LEDGER.md` | Module ownership, downstream readiness and optional-panel gate | Full repository report read directly; not independently verified; not artifact-audited |
|
||||
| `read` | `ROUND3_BOTTOM_UP/11_FINAL_CONSISTENCY_AUDIT.md` | Final ownership, scope, panel-readiness and author-synthesis disposition controls | Full repository report read directly; not independently verified; not artifact-audited |
|
||||
| `read` | `ROUND3_BOTTOM_UP/05_KARMAN_SR_LAW.md` | Required SR jobs, imports/exports and caveats | Full repository dossier; not independently verified; internal artifact audit explicitly outside this style lane |
|
||||
| `read` | N. Gautier, J.-L. Aider, T. Duriez, B. R. Noack, M. Segond & M. Abel, “Closed-loop separation control using machine learning”, *Journal of Fluid Mechanics* **770**, 442–457 (2015), DOI `10.1017/jfm.2015.95` | Closest explicit machine-learning control-law precedent | `full-PDF`; PDF available and read for law, objective, execution, figures, benchmark and limits; bibliographic metadata record checked; no independent reproduction |
|
||||
| `read` | G. Y. Cornejo Maceda, Y. Li, F. Lusseyran, M. Morzyński & B. R. Noack, “Stabilization of the fluidic pinball with gradient-enriched machine learning control”, *Journal of Fluid Mechanics* **917**, A42 (2021), DOI `10.1017/jfm.2021.301` | Closest fluidic-pinball learned-law and simpler-to-richer control hierarchy | `full-PDF`; PDF available and read for control hierarchy, explicit law, averaged-law test, figures and limits; metadata record checked; no independent reproduction |
|
||||
| `read` | J.-C. Loiseau, B. R. Noack & S. L. Brunton, “Sparse reduced-order modelling: sensor-based dynamics to full-state estimation”, *Journal of Fluid Mechanics* **844**, 459–490 (2018), DOI `10.1017/jfm.2018.147` | Closest sparse-equation/topology and model-selection precedent | `full-PDF`; PDF available and read for learned equations, constraints, model selection, validation and reconstruction cadence; metadata record checked; no independent reproduction |
|
||||
| `read` | J. Rabault, M. Kuchta, A. Jensen, U. Réglade & N. Cerardi, “Artificial neural networks trained through deep reinforcement learning discover control strategies for active flow control”, *Journal of Fluid Mechanics* **865**, 281–302 (2019), DOI `10.1017/jfm.2019.62` | Useful negative comparator: closed-loop learned policy without an explicit symbolic law | `full-PDF`; PDF available and read for action definition, policy evaluation, intervention, physical interpretation and limits; metadata record checked; no independent reproduction |
|
||||
| `mapped` | Other JFM papers already indexed in the Round-2 style guide | Wider context for results-section cadence | Identity and prior style notes mapped only in this lane; not reread and not used for new observations here |
|
||||
| `mapped` | Other Round-3 controls and audits beyond the two named controls and SR dossier read above | Wider coordinator context | Located through Round-4 controlling-input list; not read in this bounded style lane; no scientific inference drawn |
|
||||
| `not-covered` | Exhaustive coding of all figures, sentences and supplementary material in the four papers | Disproportionate to the bounded style question | No inference of universal JFM convention |
|
||||
| `not-covered` | Internal SR authorities, manifests, code, arrays and immutable artifacts | This lane must not establish or restate project results | No internal result, number, formula validity or panel readiness inferred |
|
||||
| `not-covered` | New literature search beyond the four closest records | Existing exact full-PDF set directly covers the requested dimensions | Recommendation is explicitly a bounded four-paper synthesis, not a journal-wide survey |
|
||||
| `external-inaccessible` | None essential | All four selected records had readable PDFs | No access blocker |
|
||||
|
||||
## 3. Exact literature identities and direct observations
|
||||
|
||||
### 3.1 Gautier et al. (2015): explicit law as an experimentally executed object
|
||||
|
||||
**Identity/read level.** N. Gautier, J.-L. Aider, T. Duriez, B. R. Noack, M. Segond & M. Abel, “Closed-loop separation control using machine learning”, *JFM* **770**, 442–457 (2015), DOI `10.1017/jfm.2015.95`; `full-PDF`.
|
||||
|
||||
**Observed practice.** The plant, scalar sensor and actuator are defined before the machine-learning-control section. The objective then receives its own equation before candidate laws are discussed. A law is consequently legible as a map between already physicalized quantities, not as a free-standing symbolic expression. Candidate topology is shown through expression trees, while practical admissibility is enforced by pre-evaluation that rejects saturating or extreme laws before plant testing. The selected law is presented together with its input-output graph and a closed-loop temporal response; spectral interpretation follows, and periodic forcing provides a separate benchmark. The conclusion retains an explicit non-optimality caveat.
|
||||
|
||||
**Observed claim ceiling.** Strong performance and robustness wording is attached to dedicated experimental comparisons across tested conditions, while optimality is withheld. The paper does not use coefficient readability alone as evidence of closed-loop value.
|
||||
|
||||
**Style lesson, not a transferred claim.** A concise explicit-law section earns interpretability through the sequence `defined variables → objective → law shape → executed response → frequency/physical reading → benchmark → envelope`. Admissibility checks and closed-loop tests are different rhetorical jobs and should not be collapsed.
|
||||
|
||||
### 3.2 Cornejo Maceda et al. (2021): hierarchy before the final learned law
|
||||
|
||||
**Identity/read level.** G. Y. Cornejo Maceda, Y. Li, F. Lusseyran, M. Morzyński & B. R. Noack, “Stabilization of the fluidic pinball with gradient-enriched machine learning control”, *JFM* **917**, A42 (2021), DOI `10.1017/jfm.2021.301`; `full-PDF`.
|
||||
|
||||
**Observed practice.** The results progress through nested control spaces: symmetric steady forcing, general steady forcing, then feedback control. The final learned law appears only after sensors, delayed inputs, objective and search procedure have been established. Its long component equations are displayed as the exact executable outcome, but interpretation shifts quickly away from coefficient-by-coefficient narration toward mean actuation, temporal content, flow organization and comparative performance. A test of the time-averaged control separates the value of the mean component from the remaining feedback fluctuations. A late synthesis figure compares the control-space hierarchy. The conclusion states that the feedback still lacks authority to achieve complete stabilization.
|
||||
|
||||
**Observed topology/symmetry handling.** Symmetry is explicit in the deliberately restricted steady search, not silently attributed to the richer feedback law. Increasing search-space freedom is itself part of the physical argument. The paper therefore distinguishes a chosen topology or actuation subspace from what the final optimizer discovers within a richer space.
|
||||
|
||||
**Observed deletion/ablation boundary.** The averaged-law intervention is a closed-loop comparator with a precise operational meaning. It is stronger than coefficient inspection, but it still tests one constructed replacement, not every possible decomposition of feedback value. The paper's own strong interpretation is supported by this dedicated comparison and remains bounded by incomplete stabilization.
|
||||
|
||||
**Style lesson, not a transferred claim.** If a learned law contains a persistent component plus fluctuations, test the simpler executed controller directly and explain exactly what was held or replaced. Introduce richer feedback as the answer to a limitation exposed by the simpler class; do not start with the most complex expression and reverse-engineer a hierarchy afterward.
|
||||
|
||||
### 3.3 Loiseau et al. (2018): topology, sparsity and validation as separate decisions
|
||||
|
||||
**Identity/read level.** J.-C. Loiseau, B. R. Noack & S. L. Brunton, “Sparse reduced-order modelling: sensor-based dynamics to full-state estimation”, *Journal of Fluid Mechanics* **844**, 459–490 (2018), DOI `10.1017/jfm.2018.147`; `full-PDF`.
|
||||
|
||||
**Observed practice.** A single framework schematic establishes the route from sensors to lifted coordinates, sparse dynamics, manifold mapping and full-state reconstruction before case equations appear. Learned differential and algebraic equations are introduced after their variables and candidate libraries are defined. Sparsity selection is not presented as self-validating: complexity is arbitrated with an information criterion, identified trajectories are compared with reference data, and full-state reconstruction is evaluated separately. A failed unconstrained identification motivates explicit physical constraints rather than being hidden. Limitations of the preferred reconstruction method, including storage cost, stay beside its accuracy advantage.
|
||||
|
||||
**Observed topology/deletion handling.** Term removal is embedded in sparse model selection, not described as causal ablation. Constraints may encode known physical properties; the text identifies which structure is imposed and which is identified. Model-form selection, temporal validation and full-state reconstruction are distinct evidentiary layers.
|
||||
|
||||
**Style lesson, not a transferred claim.** State imposed structure before interpreting a sparse equation. A compact topology is a selected representation, not proof that every retained term is physically necessary. If a constraint repairs a known failure, show the failure and name the constraint; this increases credibility more than presenting the constrained equation as spontaneous discovery.
|
||||
|
||||
### 3.4 Rabault et al. (2019): why policy interpretation is not an explicit-law section
|
||||
|
||||
**Identity/read level.** J. Rabault, M. Kuchta, A. Jensen, U. Réglade & N. Cerardi, “Artificial neural networks trained through deep reinforcement learning discover control strategies for active flow control”, *Journal of Fluid Mechanics* **865**, 281–302 (2019), DOI `10.1017/jfm.2019.62`; `full-PDF`.
|
||||
|
||||
**Observed practice.** Geometry, probes, action normalization, zero-net-mass-flow constraint and action smoothing are stated before performance. The neural policy itself remains a black box; the paper interprets controlled trajectories and mean/statistical fields rather than printing network weights as a law. Evaluation uses deterministic policy execution after training. Suppressing control after a controlled state is reached tests persistence of that state, while reduced-probe studies test observation sufficiency under named alternatives. Figures progress from setup and learning consistency to time histories, instantaneous flow and statistical fields. Dimensionality, Reynolds-number scope, probe-position sensitivity and black-box interpretability are visibly bounded.
|
||||
|
||||
**Observed claim ceiling.** Physical interpretation is attached to observed controlled-flow changes and intervention, not to the internal topology of the neural network. The policy is not recast as a symbolic governing equation.
|
||||
|
||||
**Style lesson, not a transferred claim.** Keep the parent policy and the explicit surrogate conceptually separate. A surrogate section must not borrow the parent's field interpretation merely because both execute in closed loop. Likewise, turning control off is a different intervention from deleting one surrogate term; the two should not share “ablation” language without qualification.
|
||||
|
||||
## 4. Cross-paper observed practice
|
||||
|
||||
The following statements are observations across this bounded set, not universal rules for JFM.
|
||||
|
||||
### Equation introduction
|
||||
|
||||
- Physical variables, action bounds and objective precede the learned equation.
|
||||
- A displayed law enters when the reader can already answer: input from what state, output to which actuator, under what normalization, and judged by which objective or comparator?
|
||||
- Long formulas are not mined coefficient by coefficient. Authors move from exact expression to lower-dimensional behavior: law shape, mean actuation, frequency content, trajectory or field consequence.
|
||||
- Equations that do not carry a later comparison are minimized; a framework schematic often removes the need for repeated algebra.
|
||||
|
||||
### Notation
|
||||
|
||||
- Symbols are attached to physical observables and actuators before algorithmic detail.
|
||||
- Normalization, index order and delays are defined locally enough to read the law without consulting an implementation inventory.
|
||||
- Present tense is used for definitions and equation meaning; past tense is used for optimization, execution and measured outcomes.
|
||||
- The executable map and the objective use separate notation. Fit or search notation is not allowed to obscure plant semantics.
|
||||
|
||||
### Symmetry and topology
|
||||
|
||||
- Restricted actuation spaces, expression-tree libraries, sparse candidate libraries and physical equality constraints are disclosed as choices.
|
||||
- Imposed structure is not narrated as a discovered property.
|
||||
- A failed or weaker unconstrained/simple class can motivate a richer or constrained law.
|
||||
- Topological compactness supports readability and model selection; by itself it does not establish physical truth, necessity or causality.
|
||||
|
||||
### Closed-loop arbitration
|
||||
|
||||
- The learned object is judged in execution: direct plant evaluation, deterministic policy rollout, reference-trajectory comparison or reconstruction against held reference data.
|
||||
- Admissibility/static checks, offline fit, search convergence and closed-loop performance answer different questions.
|
||||
- A benchmark is introduced only after the law has a declared objective and execution protocol.
|
||||
- Simpler executed controls provide more interpretable comparisons than symbolic coefficient inspection, provided replacement and acquisition semantics are explicit.
|
||||
|
||||
### Deletion and ablation caveats
|
||||
|
||||
- None of the four papers licenses a generic rule that deleting a term proves causality or necessity.
|
||||
- Sparse thresholding is model selection; control suppression tests continued actuation; averaged-law replacement tests one simpler controller; reduced sensors test an observation set. These interventions are non-equivalent.
|
||||
- Stronger words are earned only when the experiment is designed for them. Even then, finite conditions and unresolved authority/optimality limitations remain local.
|
||||
|
||||
### Figure cadence
|
||||
|
||||
A recurrent cadence is:
|
||||
|
||||
`contract/schematic → explicit equation or law graph → executed time/trajectory evidence → flow or spectral interpretation → simpler/benchmark comparison → limitation or synthesis`.
|
||||
|
||||
The equation and its first evidence are close. A summary figure arrives after the hierarchy is established, not as an opening gallery. One figure carries one main inferential job; secondary implementation details remain in captions, methods or appendices.
|
||||
|
||||
### Claim ceilings
|
||||
|
||||
- “Law” denotes an executable mapping or identified reduced dynamics under a specified contract; it is not automatically a governing physical law.
|
||||
- “Robust”, “stable” and “optimal” appear only with purpose-built tests, and explicit caveats still narrow them.
|
||||
- Interpretation of observed flow organization can be stronger than interpretation of symbolic coefficients when supported by executed comparisons.
|
||||
- Failure and incomplete authority are reported where they delimit the result, not postponed to a generic limitations paragraph.
|
||||
|
||||
## 5. Recommendation for the later SR section
|
||||
|
||||
This section translates observed practice into a bounded recommendation. It does not prescribe internal findings.
|
||||
|
||||
### Paragraph jobs and cadence
|
||||
|
||||
1. **Question and estimand.** Define the SR object as an observation-to-action surrogate under one declared plant and role contract. State immediately that it is neither a flow-field model nor a causal model. Import action order, normalization, roles and metric semantics rather than redefining them.
|
||||
2. **Recovery and deployment contract.** Give causal pairing, search/refit hierarchy, imposed symmetry and executable/static gates. Separate discovered topology from imposed deployment structure. Move exhaustive libraries, run history and package provenance to a ledger or supplement.
|
||||
3. **Canonical law.** Display it once. Decode variables, action order, signs, odd/shared mapping and persistent versus observation-dependent structure. Do not narrate every coefficient or call the expression a governing law.
|
||||
4. **Closed-loop arbitration.** State which acquisition actually compares which roles under which window. Put offline fit in a subordinate diagnostic role. If the available formula-parent execution and a standardized multi-role collection are different acquisitions, present them as different evidence contracts rather than assembling a synthetic comparison.
|
||||
5. **Deletion and limit.** Report parent-relative fixed-window term tests as a ranking of tested perturbations. Put `not causal ablation; not necessity, sufficiency or uniqueness` in the same paragraph and caption. End by exporting the unresolved field-accounting question, not by claiming mechanism.
|
||||
|
||||
### Equation and notation contract
|
||||
|
||||
- **Equation 1: symmetry/deployment map — `required`.** Owner: the Kármán SR section (Round-3 M05). Job: expose native action order and every imposed transformation before the canonical law. Caption/text ceiling: “defines the deployed mapping”; it cannot prove parent-policy equivariance, plant symmetry or discovery of symmetry. Text readiness: supported by the R3 dossier. Artifact readiness: no separate schematic should be claimed without production/provenance; the equation itself is sufficient fallback.
|
||||
- **Equation 2: canonical executable law — `required`.** Owner: the Kármán SR section (Round-3 M05). Job: provide one exact, reusable law identity and permit a bounded structural translation. Caption/text ceiling: executable surrogate under the declared contract, not governing law, mechanism or unique representation. The Round-3 interface ledger and final audit mark the equation/text ready for author synthesis; this style memo does not independently verify or artifact-audit that readiness.
|
||||
- Define observation aggregates before or directly beneath the law; never force the reader to reconstruct them from subscripts.
|
||||
- Keep parent-policy notation, surrogate notation, physical angular velocity and normalized action distinct.
|
||||
- Do not repeat reusable Reynolds, field-L2 or DTW equations owned by Contracts. A short metric name and exact imported semantics are enough.
|
||||
|
||||
### Closed-loop arbitration contract
|
||||
|
||||
- **Primary rhetorical arbiter:** executed surrogate under a named plant, comparator and evaluation window.
|
||||
- **Subordinate diagnostics:** offline action fit, symbolic complexity, search convergence, static finiteness/schema checks.
|
||||
- **Direct multi-role display — `blocked if retained`.** Include only if target, parent policy, surrogate and zero roles share the required acquisition protocol and their comparator/window provenance is explicit. Otherwise omit the panel; this optional artifact gate does not block bounded section text.
|
||||
- Failed formulas or gates should be visible when they explain why the displayed topology was retained. They need not become a package history.
|
||||
|
||||
### Deletion contract
|
||||
|
||||
- Prefer “parent-relative term deletion” or “fixed-window deletion test” over unqualified “ablation”.
|
||||
- State parent, replacement/deletion, held-fixed conditions, metric, window and tested set.
|
||||
- Rank only the tested perturbations. Do not convert a larger change into causal contribution, physical necessity, universal sensitivity or a decomposition of total performance.
|
||||
- If deleting one term changes the state distribution, acknowledge that the result is a new closed-loop trajectory, not a local coefficient derivative.
|
||||
- **Deletion display/table — `required` if the term-ranking claim is retained.** Job: make tested variants and caveats auditable. Caption ceiling: comparative degradation among tested parent-relative variants. Fallback: concise table; no decorative bar chart is needed.
|
||||
|
||||
### Figure cadence and density
|
||||
|
||||
Recommended minimum sequence:
|
||||
|
||||
1. symmetry/deployment equation and canonical law;
|
||||
2. compact closed-loop arbitration display or table only if its artifact gate passes;
|
||||
3. compact deletion table/figure.
|
||||
|
||||
A law graph or symmetry schematic is optional if it resolves sign/order ambiguity that equations cannot. Avoid placing a field figure in this section merely to strengthen the law; field accounting belongs to the following owner. Keep one principal inference per display and put the strongest limitation in the adjacent paragraph and caption.
|
||||
|
||||
### Imports, exports and bridge
|
||||
|
||||
**Imports:** plant and chain identity; action order/sign/normalization; observation definitions; role/comparator vocabulary; field-L2 versus DTW distinction; the bounded Kármán performance handoff.
|
||||
|
||||
**Owns/exports:** the Kármán SR section (Round-3 M05) owns the exact executable law identity, imposed symmetry/deployment topology, distinction between offline diagnostics and closed-loop arbitration, and bounded fixed-window deletion ranking; it exports the question passed to field-level accounting. Steady sign/magnitude context remains owned by the steady section and must not be used to imply amplitude agreement or chain continuation.
|
||||
|
||||
**Does not own:** full parent-policy performance; full field matching; mean/dynamic field attribution; CCD; steady-chain results; positive cross-case SR interpretation.
|
||||
|
||||
**Opening dependency:** the preceding performance section must have already established why action-law complexity is the next question without implying that performance reveals law structure.
|
||||
|
||||
**Closing bridge:** convert action-space structure and deletion ranking into a field-accounting question: what portion, if any, of independently defined field improvement is associated with persistent and correction components?
|
||||
|
||||
### Duplication deletions
|
||||
|
||||
- Delete repeated plant, Reynolds, action and metric equations already owned by Contracts.
|
||||
- Delete full performance numbers and field comparisons owned by neighboring sections.
|
||||
- Delete search-stage chronology, package names, hashes and library dumps from narrative; retain only details needed to distinguish fit, gate and execution.
|
||||
- Delete any steady-result detail beyond a later, explicitly separate contextual bridge.
|
||||
- Delete positive Illusion SR language and any import of CCD as confirmation of the law.
|
||||
|
||||
## 6. Claim ceiling and near-collision audit
|
||||
|
||||
### Green
|
||||
|
||||
Permissible style-level forms, conditional on later authority verification:
|
||||
|
||||
- “defines the executable surrogate and its imposed deployment map”;
|
||||
- “was evaluated in closed loop under the named contract”;
|
||||
- “the tested deletion produced the largest/smaller change among the listed variants”;
|
||||
- “offline fit is diagnostic; closed-loop execution is the arbiter.”
|
||||
|
||||
### Amber
|
||||
|
||||
Use only with the full local qualifier:
|
||||
|
||||
- “supports a bounded control-skeleton interpretation under the declared Kármán contract”;
|
||||
- “organizes the law into a persistent component and smaller observation-dependent corrections”;
|
||||
- “is consistent with”, when the comparison and non-equivalence are named.
|
||||
|
||||
### Prohibited red without new dedicated evidence
|
||||
|
||||
`governing law`, `mechanism`, `causes`, `necessary`, `sufficient`, `unique`, `optimal`, `stable`, `robust`, `universal`, `transferable`, `general Reynolds law`, `independent validation`, `term contribution to the flow`, `PPO symmetry discovered`, `DTW delay`, or positive Illusion SR efficacy/generalization.
|
||||
|
||||
### Near-collision findings
|
||||
|
||||
| Pair | Acquisition/roles/domain/window distinction | Classification and wording consequence |
|
||||
|---|---|---|
|
||||
| Offline parent-action fit vs surrogate closed-loop execution | Fixed parent-visited states versus a surrogate-generated trajectory; action error versus plant outcome | `different estimand`; fit cannot be called control success |
|
||||
| Static/schema/short execution gate vs standard closed-loop comparison | Admissibility/finiteness and short-horizon behavior versus declared evaluation roles/window | `related diagnostic`; do not call a short gate validation of the full result |
|
||||
| Formula-parent execution vs standardized target/policy/surrogate/zero acquisition | Potentially different role acquisition and windows even in the same plant | `different estimand` unless exact authority establishes sameness; no synthetic panel or arithmetic |
|
||||
| Parent-relative deletion vs causal ablation | New closed-loop trajectory after a controller intervention; no randomized/local causal estimand | `different estimand`; use deletion/ranking language only |
|
||||
| Persistent action magnitude vs separate steady-chain actuation | Different chain, solver/geometry, role, metric and temporal contract | `related contextual diagnostic` at most; no agreement, validation or amplitude equivalence |
|
||||
| SR law vs field/CCD organization | Action mapping versus separately defined field residual/subspace, potentially sharing parent lineage | `different estimand`; no independent triangulation, causality or identification of law terms with modes |
|
||||
| Control suppression vs term deletion | Whole-controller removal after a controlled state versus modification of one surrogate component | `different estimand`; do not generalize “ablation” conclusions across them |
|
||||
|
||||
## 7. Anti-patterns
|
||||
|
||||
- Opening with the full formula before defining input, output, normalization, action order and objective.
|
||||
- Calling imposed oddness, shared topology or reflection a discovered parent-policy symmetry.
|
||||
- Treating a sparse expression as self-interpreting and narrating every coefficient.
|
||||
- Presenting offline action RMSE, Pareto rank or search convergence as closed-loop control evidence.
|
||||
- Combining different acquisitions into a PPO/SR/zero comparison because their labels look compatible.
|
||||
- Using `ablation` without saying whether the intervention deletes a term, averages a controller, removes control, changes sensors or changes the model library.
|
||||
- Inferring necessity, causality or mechanism from parent-relative deletion.
|
||||
- Calling a short gate, a spatial L2 metric or a DTW score interchangeable forms of validation.
|
||||
- Importing field panels, CCD modes or steady results to manufacture independent confirmation.
|
||||
- Hiding failed candidates, imposed constraints or incomplete authority until the final Discussion.
|
||||
- A figure gallery: formula, traces, spectra, fields and bars with no single comparator-led argument.
|
||||
- Repeating equations owned by Contracts or narrating package stages, hashes and artifact inventories in prose.
|
||||
- Borrowing strong words such as robust, stable or optimal from literature without purpose-built internal tests.
|
||||
- Treating an explicit surrogate as the parent neural policy or as a physical governing law.
|
||||
|
||||
## 8. Decisions, deferrals and disposition
|
||||
|
||||
### Evidence-governed recommendations
|
||||
|
||||
- Use the five-move cadence and two required SR-owned equations.
|
||||
- Make imposed structure explicit before interpretation.
|
||||
- Let matched closed-loop execution arbitrate; subordinate offline fit and static gates.
|
||||
- Use parent-relative deletion language with same-location caveats.
|
||||
- Keep field/CCD and steady evidence outside this section except for narrow imports/bridges.
|
||||
- Enforce the bounded Kármán/cloak-only claim ceiling.
|
||||
|
||||
### Author/coordinator choices
|
||||
|
||||
- **Direct multi-role panel:** omit unless exact same-protocol role artifacts and comparator/window values are bound. The Round-3 final audit classifies the retained panel as blocked but the bounded text as ready; omission leaves the core law/deletion argument intact.
|
||||
- **Law graph or symmetry schematic:** safe default is equations only. Add a schematic only if action order/sign cannot be read quickly; it remains a production artifact, not existing evidence.
|
||||
- **Deletion visualization:** safe default is a compact table rather than bars, because the table can carry parent, variant, window and caveat without implying additive contribution.
|
||||
|
||||
### Deferrals and blockers
|
||||
|
||||
- Internal formula identity, coefficients, run lineage, comparator roles, windows, hashes and panel arrays were deliberately not audited in this style lane. Any retained manuscript detail or panel must be checked against current authorities during later assembly; the Round-3 readiness disposition remains controlling.
|
||||
- The optional direct multi-role panel is blocked if retained until matched acquisition is established; this does not block bounded section text.
|
||||
- No literature access blocker exists.
|
||||
- No first-pass style memo may declare the section frozen.
|
||||
|
||||
### Disposition
|
||||
|
||||
`researching` — the bounded style research is complete and traceable. The Round-3 controls already mark the SR equation/text ready for author synthesis and the optional standardized panel blocked if retained; those controls do not authorize this first-pass Round-4 memo to self-freeze. A later coordinator disposition, not this memo, may assign `frozen` status.
|
||||
|
||||
## 9. Validation against the Round-4 style gate
|
||||
|
||||
- Exact identities, publication venues, years, DOIs and `full-PDF` read levels are recorded for four selected papers.
|
||||
- Observed practice is separated from recommendation and no external paper is used as internal evidence.
|
||||
- Equation introduction, notation, imposed symmetry/topology, closed-loop arbitration, deletion caveats, figure cadence and claim ceilings are covered explicitly.
|
||||
- Coverage states, independent-verification state and artifact-audit state remain distinct.
|
||||
- Imports, ownership, exports, bridge and duplication deletions are explicit.
|
||||
- Display/equation recommendations state owner, job, readiness/dependency, caption ceiling and fallback.
|
||||
- Near-collisions are classified without pooling or replication/validation/agreement language.
|
||||
- No project result, coefficient, performance number or new scientific claim is introduced.
|
||||
- Gate-1 self-check: **PASS candidate for coordinator review**, not self-freeze.
|
||||
@@ -0,0 +1,195 @@
|
||||
# Round 4 style research — Field decomposition and CCD
|
||||
|
||||
**Lane:** section-style research only; not manuscript prose, evidence generation, artifact approval, or a section freeze.
|
||||
**Disposition:** `researching` — completed first-pass memo for coordinator audit.
|
||||
|
||||
## 1. Scope, question, and bounded recommendation
|
||||
|
||||
**Question.** How do close *Journal of Fluid Mechanics* papers order and qualify sections that progress from mean/reference fields through a dynamic residual to modal or correlation-based interpretation?
|
||||
|
||||
**Recommendation.** Use the narrowing sequence
|
||||
|
||||
`four-role contract → mean fields → scalar error-difference accounting → local caveat → separately centred residual → action-correlated modal organization → same-object POD boundary → non-transfer bridge`.
|
||||
|
||||
The scalar estimand and comparator must precede modes. Put the mean display before the residual/mode display. Display only equations that define an immediately used object: scalar shares, the non-paired residual, and an optional compact CCD operator. Introduce POD after CCD as a boundary on distinct field information, not a superiority contest. This memo adds no project prose, numbers, or claims; external literature supplies style precedent only.
|
||||
|
||||
## 2. Coverage ledger
|
||||
|
||||
`read` is not independently verified or artifact-audited; `mapped` is not read.
|
||||
|
||||
| Status | Source/input | Why in scope | Evidence state/reason |
|
||||
|---|---|---|---|
|
||||
| `read` | Round-4 `README.md`, `00_MASTER_AGENT_PROMPT.md`, `00_QUALITY_GATE.md`, `00_GLOBAL_DECISION_LEDGER.md`, `00_DISPLAY_EQUATION_INVENTORY.md` | Controls, ownership, statuses, display/equation gates | Read directly; workflow/decision authority, not scientific verification |
|
||||
| `read` | R2 `05_REVISED_STRUCTURE_FREEZE.md` and `06_SECTION_LEVEL_JFM_STYLE_GUIDE.md` | Author-ratified architecture and sampled JFM practice | Read directly; style guide is `full-report` |
|
||||
| `read` | R3 `06_KARMAN_FIELD_CCD.md`, `00_MODULE_INTERFACE_LEDGER.md`, `11_FINAL_CONSISTENCY_AUDIT.md` | M06 estimands, ownership, near-collisions, readiness | Read directly as repository reports; no recomputation or artifact audit |
|
||||
| `read` | Jean-Christophe Loiseau, Bernd R. Noack & Steven L. Brunton, “Sparse reduced-order modelling: sensor-based dynamics to full-state estimation”, *JFM* 844, 459–490 (2018), DOI `10.1017/jfm.2018.147` | Closest reconstruction → correlation modes → POD precedent | `full-PDF`; full-text method/results/modal passages inspected, not equation-by-equation verified |
|
||||
| `read` | Nan Deng, Bernd R. Noack, Marek Morzyński & Luc R. Pastur, “Low-order model for successive bifurcations of the fluidic pinball”, *JFM* 884, A37 (2020), DOI `10.1017/jfm.2019.959` | Plant-specific mean/fluctuation decomposition and basis choice | `full-PDF`; decomposition and POD passages inspected, not equation-by-equation verified |
|
||||
| `read` | Guy Y. Cornejo Maceda, Yiqing Li, François Lusseyran, Marek Morzyński & Bernd R. Noack, “Stabilization of the fluidic pinball with gradient-enriched machine learning control”, *JFM* 917, A42 (2021), DOI `10.1017/jfm.2021.301` | Scalar objective → law → mean field → hierarchy precedent | `full-PDF`; result-order and field-interpretation passages inspected, not equation-by-equation verified |
|
||||
| `mapped` | Bernd R. Noack, Witold Stankiewicz, Marek Morzyński & Peter J. Schmid, “Recursive dynamic mode decomposition of transient and post-transient wake flows”, *JFM* 809, 843–872 (2016), DOI `10.1017/jfm.2016.678` | Close POD/DMD comparison record | Exact record/abstract mapped; full-text retrieval failed, so no section-practice inference is assigned |
|
||||
| `mapped` | Remaining JFM papers indexed in the R2 style guide; current CCD manifests/publication siblings named by M06 | Broader cadence and panel lineage | Mapped through controlling reports; not reread or artifact-audited here |
|
||||
| `not-covered` | Raw arrays, masks, hashes, numerical arithmetic, new CFD/training/SR/CCD/OID, figure generation | Outside style lane and explicitly prohibited | No claim inferred; authority-led deferral |
|
||||
| `not-covered` | Exhaustive sentence coding of all JFM modal literature | Three closest full PDFs are proportionate | Findings are sampled practice, not universal rules |
|
||||
| `external-inaccessible` | Noack et al. (2016) full text during this pass | Retrieval failed | Non-blocking; record remains `mapped` |
|
||||
| `external-inaccessible` | Hidden/external project payloads | Unnecessary for style analysis | `artifact-audit deferred`; retained panels remain gated |
|
||||
|
||||
Requested controls, R2 style guidance, R3 M06, and relevant audits were read. Three close JFM papers were inspected at full-PDF level and one additional exact record was mapped.
|
||||
|
||||
## 3. Observed practice versus recommendation
|
||||
|
||||
### 3.1 Role ordering
|
||||
|
||||
**Observed practice.** Loiseau et al. order sensor dynamics and predictive performance before full-state reconstruction, then use feature-correlated modes as an interpretability layer. Deng et al. establish phenomenology before mean/fluctuation decomposition and refine the representation where a simpler model fails. Cornejo Maceda et al. repeatedly couple scalar objective, explicit law, and corresponding fields within an increasingly rich control hierarchy.
|
||||
|
||||
**Recommendation.** Begin with target, zero, constant, and DRL roles under one exact comparison contract. Show means and scalar accounting before defining the residual; introduce CCD only after centring and non-pairing are explicit; end with POD on the same residual object. This delays attractive modes but prevents them from being mistaken for the primary performance estimand or a causal decomposition.
|
||||
|
||||
### 3.2 Scalar versus modal estimands
|
||||
|
||||
**Observed practice.** Loiseau et al. distinguish reconstruction accuracy from modal interpretability and note that a less accurate modal expansion may still aid interpretation. Deng et al. make basis choice conditional on the represented regime: transient-near-base and asymptotic-limit-cycle estimands need not favour the same modes. Cornejo Maceda et al. use fields to interpret an optimized outcome rather than silently redefine its scalar cost.
|
||||
|
||||
**Recommendation.** Keep three propositions separate: means are compared under one role/ROI/mask/window contract; scalar error differences provide arithmetic accounting; CCD organizes a separately defined residual. Never call scalar shares modal energy, explained variance, modal contribution, or actuator attribution. Never use a scalar mean ledger to validate dynamic reconstruction.
|
||||
|
||||
### 3.3 Equations
|
||||
|
||||
**Observed practice.** In Deng et al., decomposition equations arrive where mean-field assumptions begin, after phenomenology motivates them. Loiseau et al. stage equations within the method and let comparative figures carry empirical burden. Cornejo Maceda et al. display a law when it becomes the object interpreted by scalar and field evidence.
|
||||
|
||||
**Recommendation.**
|
||||
1. **Scalar-share identity:** required only if shares are retained; place after mean errors and before interpretation. Define numerator/denominator and state “not energy or causality”.
|
||||
2. **Separately centred residual:** required; place immediately before residual/mode evidence and state non-pairing in the same paragraph.
|
||||
3. **CCD operator:** optional; retain only if its field side, action side, weighting, or rank is needed for the adjacent interpretation. Otherwise use prose and move derivation to supplement.
|
||||
|
||||
Do not duplicate reusable L2/action definitions owned by Contracts, the SR law owned by M05, a general POD derivation, or OID equations.
|
||||
|
||||
### 3.4 Figure sequence
|
||||
|
||||
**Observed practice.** Loiseau et al. proceed from method/prediction to reconstruction comparisons, estimator errors, and generalized modes. Deng et al. show phenomenology before model construction and connect modes back to the bifurcation narrative. Cornejo Maceda et al. pair law hierarchy with time/scalar behaviour and mean fields, then synthesize the hierarchy. Figures advance an inference rather than form mode galleries.
|
||||
|
||||
**Recommendation.**
|
||||
1. Mean comparison: four roles on one scale plus compact scalar ledger.
|
||||
2. Residual definition: make separate centring and non-pairing legible.
|
||||
3. Mode identity and phase organization: selected modes/action coordinates only.
|
||||
4. POD boundary: compact same-object subspace comparison, not a second gallery.
|
||||
|
||||
One paragraph should carry one principal figure claim. Captions own role, component, normalization/orientation, domain, arbitrary sign, lineage, and noncausal ceiling. If panels are unavailable, retain the scalar ledger, residual equation, and bounded comparison; do not invent redraws.
|
||||
|
||||
### 3.5 POD and other methods
|
||||
|
||||
**Observed practice.** Loiseau et al. state what generalized modes correlate with, compare reconstruction against POD alternatives, and expose modal-deformation limits. Deng et al. choose eigenmodes or POD modes according to transient versus asymptotic purpose. The mapped Noack et al. record frames RDMD against POD/DMD, but unread sections are not evidence here.
|
||||
|
||||
**Recommendation.** Compare CCD and POD only on identical residual snapshots, rank, weighting, and domain. Ask whether tested field subspaces differ and what action association adds. If subspaces are near-equivalent, say CCD is not shown to provide a distinct field basis and adds action-coordinate association/phase organization under the declared construction. Do not use `outperforms`, `unique modes`, `more physics`, or `explains more variance` without exact dedicated authority. Add DMD/SPOD/resolvent/OID only for a live boundary question, never as a catalogue.
|
||||
|
||||
### 3.6 Descriptive caveats
|
||||
|
||||
Place each limit beside the object it limits:
|
||||
|
||||
- scalar shares: arithmetic accounting, not energy or causal contribution;
|
||||
- residual: separately centred and non-paired, not a matched counterfactual or response;
|
||||
- CCD: descriptive action correlation, not lag, direction, response time, authority, or mechanism;
|
||||
- modes: global sign arbitrary; rotation/parity views do not measure wall vorticity;
|
||||
- POD comparison: one parent lineage, not independent validation or triangulation;
|
||||
- synthesis: Legacy Kármán only; no V5, steady, Vortex, Illusion, or OID transfer.
|
||||
|
||||
A final disclaimer cannot repair causal verbs used earlier.
|
||||
|
||||
### 3.7 Bridges
|
||||
|
||||
**Observed practice.** Loiseau et al. turn predictive success into reconstruction and then interpretability questions. Deng et al. let phenomenology and representational failure motivate refinement. Cornejo Maceda et al. let each richer control space pose the next question.
|
||||
|
||||
**Recommendation.**
|
||||
- SR → mean: term ranking asks what the constant reference accounts for at field level; it does not predict the answer.
|
||||
- Mean → residual: scalar accounting leaves a field difference whose organization is unresolved.
|
||||
- Residual → CCD: once centring/non-pairing are fixed, ask whether action-associated low-rank organization is present.
|
||||
- CCD → POD: association does not establish a distinct basis, motivating a same-object boundary test.
|
||||
- Closing bridge: cross-case temporal/target contracts change and cannot inherit this Kármán interpretation.
|
||||
|
||||
## 4. Paragraph jobs, ownership, and deletion
|
||||
|
||||
### Six paragraph jobs
|
||||
|
||||
1. Exact four roles and comparison contract; import definitions compactly.
|
||||
2. Mean fields, scalar errors, and error-difference identity.
|
||||
3. Local scalar limit and one concise M04/M06 different-estimand warning.
|
||||
4. Residual definition, separate centring, non-pairing, and dynamic question.
|
||||
5. CCD rank/action coordinates/phase as **action-correlated correction-field modes**, with local caveats.
|
||||
6. Same-object POD boundary, single-lineage limit, bounded export, and non-transfer bridge.
|
||||
|
||||
If compressed, combine 2–3; never combine mean and modal estimands in the opening paragraph.
|
||||
|
||||
**Owned here:** corrected four-role mean comparison; scalar error-difference semantics; separately centred non-paired residual; descriptive CCD organization; same-object POD boundary; M04/M06 near-collision rule.
|
||||
|
||||
**Imports:** chain/action/Re/metric definitions from Contracts; detailed independent field matching from M04; SR formula/deletion ranking from M05. Imports orient but do not duplicate.
|
||||
|
||||
**Exports:** bounded mean accounting plus smaller descriptively action-correlated residual organization under the named Legacy Kármán contract; explicit non-transfer; no positive OID export.
|
||||
|
||||
**Deletion instructions:** remove full SR formula/deletion detail, M04 values/seed/DTW prose, repeated package hashes/dimensions, general POD derivation, and OID results. State the full M04/M06 prohibition once here; later sections keep only a reminder.
|
||||
|
||||
## 5. Claim ceiling and near-collision audit
|
||||
|
||||
**Green:** `defines`, `compares`, `separates`, `uses separate centring`, `is non-paired`, `is associated with`, `spans the tested rank`.
|
||||
|
||||
**Amber:** `accounts for the measured error difference`, `organizes the residual`, `is action-correlated`, `is phase-organized`, `adds action-coordinate association`, `supports a bounded persistent-plus-dynamic reading`—always with exact contract and adjacent caveat.
|
||||
|
||||
**Red without dedicated evidence:** `causes`, `drives`, `responds`, `control authority`, `mechanism`, `necessary/sufficient`, `unique`, `optimal`, `stable`, `robust/general`, `transferable`, `restores momentum`, CCD `energy captured`/`explained variance`, `independently validates`, `triangulates`, or `CCD superiority`.
|
||||
|
||||
| Comparison | Contract differences | Classification / rule |
|
||||
|---|---|---|
|
||||
| M04 standardized PPO fields vs M06 corrected means | Independent three roles versus four-role target/zero/constant/DRL; geometry guard versus exact common mask; unweighted normalized RMS versus coordinate-weighted nondimensional RMS; different retained windows | **Different estimands.** No arithmetic, replication, validation, agreement, closure, or triangulation |
|
||||
| M06 scalar shares vs M06 modal quantities | Scalar mean target-error differences versus weighted residual/action decomposition; mean window versus centred cycle/phase object | **Different estimands.** Shares are not energy, variance, or modal contribution |
|
||||
| M05 SR ranking vs M06 action coordinates | Observation-to-action closed-loop/deletion contract versus field/action residual association | **Related diagnostic.** Brief descriptive bridge; no derivation/equivalence/validation |
|
||||
| CCD vs POD | Valid comparison requires identical residual, samples, rank, weighting, and domain; action-informed versus action-agnostic bases | **Same field object, different basis estimand.** May bound distinctness, not confer causality/independence |
|
||||
| Legacy M06 vs V5 steady/cross-cases | Different chain, roles, fields, metrics, and temporal contracts | **Different estimands.** Context/non-transfer only |
|
||||
|
||||
## 6. Display and equation contracts
|
||||
|
||||
No inventory entry implies an existing publication-ready artifact.
|
||||
|
||||
| Item | Owner/job/source | Ceiling | Readiness, dependency, fallback |
|
||||
|---|---|---|---|
|
||||
| Mean comparison + scalar ledger | Field/CCD; establish four roles and arithmetic accounting; exact M06 role arrays only | Registered mean comparison; no causality, energy, mechanism, necessity, or M04 validation | `dependency-gated`; text ready, redraw needs arrays/mask/window/scale/orientation/provenance. Fallback: scalar ledger + prose |
|
||||
| Scalar-share equation | Field/CCD; define retained shares from authority-bound scalar errors | Measured error-difference accounting only | `required` if shares retained; text ready, artifact audit deferred. Fallback: verbal definition/table; no unexplained percentages |
|
||||
| Residual equation | Field/CCD; define separately centred non-paired object from exact DRL cycles/constant template | Not counterfactual, response, force, or lag | `required`; text ready, panels gated. Fallback: equation plus explicit non-pairing sentence |
|
||||
| Compact CCD operator | Field/CCD; clarify field/action sides, weighting/rank if necessary | Descriptive correlation organization | `optional`; include only after equation-budget test. Fallback: prose/supplement |
|
||||
| Residual/modes/phase figure | Field/CCD; corrected parent/derived arrays and same-boundary action observable | Action association/phase; arbitrary sign; no wall-vorticity, causality, lag, or response time | `dependency-gated`; needs arrays/orientation/labels/lineage/render audit. Fallback: residual equation + bounded rank/coordinate statement |
|
||||
| Same-object POD boundary | Field/CCD; identical residual/rank/weight/domain | Tested subspace boundary; no superiority, uniqueness, independence, or CCD explained variance | `dependency-gated`; needs paired arrays/comparison metric. Fallback: sentence or compact row |
|
||||
| Estimand-boundary table | Field/CCD with Contracts terminology; prevent unsafe compression | Taxonomy, not convergence | `optional`, text-ready. Fallback: local caveat sentences |
|
||||
|
||||
## 7. Choices, deferrals, blockers, and anti-patterns
|
||||
|
||||
### Evidence-governed decisions
|
||||
|
||||
Mean precedes modes; scalar and modal estimands stay separate; non-pairing appears at first residual use; CCD remains descriptive and uses the fixed term; POD is a same-object boundary; M04 and M06 remain different estimands.
|
||||
|
||||
### Author/coordinator choices
|
||||
|
||||
| Choice | Recommendation and safe default | Status |
|
||||
|---|---|---|
|
||||
| Display CCD operator | Omit unless symbols materially clarify interpretation; prose/supplement fallback | `discuss` |
|
||||
| Combine mean fields and ledger | Combine only if roles/scale remain legible; otherwise split | `discuss` |
|
||||
| Retain POD panel | Only with exact same-object arrays; otherwise one bounded sentence | `discuss` |
|
||||
| Caption publication label | Exact preferred derived publication plus validated parent; avoid “canonical CCD” | `discuss` |
|
||||
| Compress displays | Protect mean-first order, residual equation, local caveats; drop operator/redundant modes first | `discuss` |
|
||||
|
||||
**Deferrals/conflicts.** The fourth modal record remains mapped because retrieval failed. Project arrays/payloads were not artifact-audited. Visual economy conflicts with dual CCD/POD galleries; resolve with one CCD/phase sequence and compact subspace boundary. Causal language from other correlation literature cannot transfer to this different estimand.
|
||||
|
||||
**Blockers.** None for this style memo or bounded text contract. A retained panel is blocked if exact role arrays, common object/domain, orientation, lineage, or rendering provenance are missing. POD comparison is blocked unless both methods use the identical declared residual. Optional missing panels do not block text.
|
||||
|
||||
### Anti-patterns
|
||||
|
||||
- Linear algebra or mode gallery before roles and scalar estimand.
|
||||
- Four fields without comparator, mask, window, and acquisition.
|
||||
- Percentage without scalar denominator and non-energy caveat.
|
||||
- Residual called response or paired counterfactual.
|
||||
- Negative lag called response time or directionality.
|
||||
- CCD strengths called energy/explained variance.
|
||||
- SR terms equated with CCD coordinates or one lineage called triangulation.
|
||||
- CCD superiority inferred from visual cleanliness or action labels.
|
||||
- POD used as a straw comparator or methods added as a catalogue.
|
||||
- M04/M06 called agreement because values are close.
|
||||
- Caveats delayed until Discussion.
|
||||
- Panel-by-panel narration, package-handoff prose, or inventory treated as artifact existence.
|
||||
- Closing with a section-number announcement rather than the non-transfer bridge.
|
||||
|
||||
## 8. Validation and disposition
|
||||
|
||||
**Gate-1 validation: PASS for coordinator review.** Exact identities, DOIs, and read levels are stated; the unread record remains mapped. Sampled observed practice is separate from recommendation. Role order, scalar/modal boundaries, equations, figure cadence, POD comparison, caveats, bridges, ownership, near-collisions, choices, readiness, fallbacks, and anti-patterns are explicit. Literature supplies no internal result or priority. Evidence and artifact-audit states remain distinct.
|
||||
|
||||
**Disposition:** `researching`. This is a completed first-pass style memo, not a freeze; only the Round-4 coordinator’s required second audit may advance the downstream section contract.
|
||||
@@ -0,0 +1,401 @@
|
||||
# Round 4 style research — cross-case extensions
|
||||
|
||||
**Lane:** section-style research for the cross-case extension unit.
|
||||
**Status:** `researching`; this first-pass memo cannot self-freeze.
|
||||
**Role:** decision/freeze input, not manuscript prose, project evidence, or a source of new scientific claims.
|
||||
|
||||
## 1. Scope, question and bounded recommendation
|
||||
|
||||
### Scope and question
|
||||
|
||||
This memo studies how a JFM-facing cross-case section can move from a reference case to heterogeneous, transient, retargeted, or failure-bounded extensions without turning the section into a case gallery or implying generalization. It focuses on five rhetorical operations:
|
||||
|
||||
1. change one contract dimension at a time;
|
||||
2. reintroduce the local comparator whenever the contract changes;
|
||||
3. use visual-only evidence at a visibly lower claim level than quantified comparison;
|
||||
4. place negative or failed routes beside the claim they bound; and
|
||||
5. finish with an evidence envelope rather than a breadth or universality claim.
|
||||
|
||||
External papers are used only to observe publication practice. They do not establish any project result, role, artifact, or priority.
|
||||
|
||||
### Bounded recommendation
|
||||
|
||||
Build the section as three argumentative tests rather than a catalogue of cases. Each module should begin by naming the single changed contract, then restate the target and comparator, state the strongest admissible evidence type, give one principal finding at its actual readout level, expose the nearest failure or missing dependency, and hand off the resulting boundary. Keep the reference case as a comparison anchor, but do not import its deep interpretation. When only images are available, call the evidence a named visual demonstration and let the caption carry acquisition identity and visible limitations; do not convert appearance into field matching, efficacy, robustness, or transfer. End by stating which execution interfaces were exercised under distinct contracts and which interpretations did not travel with them.
|
||||
|
||||
The trade-off is deliberate. This structure gives up the rhetorical breadth of a montage and some visual variety, but it lets each extension answer a distinct question and prevents incompatible metrics, targets, temporal coordinates, and evidence states from being read as one progression score.
|
||||
|
||||
## 2. Coverage ledger
|
||||
|
||||
`Read` means inspected directly for this lane. It does not mean independently reproduced, independently verified, or artifact-audited. `Mapped` means located and used for navigation only. Literature read levels follow the Round-2 definition: `full-PDF` is paper-level full-text reading, not equation-by-equation verification.
|
||||
|
||||
| Status | Source/input | Why in scope | Evidence state/reason |
|
||||
|---|---|---|---|
|
||||
| `read` | `ROUND4_TOP_DOWN/00_MASTER_AGENT_PROMPT.md` | Round-4 method, deliverables, fixed cautions and status rules | Read directly; control document, not independently verified or artifact-audited |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_QUALITY_GATE.md` | Style-memo and section-freeze gates | Read directly; control document |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_GLOBAL_DECISION_LEDGER.md` | Ownership, claim ladder, section order and cross-case ceiling | Read directly; coordinator authority, not scientific artifact audit |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_DISPLAY_EQUATION_INVENTORY.md` | Cross-case display readiness, equation ownership and fallbacks | Read directly; inventory does not establish panel existence |
|
||||
| `read` | `ROUND4_TOP_DOWN/README.md` | Round-4 purpose, inputs, statuses and handoff boundary | Read directly; control document |
|
||||
| `read` | `ROUND2_TOP_DOWN/05_REVISED_STRUCTURE_FREEZE.md` (relevant cross-case and supersession passages) | Controlling section placement and rejection of a mere gallery | Read directly; project architecture authority |
|
||||
| `read` | `ROUND2_TOP_DOWN/06_SECTION_LEVEL_JFM_STYLE_GUIDE.md` | Controlling JFM style synthesis and cross-case cadence | Read in full; `full-report`; its literature identities were checked against current records where used below |
|
||||
| `read` | `ROUND3_BOTTOM_UP/07_CROSS_CASE_EXTENSIONS.md` | Relevant M07 evidence, interfaces, display gates and failure boundaries | Read in full; `full-report`; no underlying project artifact independently verified or artifact-audited in this lane |
|
||||
| `read` | `ROUND3_BOTTOM_UP/00_MODULE_INTERFACE_LEDGER.md` | M07 ownership and downstream interface | Read directly; coordinator ledger |
|
||||
| `read` | `ROUND3_BOTTOM_UP/11_FINAL_CONSISTENCY_AUDIT.md` | Reconciled cross-module scope, readiness and duplication findings | Read in full; `full-report`; no artifact recomputation |
|
||||
| `read` | Xia, Zhang, Kerrigan & Rigas (2024), exact record below | Closest failure-to-repair and bounded unseen-condition example | `full-PDF`; paper-level rhetorical reading; bibliographic DOI/PDF availability checked; no numerical reproduction |
|
||||
| `read` | Paris, Beneddine & Dandois (2021), exact record below | Closest reference-case-to-robustness/sensor-extension example | `full-PDF`; paper-level rhetorical reading; bibliographic DOI/PDF availability checked; no numerical reproduction |
|
||||
| `read` | Wang et al. (2024), exact record below | Closest escalating heterogeneous-flow and degraded-route example | `full-PDF`; paper-level rhetorical reading; bibliographic DOI/PDF availability checked; no numerical reproduction |
|
||||
| `read` | Deng et al. (2020), exact record below | Closest pinball example of visible failure, refinement and synthesis rather than gallery | `full-PDF`; paper-level rhetorical reading; bibliographic DOI/PDF availability checked; no numerical reproduction |
|
||||
| `mapped` | Round-3 modules adjoining M07 and authority/manifests linked from M07 | Locate imported definitions, export boundaries and artifact gates | Mapped for interface navigation; not read as evidence in this style lane and not independently verified |
|
||||
| `mapped` | Existing Undermind `JFM/` and `JFM References/` collections | Locate exact records and PDFs | Collection/index mapping only; mapped is not read |
|
||||
| `not-covered` | Exhaustive sentence coding of the broader JFM corpus; all references cited by the four papers | Beyond the bounded closest-example sample | No general frequency claim is inferred from four papers |
|
||||
| `not-covered` | Raw project arrays, field files, evaluator outputs, source media, rendering scripts and hashes | This is style research, not artifact generation or audit | No project value, panel existence or quantitative result is inferred |
|
||||
| `not-covered` | New CFD, training, SR, CCD, experiment analysis or metric computation | Explicitly outside Round 4 | No new evidence or claim produced |
|
||||
| `external-inaccessible` | None among the four selected JFM PDFs | All four full texts were available in the literature workspace | No essential literature blocker |
|
||||
| `external-inaccessible` | External/symlink-backed project payloads named by M07 | Not opened in this style lane; access and audit state are outside its remit | Authority-led artifact-audit deferral; retained quantitative panel dependencies remain governed by M07 and Round-4 inventory |
|
||||
|
||||
Coverage is complete for the requested Round-4 controls, controlling Round-2 style guide, relevant Round-3 M07 dossier/audit interface, and four closest full-PDF examples. It is intentionally not a new project-evidence audit.
|
||||
|
||||
## 3. Exact literature identities and observed JFM practice
|
||||
|
||||
### 3.1 Exact identities and read levels
|
||||
|
||||
1. C. Xia, J. Zhang, E. C. Kerrigan & G. Rigas, “Active flow control for bluff body drag reduction using reinforcement learning with partial measurements”, *Journal of Fluid Mechanics*, **981**, A17 (2024). DOI: `10.1017/jfm.2024.69`. Read level: `full-PDF`.
|
||||
2. R. Paris, S. Beneddine & J. Dandois, “Robust flow control and optimal sensor placement using deep reinforcement learning”, *Journal of Fluid Mechanics*, **913**, A25 (2021). DOI: `10.1017/jfm.2020.1170`. Read level: `full-PDF`.
|
||||
3. Qiulei Wang, Dong Yan, Guodong Hu, Gang Chen, Jean Rabault & Bernd R. Noack, “Dynamic feature-based deep reinforcement learning for flow control of circular cylinder with sparse surface pressure sensing”, *Journal of Fluid Mechanics*, **988**, A4 (2024). DOI: `10.1017/jfm.2024.333`. Read level: `full-PDF`.
|
||||
4. N. Deng, B. R. Noack, M. Morzyński & L. R. Pastur, “Low-order model for successive bifurcations of the fluidic pinball”, *Journal of Fluid Mechanics*, **884**, A37 (2020). DOI: `10.1017/jfm.2019.959`. Read level: `full-PDF`.
|
||||
|
||||
The literature service’s cite-key years reflect prepublication records for some papers; the journal year, volume and article identity above control this memo. No quotations are used, so there is no quotation-level verification claim.
|
||||
|
||||
### 3.2 Observed practice: extensions change a declared difficulty
|
||||
|
||||
**Xia et al.** establish the fully observed reference problem before reducing the available measurements. The degraded static-controller route is shown before temporal history is introduced as the repair. Later tests vary history content, history length, reward design, sensor arrangement and unseen Reynolds number in separate steps. The rhetorical value is not merely chronological: the reader can identify which information contract changed and which comparison diagnoses the consequence.
|
||||
|
||||
**Paris et al.** first establish a reference control configuration, then test a fixed learned policy under changed flow and noise conditions, and only afterward narrow the sensor interface. The robustness and sensor-placement parts are therefore extensions of a declared reference, not additional showcase cases. Dedicated parameter tests precede the stronger vocabulary attached to them.
|
||||
|
||||
**Wang et al.** first compare the proposed feature construction with a conventional controller under a baseline sensing/flow setting. Harder Reynolds-number and three-dimensional settings arrive later, with comparison schemes renamed or restated so that the reader can still tell which method and sensing contract is being compared. A degraded sparse-sensing route and an undesirable reward-exploitation route are made visible rather than omitted.
|
||||
|
||||
**Deng et al.** build separate low-order descriptions for successive physical transitions, compare each against the numerical plant in its applicable regime, and then combine/refine them. A simpler model’s visible inability to reproduce part of the transient is used to motivate additional cross-terms. The final phenomenogram synthesizes the route through regimes; it is not a montage offered as proof that one model is universal.
|
||||
|
||||
**Observed pattern across this sample.** A heterogeneous extension is readable when the paper makes the changed difficulty explicit and preserves a stable comparison anchor. The papers do not rely on section ordering alone to communicate this: they restate the relevant sensing, flow, model or policy condition at the point where it changes. This is a sample-level observation, not a claim about all JFM papers.
|
||||
|
||||
### 3.3 Observed practice: comparators are renewed locally
|
||||
|
||||
The four papers repeatedly return to a local reference rather than assuming that a comparator introduced many pages earlier remains obvious.
|
||||
|
||||
- Xia et al. keep uncontrolled/base-flow bounds and the fully measured controller visible when evaluating partial-measurement variants and their repair.
|
||||
- Paris et al. retain the reference policy/configuration while perturbing Reynolds number, noise or sensor count, so a change in outcome can be attached to the stated test.
|
||||
- Wang et al. explicitly define comparison schemes again when the dimensionality and sensing arrangement change.
|
||||
- Deng et al. renew DNS as the plant comparator at each regime and distinguish the unrefined and refined reduced models.
|
||||
|
||||
This renewal does not mean that all cases share one metric. Its function is local identification: what is the target or baseline here, what changed, and what readout answers this module’s question?
|
||||
|
||||
### 3.4 Observed practice: visuals explain or delimit; they do not silently upgrade evidence
|
||||
|
||||
In Xia et al., wake snapshots are read beside reward, force or temporal comparisons and are used to clarify why the partial-measurement configurations differ. In Paris et al., wake structure and sensor-layout views are paired with performance curves rather than substituted for them. In Wang et al., representative and mean flow fields contextualize force changes, while training and force evidence carry the comparison. In Deng et al., state-space portraits and the final phenomenogram provide physical organization, while time histories and coefficient/model comparisons test fidelity.
|
||||
|
||||
The common rhetorical discipline is pairing: a visual can show morphology, timing, placement or qualitative difference, but the claim register follows the acquisition and readout actually supplied. None of these examples licenses treating visual resemblance as an unreported norm, field error, target match or transfer test.
|
||||
|
||||
### 3.5 Observed practice: failed routes remain in the argument
|
||||
|
||||
The closest examples use negative evidence constructively:
|
||||
|
||||
- a partial-measurement static controller fails before a history-based repair is introduced;
|
||||
- extreme sensor sparsity and some transient energetic behaviour remain visible beside successful control;
|
||||
- a conventional sparse-sensing route degrades, and an undesirable reward-exploitation strategy is identified;
|
||||
- a least-order model’s transient mismatch motivates refinement before the final synthesis.
|
||||
|
||||
These failures are local and diagnostic. They identify the limit of a specific information, actuation, reward or model contract. The papers do not automatically turn a repaired failure into robustness across all omitted conditions.
|
||||
|
||||
### 3.6 Observed practice: breadth is earned test by test
|
||||
|
||||
Paris et al. attach robustness/generalization wording to dedicated perturbation tests. Xia et al. delimit unseen-condition performance around a tested training setting and expose degradation away from it. Wang et al. qualify conclusions by the investigated flow and sensing configurations. Deng et al. identify configuration-specific and non-generic aspects of the bifurcation sequence. In this sample, the breadth word follows the breadth test; it does not follow from the number of visually different cases.
|
||||
|
||||
## 4. Project recommendation derived from the observed practice
|
||||
|
||||
The recommendations below are design choices for this paper, not descriptions of project outcomes.
|
||||
|
||||
### 4.1 Section-level rhetorical contract
|
||||
|
||||
Use three modules, each with the same six-part logic but not identical prose:
|
||||
|
||||
1. **Changed contract.** Name exactly one primary change from the preceding reference: temporal coordinate, target/reference, or evidence availability.
|
||||
2. **Local comparator.** Restate target, controlled role and baseline/zero role only to the detail needed for this comparison. If one role is absent, say so before the result.
|
||||
3. **Admissible readout.** Name whether the evidence is event-relative quantitative, periodic target-relative quantitative, mean-only, or visual-only. Do not promote a candidate definition into a result.
|
||||
4. **Principal observation.** Give one result-level sentence at the admissible claim ceiling; let the figure/caption carry secondary case labels.
|
||||
5. **Negative boundary.** Place the failed route, unavailable comparator, negative reduction route or unresolved evaluator beside the principal observation.
|
||||
6. **Consequence.** State what the changed contract demonstrates and what interpretation does not transfer.
|
||||
|
||||
This repetition is conceptual, not templated wording. Vary sentence shape and transition according to the scientific change. If every module reads as “case name → image → success,” the section has become a gallery.
|
||||
|
||||
### 4.2 Change one contract at a time
|
||||
|
||||
The opening transition should decompose heterogeneity into independent axes before presenting examples. Recommended conceptual sequence:
|
||||
|
||||
- periodic coordinate → event-relative coordinate;
|
||||
- background/reference objective → declared non-zero target;
|
||||
- supported target/comparator roles → a further objective whose quantitative role contract is incomplete.
|
||||
|
||||
At each transition, explicitly name what remains fixed only if the source authority permits that statement. Do not use “same framework” as a shortcut: it can conceal changes in chain, geometry, role acquisition, normalization, evaluation window or target construction. When several things actually change together, label the module as a named demonstration under a distinct contract rather than pretending it is a controlled one-variable extension.
|
||||
|
||||
### 4.3 Reintroduce comparators without repeating Methods
|
||||
|
||||
A local comparator sentence should answer four questions:
|
||||
|
||||
- What field or signature is the declared target?
|
||||
- Which role is the controlled observation?
|
||||
- Which baseline or physical-zero role is informative here?
|
||||
- Which metric or visual readout is admissible under this acquisition?
|
||||
|
||||
The section should import definitions from Contracts and avoid reprinting formulas, ROI details or action conventions unless the change itself makes one of them newly consequential. This gives enough re-anchoring for the reader without duplicating M03 or the detailed Kármán performance section.
|
||||
|
||||
Do not call physical zero the target in a non-zero-target module. Do not treat a missing baseline as if a visual panel supplies it. Do not preserve a comparator label while silently changing its acquisition or window.
|
||||
|
||||
### 4.4 Visual-only evidence contract
|
||||
|
||||
When the evidence is visual-only, use a two-level statement:
|
||||
|
||||
- **Observation:** identify the named scene, relative/event coordinate or target/control role visible in the source artifact.
|
||||
- **Limit:** state that no quantitative target/control/baseline field comparison follows without exact roles, common domain/mask, weighting/normalization, window and evaluator authority.
|
||||
|
||||
Recommended verbs are `shows the named visual demonstration`, `depicts`, `is visibly different under the stated acquisition`, or `supports inclusion as a capability example`. Avoid `matches`, `recovers`, `suppresses`, `tracks`, `generalizes`, `is effective`, or `performs comparably` unless the corresponding quantitative comparator and estimand are available.
|
||||
|
||||
A visual panel should not carry an unsourced scalar annotation. If the caption cannot state chain, case, role, comparator/target, temporal coordinate, source and evidence level, omit the panel. Visual-only is an evidence status, not a criticism; hiding that status is the defect.
|
||||
|
||||
### 4.5 Negative and failed routes
|
||||
|
||||
Place each negative route in the module it constrains, not in a late limitations dump.
|
||||
|
||||
- A negative surrogate/reduction route belongs beside the positive learned-policy capability, with the chain and acquisition difference made explicit; it is a boundary, not a head-to-head comparison unless protocols match.
|
||||
- A missing event evaluator or exact role set belongs immediately after any transient visual demonstration; it prevents numerical inference but need not erase the visual observation.
|
||||
- An unsupported objective/target construction should be presented as a route that cannot sustain quantitative efficacy, not as an empty placeholder for a future success.
|
||||
|
||||
Use failures to answer “what would have to change for a stronger claim?” They should not be dramatized as universal impossibility, and a successful neighbouring case should not be presented as having repaired a failure in a different chain or estimand.
|
||||
|
||||
### 4.6 Avoid gallery and generalization claims
|
||||
|
||||
A cross-case figure becomes a gallery when panels are grouped by appearance but lack a shared argumentative question. Prevent this by giving each retained panel one job:
|
||||
|
||||
- establish the changed temporal coordinate;
|
||||
- establish the changed target/reference;
|
||||
- expose a boundary or unavailable comparison.
|
||||
|
||||
The prose should compare contracts, not visual attractiveness. Do not count cases as evidence of generality. Do not use a common colour scale, aligned panel geometry or shared layout to imply a common estimand. A synthesis sentence may say that the execution objective was exercised in named cases under distinct contracts; it may not say that the policy, interpretation, mechanism or efficacy generalized across them.
|
||||
|
||||
### 4.7 Evidence density, cadence and transitions
|
||||
|
||||
Recommended density is one principal evidentiary object per module. Put secondary labels, acquisition details and non-claims in the caption or a compact case-contract table. A paragraph should not enumerate every case variant, target size, offset or file. Move complete inventories to the supplement if they remain scientifically useful.
|
||||
|
||||
Use transitions as logical changes:
|
||||
|
||||
- reference periodicity established → ask what replaces phase for an incoming event;
|
||||
- event-relative execution established → ask what changes when the desired signature is non-zero;
|
||||
- retargeting shown at capability level → ask which extension routes survive exact comparator and evidence requirements;
|
||||
- bounded extensions assembled → return to the deep case’s context without transferring its interpretation.
|
||||
|
||||
Avoid transitions of the form “the next case is…” or “Figure X also shows…”.
|
||||
|
||||
## 5. Paragraph jobs, ownership and interfaces
|
||||
|
||||
### Opening dependency
|
||||
|
||||
Import from the contracts section only the already frozen distinctions among chains, target/control/baseline roles, temporal coordinates and case-specific metrics. Import from the capability section only the existence and placement of the named cross-case preview. Import from the Kármán deep block only the fact that it is the reference/deep case; do not import detailed values, surrogate structure or dynamic-field interpretation.
|
||||
|
||||
### Recommended paragraph-function map
|
||||
|
||||
| Paragraph job | Opening move | Evidence function | Local limitation | Export/bridge |
|
||||
|---|---|---|---|---|
|
||||
| 1. Decompose the extension problem | State that temporal and target contracts change independently | Establish the comparison logic and prohibit a pooled score | “Shared interface” cannot imply same chain or estimand | Opens event-relative module |
|
||||
| 2. Event-relative extension | Replace periodic phase with a declared event coordinate; renew target/control/baseline roles | One bounded event-relative result if authority exists; otherwise one visual-only observation | No phase averaging, arbitrary-disturbance claim or imported Kármán interpretation | Asks whether the objective itself can be retargeted |
|
||||
| 3. Non-zero-target extension | Define the local non-zero target and comparator anew | One bounded learned-policy capability observation; quantitative field statement only if exact role authority exists | Keep negative reduction route local and chain-separated; no target-tracking inference from appearance | Asks what happens when the target/reference changes again |
|
||||
| 4. Failure-bounded further objective | State the required target/comparator contract before any panel | Either a supported mean/target comparison or an explicit unsupported-route boundary | Filenames, plot navigation and visual similarity do not establish roles or efficacy | Builds the evidence envelope |
|
||||
| 5. Cross-case synthesis | Compare what changed and what did not travel | Synthesize named capability under distinct contracts | No common score, transfer, robustness, mechanism or universality | Bridges to later steady context as context only |
|
||||
|
||||
Five paragraphs is a functional target, not a quota. Split a paragraph only when the comparator and failure boundary cannot remain legible together; combine only when no evidence distinction is lost.
|
||||
|
||||
### Result ownership
|
||||
|
||||
- **Cross-case section owns:** the changed-contract sequence; local comparator renewal; event-relative versus periodic-target semantics at a high level; visual-only evidence language; the negative/unsupported extension envelope; and the final non-transfer statement.
|
||||
- **Contracts owns:** reusable chain, role, action, metric, ROI/mask/window and normalization definitions.
|
||||
- **Capability owns:** the early preview/progression and Fig. 1 selection logic.
|
||||
- **Kármán performance owns:** detailed reference-case field and signal results.
|
||||
- **SR and field/CCD own:** deep Kármán interpretation and its exact evidentiary components.
|
||||
- **Conclusion owns:** one compressed restatement of breadth with asymmetric interpretation, not the detailed negative-route ledger.
|
||||
|
||||
### Imports
|
||||
|
||||
1. Definitions of chain, role, comparator, temporal coordinate and metric semantics from Contracts.
|
||||
2. The reference/deep-case designation from Capability/Kármán performance.
|
||||
3. The positive-scope boundary from SR and non-transfer boundary from Field/CCD.
|
||||
4. M07’s authority-led status for visual, quantitative, conditional and blocked routes.
|
||||
|
||||
Imports must remain brief. They cannot be rewritten as cross-case evidence.
|
||||
|
||||
### Exports
|
||||
|
||||
1. A named, case-bounded capability statement under distinct temporal/target contracts.
|
||||
2. A local evidence envelope distinguishing quantitative, visual-only, negative and unsupported routes.
|
||||
3. A prohibition on carrying the deep-case interpretation or one score across extensions.
|
||||
4. A bridge to later steady context that leaves transient and target-transfer gaps open.
|
||||
|
||||
### Duplication deletions for assembly
|
||||
|
||||
- Delete detailed Kármán values and metric contracts from this section; retain only the reference function.
|
||||
- Delete full action-law or correction-field explanations; retain only their non-transfer boundary.
|
||||
- Delete the case inventory from main prose; place exact variants in a contract table or supplement if retained.
|
||||
- Delete any second full definition of spatial L2 or DTW; use the contracts notation by reference.
|
||||
- Delete repeated negative-route detail from Conclusion; Conclusion keeps only the envelope.
|
||||
- Delete caption-style narration from prose when the panel already carries role and acquisition identity.
|
||||
|
||||
### Closing bridge
|
||||
|
||||
The bridge should say, in function rather than polished prose: heterogeneous cases exercise execution under changed temporal or target contracts, while the deep interpretation remains local; the manuscript can now return to steady evidence only as contextual reference for that local interpretation, not as closure of transient or target-transfer questions.
|
||||
|
||||
## 6. Claim ceiling and near-collision audit
|
||||
|
||||
### Green: admissible at named scope
|
||||
|
||||
- The section may distinguish periodic, event-relative, non-zero-target and unsupported/conditional comparison contracts.
|
||||
- It may report a named visual demonstration as visual-only when exact provenance and roles are available.
|
||||
- It may state that a negative or unsupported route bounds the extension claim under its own chain and acquisition.
|
||||
- It may state that cross-case evidence supports named capability breadth under distinct contracts, provided “breadth” is not rewritten as robustness or generalization.
|
||||
|
||||
### Amber: allowed only with explicit bounds
|
||||
|
||||
- `extends the objective` or `exercises the execution interface under a changed contract` when the unchanged and changed elements are authority-bound.
|
||||
- `supports a comparable-scale hypothesis` only if the source dossier licenses that interpretation and the named tested disturbances remain explicit.
|
||||
- `is consistent with retargeting at capability level` only when target and controlled roles are identified and the statement does not imply independent field matching.
|
||||
- `provides a failure boundary` when the negative route’s chain, protocol and comparator are stated.
|
||||
|
||||
### Prohibited red without new authoritative evidence
|
||||
|
||||
- `generalizes`, `transfers`, `robust`, `universal`, `arbitrary-flow`, `cross-size law`, `common efficacy`, `same mechanism`, `same control skeleton`, `validates`, `replicates`, `agrees with`, `triangulates`, `target tracking`, or `field matching` from visual resemblance alone.
|
||||
- Any positive extension of the Kármán-only surrogate or dynamic-field interpretation.
|
||||
- Any quantitative result inferred from filenames, plot navigation, panel existence, a candidate equation or an unverified role mapping.
|
||||
- Any pooled arithmetic across periodic, event-relative, mean-only, visual-only or chain-separated readouts.
|
||||
|
||||
### Near-collision findings
|
||||
|
||||
The section must perform the near-collision check before putting two values or panels into comparative syntax. For each pair, compare:
|
||||
|
||||
1. acquisition and executable chain;
|
||||
2. target, controlled and baseline role identities;
|
||||
3. temporal coordinate and retained window;
|
||||
4. ROI/domain and mask;
|
||||
5. weighting and normalization;
|
||||
6. metric definition and orientation.
|
||||
|
||||
Classification rules:
|
||||
|
||||
- **Same estimand:** only when all six items are authority-bound as the same comparison.
|
||||
- **Related diagnostic:** same acquisition/question but a deliberately different readout, such as field and signal evidence that remain separately interpreted.
|
||||
- **Different estimand:** any changed chain, role acquisition, target, temporal coordinate, mask/weighting, normalization or window that changes the measured object.
|
||||
|
||||
For this style lane, periodic versus event-relative evidence is `different estimand`; visual-only versus quantitative field comparison is `different evidence/readout`, not agreement; positive learned-policy capability versus a negative surrogate route from another chain/protocol is a boundary juxtaposition, not a direct comparator; and any further objective lacking exact target/control/baseline roles is not classifiable as an efficacy result. No arithmetic, replication, validation, agreement or triangulation wording is allowed merely because values or images look similar.
|
||||
|
||||
## 7. Figure, table and equation contracts
|
||||
|
||||
### Cross-case display
|
||||
|
||||
- **Owner:** Cross-case section.
|
||||
- **Argumentative job:** show the sequence of contract changes and their evidence ceilings, not maximize the number of cases.
|
||||
- **Source:** only authority-bound existing visual or quantitative artifacts identified by M07 and later production audit.
|
||||
- **Roles/arrays:** each retained panel must identify chain, case, target, controlled role, baseline/zero role if present, temporal coordinate, source artifact and readout. Exact arrays remain a production/audit dependency.
|
||||
- **Caption ceiling:** named event-relative or non-zero-target capability; visual-only where no exact quantitative comparison exists; negative/unsupported route visible locally.
|
||||
- **Classification:** overall `dependency-gated`; quantitative transient and unsupported-objective components remain `blocked` where the Round-4 inventory says so; a non-zero-target field metric remains conditional on exact authority.
|
||||
- **Text readiness:** the rhetorical contract is text-ready as a recommendation.
|
||||
- **Artifact readiness:** not panel-ready in this memo; no artifact was inspected or generated.
|
||||
- **Fallback:** retain only sourced visual panels with explicit evidence level; omit blocked quantitative annotations and unsupported panels. If the remaining display becomes a gallery, replace it with the compact case-contract table.
|
||||
- **Does not establish:** common efficacy, arbitrary-flow capability, cross-case transfer, robustness, positive surrogate scope, causal mechanism or quantitative success from appearance.
|
||||
|
||||
### Compact case-contract table
|
||||
|
||||
- **Owner:** Cross-case section; detailed definitions imported from Contracts.
|
||||
- **Argumentative job:** make the changed dimension, comparator, readout, evidence state and local limit visible in one place.
|
||||
- **Source:** frozen contracts plus M07 authority ledger; no new values.
|
||||
- **Rows/fields:** module/question; changed contract; target; controlled role; baseline/comparator; temporal coordinate; admissible readout; evidence status; strongest local limit.
|
||||
- **Caption ceiling:** organizational aid for non-equivalent extensions.
|
||||
- **Classification:** `optional`, but preferred fallback if a multi-panel display cannot avoid gallery reading.
|
||||
- **Text readiness:** table schema ready.
|
||||
- **Artifact readiness:** content requires coordinator production audit against current authorities.
|
||||
- **Fallback:** concise opening paragraph plus local comparator sentences.
|
||||
- **Does not establish:** comparability of metrics merely because they occupy adjacent rows.
|
||||
|
||||
### Event-relative metric equation
|
||||
|
||||
- **Owner:** Cross-case section for the event-specific definition; Contracts owns reusable spatial-error notation.
|
||||
- **Argumentative job:** define an event-relative candidate estimand only if a retained quantitative transient comparison requires it.
|
||||
- **Source/roles:** exact target/control roles, event offsets, common ROI/domain/mask, weighting/normalization and evaluator authority must be bound before use.
|
||||
- **Caption/text ceiling:** definition only; it creates no role acquisition, evaluator output or result.
|
||||
- **Classification:** `optional`; currently dependency-bound and not required for the section’s bounded visual text.
|
||||
- **Text readiness:** semantic description is ready; notation is not frozen here.
|
||||
- **Artifact/result readiness:** blocked wherever exact roles/evaluator are absent.
|
||||
- **Fallback:** prose or the compact case-contract table.
|
||||
|
||||
### Negative-route callout
|
||||
|
||||
- **Owner:** Cross-case section.
|
||||
- **Argumentative job:** keep negative surrogate evidence and unsupported objective routes adjacent to the positive capability they bound without presenting unlike protocols as matched comparisons.
|
||||
- **Source:** M07’s current authority ledger and final Round-3 audit.
|
||||
- **Classification:** `required` as prose; `optional` as a boxed/table display.
|
||||
- **Caption ceiling:** historical/negative or unsupported under the named acquisition; no cross-chain efficacy comparison.
|
||||
- **Text readiness:** ready as a rhetorical requirement.
|
||||
- **Artifact readiness:** any numerical callout remains dependent on exact authority and coordinator choice; this memo introduces none.
|
||||
- **Fallback:** one local limiting sentence in each affected module.
|
||||
|
||||
No other equation is recommended. The section should reuse frozen notation and spend display space on comparison logic, not derive a common metric that the evidence does not support.
|
||||
|
||||
## 8. Anti-patterns and corrective actions
|
||||
|
||||
| Anti-pattern | Why it fails | Corrective action |
|
||||
|---|---|---|
|
||||
| One panel or paragraph per case | Converts heterogeneity into inventory and encourages case counting | Organize by changed contract and retain one principal evidentiary job per module |
|
||||
| “The framework generalizes to…” | A changed case is not a dedicated transfer/generalization test | Use named-capability wording and state the tested contract |
|
||||
| Common score bar across extensions | Periodic, event-relative, target-relative and mean-only quantities are different estimands | Use a contract/evidence-state table; keep metrics local |
|
||||
| Visual similarity described as matching | Appearance does not establish target/control/baseline field error | Label visual-only and state the missing quantitative dependencies |
|
||||
| Comparator introduced once, then assumed | The target, baseline or acquisition may have changed | Renew the local comparator at every contract change |
|
||||
| Positive capability and negative route treated as matched | Different chain, role or window can make the juxtaposition non-comparative | Present the negative route as a local boundary and state protocol separation |
|
||||
| Failure hidden in Limitations | The reader overreads the preceding success before reaching the caveat | Put the failed or unsupported route beside the claim it limits |
|
||||
| Failure called proof of impossibility | A local failed configuration does not exhaust alternatives | Name the failed contract and the next evidence needed |
|
||||
| Kármán interpretation reused as explanation | Capability breadth does not transfer deep-case SR/field interpretation | Import only the non-transfer boundary |
|
||||
| Candidate equation presented as evidence | A definition cannot create missing roles, arrays or evaluator output | Mark the equation optional/conditional and separate result readiness |
|
||||
| Filenames or manifests treated as role proof | Navigation metadata need not bind target identity or comparability | Require authority-bound roles and artifact audit before quantitative prose |
|
||||
| Captions narrate success while prose lists files | Splits the argument between promotional captions and inventory prose | Give prose the inference; captions carry provenance and limits |
|
||||
| Every module uses identical six-sentence form | Reads as generated template rather than scientific progression | Preserve the six jobs but vary cadence and emphasis by the changed question |
|
||||
| Strong caveat only in final synthesis | Allows local overclaim and makes limits look retrofitted | Pair each major observation with its nearest limitation |
|
||||
|
||||
## 9. Author decisions, conflicts, deferrals and blockers
|
||||
|
||||
### Evidence-governed decisions
|
||||
|
||||
These are not author choices:
|
||||
|
||||
- Keep periodic, event-relative, non-zero-target, mean-only and visual-only evidence states distinct.
|
||||
- Keep executable chains and role acquisitions distinct.
|
||||
- Keep positive surrogate interpretation confined to its authorized deep case; retain the negative extension route as historical/negative.
|
||||
- Do not infer quantitative efficacy from an unsupported target construction, filenames or visual navigation.
|
||||
- Do not call a multi-case display generalization, robustness, transfer or a common score without dedicated authority.
|
||||
- Preserve text readiness separately from panel/artifact readiness.
|
||||
|
||||
### Decision-ready author/coordinator choices
|
||||
|
||||
1. **Main display form.** Recommendation: use the cross-case display only if each retained panel has one distinct argumentative job and complete caption provenance. Safe default: use the compact case-contract table plus at most the strongest sourced visual examples. Consequence: fewer images, clearer evidentiary hierarchy. Status: `discuss`.
|
||||
2. **Retain an event equation.** Recommendation: omit unless a quantitative event result becomes authority-bound and remains in the main text. Safe default: prose semantics. Consequence: avoids displaying a definition that has no result. Status: `discuss`.
|
||||
3. **How prominently to show the negative route.** Recommendation: retain it in the same module as the positive capability, in prose or a small callout. Safe default: one local sentence, not a full comparative panel. Consequence: the boundary stays visible without implying matched protocols. Status: `discuss`.
|
||||
4. **Further-objective panel.** Recommendation and safe default: omit while its exact quantitative target/control/baseline contract is unsupported. Consequence: the failure envelope remains in prose and the section avoids a speculative gallery. Status: `blocked` if the panel is retained; omission is nonblocking.
|
||||
|
||||
### Conflicts
|
||||
|
||||
No conflict was found between the controlling Round-4 ledger, Round-2 style guide, M07, the Round-3 final audit and the observed practices in the four selected papers. The literature sample supports the rhetorical direction—reference first, changed difficulty declared, comparator renewed, failure local, breadth test-bounded—but it does not decide project evidence or elevate any project claim.
|
||||
|
||||
### Deferrals
|
||||
|
||||
- Project artifact inspection, role-array audit, mask/window verification, evaluator availability and rendering provenance are deferred to the owning evidence/production lanes.
|
||||
- Exact sentence/page coding of the four papers is not required because this memo uses no quotations or paper-derived numerical claims.
|
||||
- Broader JFM corpus coverage is deferred; the four-paper sample is treated as closest examples, not a universal style law.
|
||||
|
||||
### Blockers
|
||||
|
||||
There is no blocker to this style recommendation. Retained quantitative transient or unsupported-objective panels remain blocked by the dependencies already recorded in M07 and the Round-4 display inventory. Those panel blockers do not justify stronger visual prose and do not block a bounded text contract with omission fallbacks.
|
||||
|
||||
## 10. Disposition
|
||||
|
||||
**Disposition: `researching`.** The requested style evidence and section recommendation are complete for first-pass review, but Round-4 controls prohibit a first-pass lane from self-freezing. A coordinator second audit should re-read this memo with the controls and M07, confirm exact literature identities/read levels, test the ownership and near-collision boundaries, and decide the display/table/equation options. Artifact readiness remains separate and does not control the bounded style memo.
|
||||
@@ -0,0 +1,165 @@
|
||||
# Round 4 style research — steady context, physical synthesis and theory limits
|
||||
|
||||
**Status:** `researching` — first-pass style memo, not a section freeze or manuscript prose.
|
||||
**Owner:** steady/theory section style lane.
|
||||
**Question:** How should a short late return to steady control compare finite-Re observations with an outer theory, retain failed sign/closure routes as scientific boundaries, and close into experiment/conclusion without turning a contextual section into a mechanism chapter?
|
||||
|
||||
## 1. Bounded recommendation
|
||||
|
||||
Use a **contextual return, not a second results climax**. Open by recalling the earlier steady seed and narrowing the question to what the separate steady chain can contextualize. Establish the finite-Re contract and accepted/reverse branch before interpretation; then introduce one short outer-reference equation, confront it immediately with the sign or closure failure, and end with a synthesis that transfers only compatible context. The section should carry one required action/sign display, one compact endpoint comparison, and one theory-failure display or equation; a new MFS field remains blocked. Caveats should sit beside the exact inference they limit, not be postponed to the final Discussion.
|
||||
|
||||
This recommendation governs style and argumentative order only. It supplies no project prose, numerical result, or new scientific claim.
|
||||
|
||||
## 2. Coverage ledger
|
||||
|
||||
| Status | Source/input | Why in scope | Evidence state/reason |
|
||||
|---|---|---|---|
|
||||
| `read` | Round-4 `README.md`, `00_MASTER_AGENT_PROMPT.md`, `00_QUALITY_GATE.md`, `00_GLOBAL_DECISION_LEDGER.md`, `00_DISPLAY_EQUATION_INVENTORY.md` | Controls status, ownership, claim ceiling, section order, display/equation readiness and mandatory coverage fields | Read directly; no independent scientific verification or artifact audit |
|
||||
| `read` | Round-2 `05_REVISED_STRUCTURE_FREEZE.md` | Controls the two-pass steady chronology, 6% late-return role, separate-chain boundary and theory ceiling | Full-report; read directly; not independently verified or artifact-audited |
|
||||
| `read` | Round-2 `06_SECTION_LEVEL_JFM_STYLE_GUIDE.md` | Controlling R2 style guide and existing exact literature/read-level ledger | Full-report; read directly; literature observations selectively checked against four PDFs below |
|
||||
| `read` | Round-3 `08_STEADY_SYNTHESIS.md` | Relevant M08 section dossier: paragraph jobs, provenance, near-collision, equations, displays, failures and blockers | Full-report; read directly; reported project facts were not recomputed and artifact-leaf audit was deferred |
|
||||
| `read` | Round-3 `11_FINAL_CONSISTENCY_AUDIT.md` | Controlling R3 ownership, duplication, text/panel readiness and synthesis-stage deletions | Full-report; read directly; no artifact audit |
|
||||
| `read` | `src/steady_pinball_theory/LITERATURE.md` and relevant portions of `DERIVATION.md` | Current theory-interface authority used only to understand the section’s failure/closure and equation-selection constraints | Full-report; read directly; no new derivation, recomputation or universal artifact audit |
|
||||
| `read` | Deng, Noack, Morzyński & Pastur (2020), “Low-order model for successive bifurcations of the fluidic pinball”, *Journal of Fluid Mechanics* **884**, A37, DOI `10.1017/jfm.2019.959` | Closest pinball precedent for physical chronology, model refinement, visible failure and synthesis figure | `full-PDF`; PDF read for sections, equations and figures; no independent reproduction of DNS or ROM |
|
||||
| `read` | Cornejo Maceda, Li, Lusseyran, Morzyński & Noack (2021), “Stabilization of the fluidic pinball with gradient-enriched machine learning control”, *Journal of Fluid Mechanics* **917**, A42, DOI `10.1017/jfm.2021.301` | Closest pinball precedent for simple steady forcing → richer steady forcing → feedback and for failed/weaker branch placement | `full-PDF`; PDF read for rhetorical and comparison practice; no independent control-result verification |
|
||||
| `read` | Mao, Blackburn & Sherwin (2015), “Nonlinear optimal suppression of vortex shedding from a circular cylinder”, *Journal of Fluid Mechanics* **775**, 241–265, DOI `10.1017/jfm.2015.304` | Closest precedent for theory/optimization versus nonlinear simulation, regime-of-validity failure and realizability caveats | `full-PDF`; PDF read for equation/figure cadence and local limits; no numerical or stability reproduction |
|
||||
| `read` | Ueda, Sellier, Kida & Nakanishi (2003), “On the low-Reynolds-number flow about two rotating circular cylinders”, *Journal of Fluid Mechanics* **495**, 255–281, DOI `10.1017/S002211200300627X` | Closest analytical rotating-cylinder precedent for viscous wall conditions, asymptotic branches, order-of-limits failure and incomplete higher-order closure | `full-PDF`; PDF read for asymptotic structure and negative-route treatment; no derivation verification |
|
||||
| `mapped` | Remaining ten-paper R2 style-guide corpus and Undermind `JFM/`, `JFM References/01`, `/02`, `/07` indexes | Broader JFM style/context pool and source navigation | Located and metadata-mapped only in this lane; not reread and no new practice inferred from mapped-only papers |
|
||||
| `not-covered` | New literature/deep search; exhaustive sentence coding; full bibliography expansion; all Round-3 modules other than M08 and controlling audit | Not needed for this bounded section-style question | No claim inferred |
|
||||
| `not-covered` | New CFD, MFS, continuation, budgets, stability, figure generation, raw-field/NPZ review, hash recomputation or experiment analysis | Explicitly outside Round 4 style research | No result inferred; no panel promoted |
|
||||
| `external-inaccessible` | Authority-linked hidden/symlink-backed endpoint fields, dense MFS/NS payloads and facility media | Not needed to study style; required for any unsupported retained panel or reproducibility claim | Artifact audit deferred; this blocks stronger artifact/panel claims, not this style memo |
|
||||
|
||||
**Coverage finding.** All requested controls, R2 style/freeze inputs, R3 M08 and four closest JFM full PDFs were read. Evidence states remain distinct: `full-PDF`/`read` does not mean independently verified, and no literature paper establishes an internal result.
|
||||
|
||||
## 3. Observed JFM practice
|
||||
|
||||
The observations below are paper-specific; they are not universal rules inferred from four papers.
|
||||
|
||||
### 3.1 Contextual return and section progression
|
||||
|
||||
- **Deng et al. (2020), observed:** physical chronology structures the paper: flow configuration and bifurcation sequence precede incremental low-order modelling; primary and secondary regimes then test the corresponding models. The final phenomenogram (figure 14) synthesizes the chronology after the model comparisons. This is a genuine return to the physical sequence, not an isolated theory appendix.
|
||||
- **Cornejo Maceda et al. (2021), observed:** the results move from the uncontrolled baseline to one-parameter symmetric steady forcing (§4.1), general asymmetric steady forcing (§4.2), and learned feedback (§4.3). The simpler controls remain visible after the richer controller appears because they supply the comparison hierarchy and physical vocabulary.
|
||||
- **Mao et al. (2015), observed:** theory and numerical integrity are established before results; the results then move sensitivity → nonlinear optimum → control effect → physical interpretation → drag. Theory stays early because it defines the optimization actually used later.
|
||||
- **Ueda et al. (2003), observed:** governing equations and matched asymptotic regions lead directly to limiting procedures and force branches. The theory is the result rather than late context, so its long derivation is not a suitable structural template for a 6% contextual return.
|
||||
|
||||
**Style inference:** a late return works when it answers a narrower question created by earlier results. It should recover a simple reference or historical route and then expose what that reference can and cannot organize. Importing the full architecture of a theory-led paper into a short synthesis section would invert the paper’s evidence hierarchy.
|
||||
|
||||
### 3.2 Theory/model comparison
|
||||
|
||||
- **Deng et al., observed:** the idealized uncoupled model (equation (3.31)) is compared directly with DNS in figure 12 before the refined cross-coupled model (equation (5.5)) is shown in figure 13. The failed comparison is part of the proof for refinement, not omitted as an obsolete route.
|
||||
- **Cornejo Maceda et al., observed:** steady and feedback search spaces are compared under a common declared cost, with a compact final summary table/figure. Physical labels follow tested actuation branches; they do not substitute for the comparison.
|
||||
- **Mao et al., observed:** linear sensitivity and nonlinear optimum coincide at low control cost and diverge when the linear assumptions cease to hold (figure 4). The comparison defines the validity range of the simpler theory rather than declaring one model globally wrong.
|
||||
- **Ueda et al., observed:** alternative orders of limits are compared explicitly. One route retains an orientation-dependent limit and fails to recover the desired quiescent far field; the other yields the admissible steady limit. The failure identifies a mathematical non-commutation, not a generic physical mechanism.
|
||||
|
||||
**Style inference:** state the comparison contract first, show the attractive reference and finite-Re/observed counterpart in the same argumentative unit, and report the failed sign/closure test before offering synthesis. A theory curve or field without the decisive incompatibility beside it invites mechanism overreading.
|
||||
|
||||
### 3.3 Failed sign/closure routes and negative results
|
||||
|
||||
- **Deng et al., observed:** failure appears at the point where the uncoupled model misses transient and asymptotic features (§5.2, figure 12); the authors then add cross-terms and retain a residual limitation after refinement.
|
||||
- **Cornejo Maceda et al., observed:** the best symmetric base-bleed forcing is reported as failing to stabilize, while a strong Coanda/boat-tail branch can stabilize but remains far from the target and costly. A weaker route is not erased merely because a richer controller performs better.
|
||||
- **Mao et al., observed:** the linear sensitivity route is useful at small magnitude and explicitly loses validity at larger forcing; non-zero-frequency forcing is less effective; continuously distributed transpiration is acknowledged as non-realizable in the abstract and interpreted as a map for future localized actuation.
|
||||
- **Ueda et al., observed:** the Jeffery-paradox route is retained through the order-of-limits comparison; leading-order lift is zero, and higher-order lift closure is deferred because nonlinear inertial analysis is substantially more involved.
|
||||
|
||||
**Style inference:** classify each negative route by what exactly failed—sign, closure, approximation range, target, realizability or missing order—and preserve what remains usable. “Failed theory” is too coarse if a mathematical identity or outer reference survives while its finite-Re transfer fails.
|
||||
|
||||
### 3.4 Caveat placement
|
||||
|
||||
- **Observed across the four papers:** decisive caveats are local. Deng places missing coupling beside the failed ROM comparison; Cornejo Maceda places incomplete stabilization beside the feedback result; Mao flags non-realizability in the abstract and actuator-efficiency limits with the relevant effectiveness claim; Ueda places asymptotic-order and missing-higher-order limits with the force conclusions.
|
||||
- **Observed distinction:** broader scope limits may recur in conclusions, but conclusions do not perform the first repair of an overstrong result paragraph.
|
||||
|
||||
**Style inference:** use a two-level caveat pattern: (i) a short local limit in the same paragraph/caption as every attractive comparison; (ii) one closing envelope that names the remaining mechanism, stability and transfer boundaries. Do not stack all warnings in the last paragraph.
|
||||
|
||||
### 3.5 Equations and figure selection
|
||||
|
||||
- **Deng et al., observed:** equations are introduced by modelling need; the argument is carried by the uncoupled/refined equation pair and their adjacent DNS comparisons, then compressed by one phenomenogram.
|
||||
- **Cornejo Maceda et al., observed:** one reusable objective equation precedes the search-space progression; branch figures and a compact synthesis table/figure carry comparison. The complex learned law is displayed only because it is itself an object of analysis.
|
||||
- **Mao et al., observed:** the methodology legitimately carries many equations because adjoint derivation is a principal contribution; results figures then test linear/nonlinear overlap, saturation, fields and stability. This equation density is inappropriate when theory is only an outer reference.
|
||||
- **Ueda et al., observed:** a high equation count is justified by matched asymptotics; a domain map and a small number of force-versus-gap plots compress the derivation into interpretable regimes and branch consequences.
|
||||
|
||||
**Style inference:** equation count should follow argumentative ownership, not mathematical availability. For a contextual steady/theory return, retain only definitions reused to read the displays and the shortest equation that exposes the outer-reference boundary. Move MFS, strip-image, multipole and control-volume derivations to supplement unless the main-text inference truly depends on them.
|
||||
|
||||
## 4. Project recommendation: paragraph jobs and interfaces
|
||||
|
||||
### Recommended five-paragraph progression
|
||||
|
||||
1. **Return and narrowed question.** Recall the early steady discovery seed in one sentence; state that this late return asks whether a separate steady branch supplies context for the already established near-constant component. Import only the minimum Kármán conclusion and immediately prohibit continuation/equivalence.
|
||||
2. **Finite-Re contract and comparison.** Establish V5 chain, solver/geometry, Reynolds convention, body/action/sign order, comparator, settling/evaluation window and steady estimand. Present the accepted endpoint against stationary/reverse or nearby endpoint context only to the extent required by the selected display. Put endpoint-not-optimum and reverse-amplitude-mismatch caveats here.
|
||||
3. **Bounded physical description.** Read the action-sign schematic before fields. Describe gap-facing/base-bleed or gap-jet-family compatibility as observation/hypothesis. Put absent causal arrow, force ownership and closed budget in this paragraph; do not defer them.
|
||||
4. **Outer reference and failed route.** Introduce the shortest one-circle or partial-budget equation, define what it constrains, then place circulation independence/rotating-no-slip limitation and the failed sign comparison immediately beside it. Preserve surviving mathematics while marking the finite-Re closure as failed or incomplete. No new MFS field unless all validation gates close.
|
||||
5. **Synthesis export.** Relate separate steady, SR and field observations only as compatible context; state the different-chain/different-estimand boundary once. End with the decisive missing test and a bridge to qualitative experiment/conclusion, not with another result inventory.
|
||||
|
||||
### Result ownership and duplication deletions
|
||||
|
||||
- **Owned here:** V5 steady contract/sign, contextual endpoint status, accepted/reverse boundary, shortest outer-reference/failure boundary, incomplete finite-Re closure/budget, and the late-return bridge.
|
||||
- **Imports:** contracts section supplies reusable body/action/Re/metric definitions where already frozen; SR supplies only the bounded near-constant-term conclusion; field/CCD supplies only the bounded mean/dynamic organization; cross-case supplies the reason interpretation did not transfer.
|
||||
- **Exports:** a constrained compatible-context statement, explicit theory-failure boundary, decisive test, and qualitative-experiment bridge.
|
||||
- **Delete during assembly:** repeated SR formula/deletion details; detailed CCD shares/residual construction; full M04/M06 near-collision explanation; full steady sweep; package/manifest histories; repeated experiment provenance checklist. Those belong to their owners or supplement.
|
||||
- **Opening dependency:** Kármán persistent/smaller-correction interpretation and cross-case non-transfer must already be established.
|
||||
- **Closing bridge:** the computational evidence supplies bounded context, while experiment addresses qualitative deployability only and cannot close the quantitative or causal gap.
|
||||
|
||||
## 5. Claim ceiling and near-collision audit
|
||||
|
||||
### Claim ceiling
|
||||
|
||||
- **Green:** `defines`, `shows under the stated contract`, `has the declared sign`, `passes/fails the stated gate`, `is a separate endpoint observation`, `the outer reference leaves circulation unspecified`, `the sign/closure route fails`.
|
||||
- **Amber:** `is compatible with`, `provides context for`, `belongs to a base-bleed/gap-jet-family description`, `constrains the interpretation`, `motivates a bounded control-skeleton reading`.
|
||||
- **Prohibited red without new authority:** `explains`, `mechanism`, `causes`, `restores momentum`, `validates the Kármán skeleton`, `selects circulation from wall speed`, `optimal`, `stable`, `robust`, `universal`, `equivalent`, `replicates`, `triangulates`, or any quantitative experimental-validation wording.
|
||||
|
||||
### Near-collision findings
|
||||
|
||||
| Pair | Acquisition/roles | Domain, mask, weighting and window | Classification | Writing rule |
|
||||
|---|---|---|---|---|
|
||||
| Legacy SR persistent-action magnitude vs V5 steady spin ratio | Separate chains, action parameterizations, plants and objectives | Action-law/window diagnostic versus steady endpoint profile evaluation | `related diagnostic` | No amplitude agreement, arithmetic, continuation, validation or replication wording |
|
||||
| Legacy corrected constant/DRL mean-field comparison vs V5 steady endpoint | Different role sets and acquisitions | Legacy ROI/common mask/coordinate weighting/retained window versus V5 full-cross-section steady profile metric/settling window | `different estimand` | Only compatible contextual organization; no shared percentage or closure |
|
||||
| V5 accepted vs reverse branch | Separate endpoint acquisitions; reverse is not an amplitude-matched intervention | Same broad steady family but different sign/amplitude and endpoint status | `related diagnostic` | Use as a sign/branch boundary, not causal mechanism, stability or symmetry proof |
|
||||
| Outer potential/MFS reference vs finite-Re V5 endpoint | Prescribed-circulation mathematical family versus rotating no-slip CFD observation | Different governing equations, wall conditions, fields and metrics | `different estimand` | Compare only through an explicitly defined sign/projection test; disagreement bounds the route and agreement would not validate mechanism |
|
||||
|
||||
No arithmetic combination, replication, validation, agreement or independent-triangulation language is licensed by these comparisons.
|
||||
|
||||
## 6. Figure, table and equation contracts
|
||||
|
||||
| Item | Owner / argumentative job | Source and required roles/arrays | Caption ceiling | Classification and readiness | Fallback |
|
||||
|---|---|---|---|---|---|
|
||||
| Action/sign schematic | Steady/theory; establish body order, wall law and gap-/outer-facing directions before physical labels | Current V5 contract/manifest-bound wall oracle; no field array required | Establishes action geometry only; not circulation, gap-jet causality, boat-tail equivalence or force direction | `required`; text/source-ready, redraw/production pending; artifact audit deferred | Compact contract table with cardinal directions |
|
||||
| Endpoint comparison | Steady/theory; show accepted contextual endpoint and bounded reverse/stationary comparison under the steady metric | Exact V5 roles, comparator, profile arrays, settling window and source/render lineage | Separate chain and metric; endpoint not optimum/stability; reverse not amplitude matched; no Legacy comparison | `dependency-gated`; text ledger ready, panel needs bound arrays/render provenance | Endpoint table with status/caveat, or prose-only comparison |
|
||||
| Outer-reference/sign-failure display | Steady/theory; make the attractive theoretical route and decisive incompatibility visible together | Existing one-circle identity and authority-bound outer/finite-Re sign test; no newly generated field | Outer impermeability/reference only; no rotating no-slip finite-Re mechanism or closure | `required` as an argumentative element; equation/sign schematic ready, new field blocked | One-circle trace plus compact sign-conflict annotation |
|
||||
| MFS field | Steady/theory; only if it adds a validated spatial outer-reference comparison | Dense off-grid body/sector residuals, conditioning/rank, source-radius/order/image-layer convergence, independent reference, exact source arrays and rendering provenance | Mathematical outer reference only; never finite-Re no-slip physics or mechanism | `blocked` until all gates close; no current publication-ready field implied | Omit; retain one-circle/sign-failure boundary |
|
||||
| Integrated evidence map | Conclusion or steady/theory by coordinator choice; show compatible and broken links without causality | Chain/estimand/status labels from owning sections; no pooled values | Dashed links mean context, not proof; broken links mark missing closure | `optional`; text concept ready, visual design dependency-gated | One synthesis paragraph plus limitations table |
|
||||
|
||||
### Minimum equation budget
|
||||
|
||||
1. **Wall law/spin ratio:** include only if not already reusable from contracts; otherwise cite/import the definition and use the sign schematic.
|
||||
2. **Steady metric:** include its definition only if readers cannot interpret the endpoint display from the contracts section; do not redefine spatial L2 or DTW here.
|
||||
3. **Outer-reference boundary:** prefer the one-circle tangential trace because one equation simultaneously shows impermeability, free circulation and the nonzero-inflow obstruction. If the partial Navier–Stokes budget is more central in final assembly, substitute one compact budget identity rather than adding both.
|
||||
4. **Supplement:** full MFS system, source constraints, strip images, multipole expansion, convergence diagnostics and complete control-volume derivation.
|
||||
|
||||
**Cadence:** definition/sign → empirical comparison → bounded reading → outer-reference equation → failed sign/closure → consequence. Never place several pages of derivation between the finite-Re observation and the failure that limits it.
|
||||
|
||||
## 7. Anti-patterns
|
||||
|
||||
- A detached “theory chapter” that restarts the paper after the Kármán and cross-case argument.
|
||||
- Repeating the early steady discovery as a full second capability section.
|
||||
- Presenting a potential/MFS field first and revealing its sign conflict several paragraphs later.
|
||||
- Treating `failed theory` as total invalidation when the outer mathematical identity survives but its finite-Re closure fails.
|
||||
- Treating a visually attractive outer field or small collocation residual as finite-Re physical validation.
|
||||
- Imposing rotating-wall tangential speed as an inviscid boundary condition or asserting a universal circulation–spin law.
|
||||
- Naming a project branch from another paper’s body order/sign convention without a direct wall-velocity map.
|
||||
- Calling a low-deficit endpoint an optimum, stable branch or momentum-restored state.
|
||||
- Using the reverse branch as an amplitude-matched intervention when it is not.
|
||||
- Joining Legacy SR, Legacy CCD and V5 steady values because dimensionless labels look similar.
|
||||
- Equation dumping: deriving MFS, images, multipoles and budgets in main text although only one boundary identity changes the inference.
|
||||
- Figure inventory prose, package histories, or captions that imply a proposed panel already exists.
|
||||
- Delaying all mechanism/stability/closure caveats to Discussion or converting every negative result into generic future work.
|
||||
|
||||
## 8. Author/coordinator decisions, conflicts, deferrals and blockers
|
||||
|
||||
1. **Outer-reference equation choice — `discuss`.** Recommendation: use the one-circle trace because it exposes the exact inviscid boundary limitation with the least machinery. Consequence: the partial NS budget moves to text/supplement. Safe default: one-circle trace only.
|
||||
2. **Endpoint panel compression — `discuss`.** Recommendation: retain a compact sign schematic and endpoint table/one profile comparison; avoid a field gallery. Consequence: preserves space for the failed route. Safe default: table plus sign schematic.
|
||||
3. **MFS field — `blocked`.** Do not retain unless all declared validation, source-array and rendering gates close. Safe default: omit; this does not block the text contract.
|
||||
4. **Integrated evidence map — `discuss`.** Recommendation: let the conclusion own it if used, because this section should not visually imply causal closure. Safe default: omit and use one synthesis paragraph.
|
||||
5. **Conflict to preserve, not repair:** the outer-reference/finite-Re sign conflict blocks that proposed cancellation route. It is an argumentative result boundary, not a reason to redefine circulation, swap branch labels or add an unsupported closure.
|
||||
6. **Authority-led deferrals:** raw endpoint leaves, dense theory arrays and universal hashes remain artifact-audit deferred. They block unsupported panels/reproducibility claims, not the bounded style recommendation.
|
||||
|
||||
## 9. Disposition
|
||||
|
||||
**`researching`.** The requested literature and controlling documents have been read and the style recommendation is decision-ready, but this first-pass lane cannot self-freeze. A coordinator must perform the required second audit for ownership, equation/display compression, DOI/read-level fidelity, near-collision language and text-versus-artifact readiness.
|
||||
@@ -0,0 +1,138 @@
|
||||
# Round 4 style research — qualitative experiment
|
||||
|
||||
**Status:** `researching` (first-pass style memo; not manuscript prose or a freeze). **Owner:** experiment style lane; Frank retains authorship. External literature supplies style/context only and establishes no internal apparatus, result, artifact or priority.
|
||||
|
||||
## 1. Question and bounded recommendation
|
||||
|
||||
**Question.** How should a JFM-facing qualitative experiment move from apparatus to observation while preserving provenance and preventing computational evidence from becoming experimental validation?
|
||||
|
||||
**Recommendation.** Use **purpose/boundary → apparatus, actuation and acquisition contract → named visual comparison → immediate measurement limit → next quantitative test**. A traceable image may support only a visible named-condition observation or bounded physical feasibility. It cannot inherit field, force, mechanism or validation language from computation. Omit media if provenance fails. Use a supplementary movie only when motion is essential and trial linkage is complete.
|
||||
|
||||
## 2. Coverage ledger
|
||||
|
||||
| Status | Source/input | Why in scope | Evidence state/reason |
|
||||
|---|---|---|---|
|
||||
| `read` | Round-4 `00_MASTER_AGENT_PROMPT.md`, `00_QUALITY_GATE.md`, `00_GLOBAL_DECISION_LEDGER.md`, `README.md` | Claim, ownership and disposition controls | Read directly; not independently verified or artifact-audited. |
|
||||
| `read` | Round-4 `00_DISPLAY_EQUATION_INVENTORY.md` | Gate-A experiment-display classification, ownership, dependencies and fallbacks | Read directly; inventory entry reconciled below; no panel, source media or artifact independently verified or audited. |
|
||||
| `read` | Round-2 `05_REVISED_STRUCTURE_FREEZE.md`, `06_SECTION_LEVEL_JFM_STYLE_GUIDE.md` | Experiment ceiling and section cadence | Full reports read; not independent verification. |
|
||||
| `read` | Round-3 `09_EXPERIMENT_AUTHOR_INTERFACE.md`, `00_MODULE_INTERFACE_LEDGER.md`, `11_FINAL_CONSISTENCY_AUDIT.md` | Author ownership, provenance gate and fallback | Full reports read; internal experiment remains author-described and unaudited. |
|
||||
| `read` | Four exact JFM papers listed in §3 | Closest qualitative visualization/control precedents | Targeted `full-PDF` paper-level read of metadata, apparatus, acquisition, visual results, captions and supplement practice; no data audit or replication. |
|
||||
| `mapped` | Supplementary-movie records linked by Bhattacharyya et al. and Khan et al. | Motion-dependent reporting precedent | Identities and links mapped; movies not watched or artifact-audited. |
|
||||
| `mapped` | Internal raw-media/provenance slots named by M09 | Defines possible gates | Located through M09 only; mapping establishes no fact. |
|
||||
| `not-covered` | Exhaustive JFM facility literature; sentence coding of all captions; new experiment/image/force/PIV analysis | Outside bounded style task | No general frequency or internal claim inferred. |
|
||||
| `external-inaccessible` | Hidden/unexposed internal media, trial logs, calibration, synchronization and permissions | Needed for panel/observation promotion | Provenance unresolved; no claim inferred. |
|
||||
|
||||
All requested controls were read. `Read` is not independently verified; no literature data or internal artifact was audited.
|
||||
|
||||
## 3. Exact literature identities
|
||||
|
||||
1. **Soumarup Bhattacharyya, Izhar Hussain Khan, Shivam Verma, Sanjay Kumar & Kamal Poddar**, “Experimental investigation of three-dimensional modes in the wake of a rotationally oscillating cylinder”, *Journal of Fluid Mechanics* **950**, A10 (2022), DOI `10.1017/jfm.2022.792`, `full-PDF`. Closest single rotating-cylinder example: qualitative hydrogen-bubble/LIF, phase selection, separate PIV/hot-wire, validation and movies.
|
||||
2. **Izhar Hussain Khan, Puja Sunil, Soumarup Bhattacharyya, Rahul Yadav, Kamal Poddar & Sanjay Kumar**, “Flow past two rotationally oscillating cylinders”, *Journal of Fluid Mechanics* **969**, A16 (2023), DOI `10.1017/jfm.2023.549`, `full-PDF`. Closest geometry/actuation family: visual-first sequence, distinct PIV/force acquisition and extensive movie use.
|
||||
3. **Lukasz Klotz, Karol Bukowski & Konrad Gumowski**, “Influence of porous material on the flow behind a backward-facing step: experimental study”, *Journal of Fluid Mechanics* **998**, A31 (2024), DOI `10.1017/jfm.2024.639`, `full-PDF`. Explicitly qualitative dye entry followed by a separate quantitative PIV campaign.
|
||||
4. **Yu Zhou, Dewei Fan, Bingfu Zhang, Ruiying Li & Bernd R. Noack**, “Artificial intelligence control of a turbulent jet”, *Journal of Fluid Mechanics* **897**, A27 (2020), DOI `10.1017/jfm.2020.392`, `full-PDF`. Machine-learning-control experiment in which facility/actuator/sensor provenance precedes performance and visualization remains subordinate to quantitative evidence.
|
||||
|
||||
## 4. Observed JFM practice
|
||||
|
||||
These are observations from this four-paper sample, not universal JFM rules.
|
||||
|
||||
### Apparatus-to-observation sequence
|
||||
|
||||
- Papers fix the question, then schematic/facility/geometry, actuation, acquisition, and only then observation. Hardware is organized by evidentiary function, not as a build inventory.
|
||||
- Visualization has its own contract: tracer, illumination, camera/view, plane/field of view, sampling, phase/frame choice and processing.
|
||||
- Comparator structure precedes interpretation. Khan et al. vary declared parameters while stating held conditions; Klotz et al. pair reference and modified cases.
|
||||
- Different acquisitions are disclosed. Bhattacharyya et al. call PIV and visualization different realizations under the same conditions; Klotz et al. say visualization and PIV came from different runs. “Same conditions” is not treated as synchronized or paired evidence.
|
||||
|
||||
### Provenance and qualitative wording
|
||||
|
||||
- Provenance is distributed deliberately: methods carry facility/acquisition/processing; captions carry exact condition, view, phase/scaling and annotations.
|
||||
- Intrusive or consequential choices are visible. Bhattacharyya et al. discuss visualization-wire placement and possible wake modification; Klotz et al. disclose dye delivery and background-intensity subtraction.
|
||||
- Klotz et al. explicitly call dye panels qualitative and use `illustrate`. Visual prose names morphology—streaklines, rows, coherence, symmetry, deflection, repetition—under a condition, without automatically converting appearance into velocity or force.
|
||||
- Stronger conclusions follow distinct diagnostics: image analysis, hot-wire, PIV, spectra or force. Strong words used in those papers are not transferable to a qualitative-only demonstration.
|
||||
|
||||
### Quantitative-validation boundary
|
||||
|
||||
- Adjacent visual and quantitative evidence remains acquisition-specific: camera, tracer, plane, run, phase and processing differences are stated.
|
||||
- Visual observations organize subsequent measurements but do not substitute for them.
|
||||
- Visual morphology becomes quantitative only after an estimator is defined. Bhattacharyya et al. specify sampled line, pixel intensity and spectral procedure, and note position dependencies.
|
||||
- Quantitative uncertainty belongs to the relevant calibrated instrument/estimator, not retroactively to a dye image.
|
||||
|
||||
### Captions, supplement and omission
|
||||
|
||||
- Effective captions identify comparison variables and fixed conditions, view/plane, flow direction, phase/averaging state, scaling/extent and annotations. They distinguish visualization from PIV and instantaneous from averaged fields.
|
||||
- Acquisition hardware and algorithms stay in Methods unless essential to reading a panel; comparator, condition, view and phase cannot be omitted from the caption.
|
||||
- Bhattacharyya et al. and Khan et al. use movies for cycle evolution, repetition, alternation or coalescence that a still cannot establish. The interpretive sentence points to a named movie.
|
||||
- Main figures select representative comparisons/maps rather than every condition. A movie supplements a traceable observation; it does not repair an unidentified still.
|
||||
- Klotz et al. retain qualitative plates because they provide the visual entry to later PIV. Zhou et al. subordinate visualization when quantitative performance and fields already carry the argument.
|
||||
|
||||
## 5. Project recommendation from observed practice
|
||||
|
||||
### Paragraph jobs
|
||||
|
||||
1. **Purpose/boundary:** why physical deployment is shown; explicitly qualitative and open-loop. Import no computational values or interpretation.
|
||||
2. **Apparatus/acquisition:** facility/geometry, command/sign, trial/reference roles, camera/view/scale/timing, tracer and synchronization status. Avoid construction history and brands unless evidence-determining.
|
||||
3. **Observation:** one comparator pair at a time; visible morphology, trial/media identity, frame-selection and repeat status. Use past tense and bounded verbs: `appeared`, `was visibly altered`, `a qualitative change was observed`.
|
||||
4. **Limit/next test:** immediately state no quantitative field/force/PIV or closed-loop validation; name the acquisition needed to raise the ceiling.
|
||||
|
||||
Two paragraphs may combine 1–2 and 3–4. A one-paragraph image anecdote is not ready.
|
||||
|
||||
| Dimension | Recommendation | Safe fallback |
|
||||
|---|---|---|
|
||||
| Apparatus | Functional schematic or compact facility → actuation → acquisition prose | Omit schematic if it becomes a hardware diary. |
|
||||
| Observation | One visual claim per named comparator; trial, view and selection rule local | Weakest wording or omission if repeat/selection is unknown. |
|
||||
| Provenance | Bind raw file, trial, condition, timing/scale, processing, permission and repeats | Omit; never reconstruct from memory/derivative plots. |
|
||||
| Wording | Morphology/appearance only; distinguish visible temporal change from measured frequency | No frequency without timing, extraction and uncertainty. |
|
||||
| Quantitative boundary | Keep computational/experimental chains separate; no overlay/shared metric/validation without same-plant registration and uncertainty | Feasibility only. |
|
||||
| Caption | Condition/role, view, frame/window, scale/calibration status, processing, trial and qualitative ceiling | Move extended provenance to supplement, not comparator/ceiling. |
|
||||
| Supplement | Named movie only when motion matters; include provenance index | Omit unlinked movie. Supplement does not lower standards. |
|
||||
| Omission | Keep media only if it changes the argument beyond hardware existence | Text-only, supplement-only or total omission. |
|
||||
|
||||
**Imports:** fixed vocabulary and qualitative/open-loop/provenance-gated boundary only. **Experiment owns:** apparatus/media provenance, trial identity, observation, local confounds, panel gate and next measurement. **Exports:** one conditional feasibility statement and one decisive measurement boundary. **Delete here:** computational values, SR/CCD, field metrics and mechanistic synthesis. **Delete elsewhere:** apparatus checklist/provenance detail; conclusion keeps only the boundary. **Bridge:** deployment does not alter computational evidence; it identifies the next synchronized/calibrated/repeated test.
|
||||
|
||||
## 6. Claim ceiling and near-collision audit
|
||||
|
||||
**Green:** `qualitative`, `open-loop`, `author-described`, `provenance-gated`, `feasibility`; `selected media appear to show [visible morphology] under the named trial/comparator` after provenance passes.
|
||||
|
||||
**Amber:** `is consistent with qualitative wake modification under the named condition`; `supports a bounded physical-deployability reading`; `motivates synchronized, calibrated and repeated measurement`. No computational mechanism or policy-validity import.
|
||||
|
||||
**Red without new evidence:** `experimentally validates`, `confirms CFD`, field matching, force reduction, closed-loop control, policy validation, measured frequency from unaudited video, causation/mechanism, restoration, stability, optimality, robustness/generality, or quantitative efficacy/uncertainty from appearance.
|
||||
|
||||
| Pair | Check | Classification / consequence |
|
||||
|---|---|---|
|
||||
| Dye/streakline image vs computational field | Different plant/acquisition; intensity/morphology is not velocity; no common roles, coordinates, mask, weighting or window | `different estimand`; no overlay, arithmetic, agreement, replication or validation. |
|
||||
| Visible shedding change vs temporal metric | Timing, extraction, trial role and window unbound | `different estimand`; visible temporal change only, no frequency/DTW equivalence. |
|
||||
| Open-loop physical actuation vs closed-loop computation | Commanded/realized motion, sensing and feedback differ | `related diagnostic` at most; no policy validation. |
|
||||
| Apparatus schematic vs result image | Configuration documentation vs observation | `different estimand`; schematic never proves outcome. |
|
||||
|
||||
## 7. Display contracts and readiness
|
||||
|
||||
| Display | Owner/job/source | Caption ceiling | Status/fallback |
|
||||
|---|---|---|---|
|
||||
| Apparatus/signal schematic | Experiment; document facility, geometry, open-loop command/media path from author record | Context only; no calibration, accuracy or result | `author-supply-gated`; prose fallback. |
|
||||
| Qualitative comparator plate | Experiment; candidate vehicle for one Frank-authored, named visible observation only after raw traceable minimally processed media pass the gate | Trial/comparator, view, timing/phase, scale, processing, repeats and qualitative ceiling; no field/force claim | `author-supply-gated`; currently no observation is promoted and no panel is ready. Frank may supply a provenance-passing bounded observation and panel; otherwise use a prose boundary or omit (supplement-only media remain equally gated). |
|
||||
| Supplementary movie | Experiment; candidate support for temporal evolution that a still cannot establish | Qualitative motion only; trial/condition/timing identified; no measured frequency or repeatability claim without a declared estimator and audit | `author-supply-gated`; Frank-supplied provenance must pass before use; otherwise use a gated still, bounded prose, or omission. |
|
||||
| Provenance index | Experiment/supplement; bind media and acquisition from Frank's records | Documentation, not efficacy and not an observation | `author-supply-gated`; Frank supplies the binding record; otherwise use only a bounded caption/data note for retained gated media, or omit media. |
|
||||
| Force/PIV/registered-CFD panel | Not owned in qualitative pass | Requires independent quantitative contract | `blocked`; omit. |
|
||||
|
||||
**Inventory reconciliation.** Round-4 `00_DISPLAY_EQUATION_INVENTORY.md` classifies the experiment display family as `author-supply-gated`: qualitative open-loop apparatus and candidate steady/Illusion dye media require Frank's conditions, action/sign, trial identity, raw media, scale/frame rate, processing, repeats and permissions. This memo does not treat the inventory wording as evidence that either dye observation exists or is publishable. The apparatus/signal schematic, comparator plate, supplementary movie and provenance index therefore remain `author-supply-gated`; the force/PIV/registered-CFD panel is `blocked`. Each row above states its prose or omission fallback. Frank retains authorship of any apparatus account, observation and media selection. The only ceiling is qualitative, open-loop and provenance-gated; no observation is promoted here and no display provides quantitative validation.
|
||||
|
||||
No experiment equation is recommended; introducing one here would imply an unsupported quantitative experimental estimand. Any quantitative experiment equation or main-text results table is `blocked` absent a separately authorized, calibrated estimator and evidence contract, with prose/omission as fallback. Text-interface readiness does not imply any panel, movie, permission or caption is artifact-ready.
|
||||
|
||||
## 8. Anti-patterns
|
||||
|
||||
Hardware diary; aesthetic image without trial/comparator/selection provenance; “with/without control” caption lacking condition/view/timing; dye as velocity/vorticity/force; unregistered CFD overlay; “same conditions” as simultaneous; unlinked movie as proof; quantitative verbs without estimator/uncertainty; single-cause background attribution without controls; open-loop playback as policy validation; limitation delayed to conclusion; supplement used to relax provenance; retaining an untraceable panel merely to lengthen the section.
|
||||
|
||||
## 9. Decisions, deferrals and blockers
|
||||
|
||||
**Evidence-governed:** qualitative/open-loop/provenance-gated ceiling; no quantitative or causal ownership; missing provenance blocks panel promotion, not this style interface.
|
||||
|
||||
**Author choices:** (1) retain subsection only if provenance supports a named observation, otherwise a brief limit/outlook sentence; (2) main-text still, movie or no media—safe default omit; (3) schematic only if more efficient than prose—safe default prose; (4) Frank supplies exact observation after raw-media review—safe default no sentence yet.
|
||||
|
||||
No control conflict was found. External papers often move from visualization to PIV/force/spectra; only their reporting discipline transfers when those measurements are absent. Raw-media, apparatus, synchronization, calibration, repeatability and permission audits are deferred to author supply.
|
||||
|
||||
**Blockers to media/observation:** missing raw identity, trial/comparator, condition, view/timing/scale, processing, selection/repeats or permission. **Blocker to quantitative wording:** no synchronized calibrated force/PIV or declared image estimator with uncertainty. These do not block the style memo.
|
||||
|
||||
## 10. Disposition and validation
|
||||
|
||||
**Disposition:** `researching`. Recommendation is decision-ready, but a first pass cannot self-freeze. Coordinator second audit is required; all candidate qualitative experiment displays remain `author-supply-gated`, while the quantitative force/PIV/registered-CFD panel remains `blocked`.
|
||||
|
||||
Validated: exact requested output only; required R4/R2/R3 controls and Round-4 `00_DISPLAY_EQUATION_INVENTORY.md` read; every experiment display reconciled as `author-supply-gated` or `blocked` with prose/omission fallback; Frank authorship preserved; qualitative open-loop provenance-gated ceiling preserved; no observation promoted; four close JFM full-PDF records retain exact identities/DOIs/read levels; observed practice remains separated from recommendation; no internal numbers, artifacts, quantitative validation or new scientific claims; text/artifact readiness remains separate.
|
||||
@@ -0,0 +1,274 @@
|
||||
# Round 4 style research — Discussion and Conclusion
|
||||
|
||||
**Role.** Section-style research memo for the Discussion/Conclusion unit. This is not manuscript prose, scientific evidence, or a section freeze. External papers establish rhetorical practice only; repository controls set the project's evidence and claim boundaries.
|
||||
|
||||
**Disposition:** `researching` — the required style pass is complete, but a first-pass style agent cannot self-freeze the later section contract.
|
||||
|
||||
## 1. Scope, question, and bounded recommendation
|
||||
|
||||
**Question.** How should the final Discussion and Conclusion be separated and paced so that synthesis follows evidence order, limitations remain local and visible, future work names decisive claim-changing tests, and the closing cadence does not exceed the fixed claim ceiling?
|
||||
|
||||
**Bounded recommendation.** Use a distinct integrated Discussion followed by a short Conclusion. The Discussion should synthesize in the same order in which the evidence earns progressively narrower inferences, distinguish compatible observations from same-estimand evidence, place each dangerous limitation beside the inference it limits, and end with a small set of tests tied explicitly to unresolved links. The Conclusion should answer the paper question in three compact moves—named capability, bounded evaluation/interpretation, strongest boundary plus decisive next test—without replaying methods, values, citations, equations, or panel descriptions.
|
||||
|
||||
This recommendation is evidence-governed rather than an author-style choice because the fixed ownership map assigns integrated answer, limits, and decisive tests to this unit while prohibiting new evidence. The only later author choice is whether an optional synthesis aid improves comprehension enough to justify space; the safe default is a compact limitations table and no new summary figure.
|
||||
|
||||
## 2. Coverage ledger
|
||||
|
||||
| Status | Source/input | Why in scope | Evidence state/reason |
|
||||
|---|---|---|---|
|
||||
| `read` | `ROUND4_TOP_DOWN/README.md` | Round-4 purpose, phases, status semantics, controlling-input order | Read directly; not independently verified or artifact-audited |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_MASTER_AGENT_PROMPT.md` | Required method, fixed claim ladder/cautions, deliverable contract | Read directly; not independently verified or artifact-audited |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_QUALITY_GATE.md` | Style PASS/BLOCK criteria and first-pass no-self-freeze rule | Read directly; not independently verified or artifact-audited |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_GLOBAL_DECISION_LEDGER.md` | Conclusion ownership, duplicate limits, claim ceiling, open choices | Read directly; not independently verified or artifact-audited |
|
||||
| `read` | `ROUND4_TOP_DOWN/00_DISPLAY_EQUATION_INVENTORY.md` | Conclusion display/equation ownership and readiness distinctions | Read directly; inventory is not panel evidence; no artifact audit |
|
||||
| `read` | `ROUND2_TOP_DOWN/05_REVISED_STRUCTURE_FREEZE.md` | Author-ratified paper identity, order, scope and section-space boundary | Read directly as controlling full report; not independently verified or artifact-audited |
|
||||
| `read` | `ROUND2_TOP_DOWN/06_SECTION_LEVEL_JFM_STYLE_GUIDE.md` | Prior ten-paper style synthesis and Discussion/Conclusion contract | Read directly as full report; its literature read levels were checked against the four PDFs below, not universally re-audited |
|
||||
| `read` | `ROUND3_BOTTOM_UP/10_LIMITATIONS_CONCLUSION.md` | Closest section dossier: evidence order, limits, tests, interfaces and display proposals | Read directly as full report; project claims were not recomputed and artifacts were not audited |
|
||||
| `read` | `ROUND3_BOTTOM_UP/00_MODULE_INTERFACE_LEDGER.md` | Import/export readiness and conclusion dependency chain | Read directly; readiness is dossier readiness, not manuscript or artifact approval |
|
||||
| `read` | `ROUND3_BOTTOM_UP/11_FINAL_CONSISTENCY_AUDIT.md` | Final ownership, duplication, near-collision and panel gates | Read directly; audit findings not independently recomputed |
|
||||
| `read` | N. Deng, B. R. Noack, M. Morzyński & L. R. Pastur, “Low-order model for successive bifurcations of the fluidic pinball”, *Journal of Fluid Mechanics* **884**, A37 (2020), DOI `10.1017/jfm.2019.959` | Closest pinball paper; final phenomenogram and combined conclusion/outlook architecture | `full-PDF`; PDF available and final sections read directly through Undermind; rhetorical reading, not equation/result verification |
|
||||
| `read` | G. Y. Cornejo Maceda, Yiqing Li, François Lusseyran, Marek Morzyński & Bernd R. Noack, “Stabilization of the fluidic pinball with gradient-enriched machine learning control”, *Journal of Fluid Mechanics* **917**, A42 (2021), DOI `10.1017/jfm.2021.301` | Closest pinball-control paper; explicit Discussion/Conclusion split and closing synthesis displays | `full-PDF`; PDF available and final sections read directly through Undermind; rhetorical reading, not result verification |
|
||||
| `read` | Ji-chao Li & Mengqi Zhang, “Reinforcement-learning-based control of confined cylinder wakes with stability analyses”, *Journal of Fluid Mechanics* **932**, A44 (2022), DOI `10.1017/jfm.2021.1045` | Physics/control integration, distributed discussion, explicit limits and outlook | `full-PDF`; PDF available and final sections read directly through Undermind; rhetorical reading, not result verification |
|
||||
| `read` | Jean Rabault, Miroslav Kuchta, Atle Jensen, Ulysse Réglade & Nicolas Cerardi, “Artificial neural networks trained through deep reinforcement learning discover control strategies for active flow control”, *Journal of Fluid Mechanics* **865**, 281–302 (2019), DOI `10.1017/jfm.2019.62` | Compact control paper with interpretation before a short textual Conclusion | `full-PDF`; PDF available and final sections read directly through Undermind; rhetorical reading, not result verification |
|
||||
| `mapped` | Remaining six exact JFM papers in the Round-2 style guide and `JFM/`, `JFM References/` indexes | Broader precedent already synthesized by the controlling style report | Identity and prior read level mapped only in this lane; PDFs not re-read here, so no new practice claim depends on them |
|
||||
| `mapped` | R3 Modules 03–09 and authority/manifests referenced by M10 | Underlying owners imported by the final synthesis | Located through M10/interface/audit; not re-read as section evidence in this style-only lane |
|
||||
| `not-covered` | New CFD, training, SR/CCD analysis, field arrays, masks, hashes, experiment media and new figures | Outside style research; literature cannot establish internal results | No claim inferred; no independent verification or artifact audit attempted |
|
||||
| `not-covered` | Exhaustive sentence coding across JFM or a new broad literature search | Four closest full PDFs plus the controlling ten-paper report are sufficient for this bounded rhetorical question | No universality claim about all JFM practice |
|
||||
| `external-inaccessible` | Hidden/external project artifacts and provenance-gated experiment media named by R3 | Not needed to determine closing-section style; required only if later prose retains dependent claims/panels | Explicitly deferred; no artifact readiness inferred |
|
||||
|
||||
## 3. Observed JFM practice
|
||||
|
||||
The observations below are sample-bounded. They describe these four papers, not a universal JFM rule.
|
||||
|
||||
### 3.1 Discussion/Conclusion separation
|
||||
|
||||
- **Deng et al.** use one final top-level section, “Conclusions and outlooks”, subdivided into concluding discussion and outlook. The architecture separates synthesis from prospective work without creating independent top-level Discussion and Conclusion sections.
|
||||
- **Cornejo Maceda et al.** explicitly separate a thematic Discussion from a continuous Conclusions section. Discussion moves through feedback, sensitivity/complexity and applicability; Conclusions compress the study through the hierarchy of control-law spaces.
|
||||
- **Li & Zhang** use no separate Discussion section. Interpretation is distributed through results, and a dedicated Conclusions section mirrors the paper's two main evidence blocks before moving to limits and outlook.
|
||||
- **Rabault et al.** place the physical reading in the last results subsection, then use a compact Conclusion for achievement, constraint and prospective extension.
|
||||
|
||||
**Observed pattern.** Structural labels vary, but interpretive work is done before the final closing paragraph. A distinct Discussion is most useful when several non-equivalent evidence streams must be integrated; a short Conclusion works when local interpretation and caveats have already been earned upstream.
|
||||
|
||||
### 3.2 Evidence-order synthesis
|
||||
|
||||
- Deng et al. move from the physical sequence to the low-order representation, then broaden to model extensions; the phenomenogram compresses the physical chronology into a final visual argument.
|
||||
- Cornejo Maceda et al. synthesize by increasing control-space complexity and then distil methodological consequences; their conclusion follows the hierarchy established by the results rather than listing sections.
|
||||
- Li & Zhang preserve the order physics/stability analysis → learned-control findings → coupled methodological implications.
|
||||
- Rabault et al. move from named achievement → physical strategy → practical constraint → broader prospective use.
|
||||
|
||||
**Observed pattern.** Final synthesis normally follows the inferential order already established in the body. It does not start with the most speculative physical reading and work backward to evidence. Where multiple layers exist, the closing section steps from capability or baseline result toward interpretation and only then toward general implications.
|
||||
|
||||
### 3.3 Limitations and future decisive tests
|
||||
|
||||
- All four papers place important limitations before or within the final synthesis rather than saving every caveat for the last sentence.
|
||||
- Deng et al. pair model success with missing-detail and fixed-mode limits, then name higher-dimensional or data-driven model extensions.
|
||||
- Cornejo Maceda et al. distribute limits by theme and place prospective algorithm/robustness directions at discussion transitions and the conclusion end.
|
||||
- Li & Zhang explicitly call parts of the design crude or situation-dependent and propose targeted changes to reward and probe-placement logic.
|
||||
- Rabault et al. identify regularization/control constraints before proposing higher-fidelity simulation, experiment and transfer-learning routes.
|
||||
|
||||
**Observed pattern.** Effective outlooks are attached to an identified unresolved link. The strongest examples say what design or evidence would test that link; weaker closing cadence drifts into broad invitations without an explicit decision criterion.
|
||||
|
||||
### 3.4 Claim ceiling and verbal strength
|
||||
|
||||
- The sample often uses confident local verbs for tested achievements while bounding scope by plant, parameter range, dimensionality or model fidelity.
|
||||
- Deng et al. combine a strong model characterization with explicit loss of detail in harder regimes.
|
||||
- Cornejo Maceda et al. make operational/methodological claims but stop short of presenting a universal turbulent-flow solution.
|
||||
- Li & Zhang qualify transfer and real-world relevance as prospective and retain tested-condition bounds.
|
||||
- Rabault et al. frame the contribution as a proof of concept and a route to a class of methods, not completed general applicability.
|
||||
|
||||
**Observed pattern.** Strong nouns such as robustness, stability, optimality or generality are credible only when a dedicated test supports them. A final section may broaden significance, but its grammatical scope must remain narrower than or equal to the actual test design.
|
||||
|
||||
### 3.5 Figure/table use in final synthesis
|
||||
|
||||
- Deng et al. use a late phenomenogram as an integrative visual that reorganizes the evidence rather than introducing a new result.
|
||||
- Cornejo Maceda et al. use late summary displays—a performance/search-space synthesis and a hierarchy schematic—as part of the closing argument.
|
||||
- Li & Zhang and Rabault et al. introduce no new final-section displays; they rely on earlier figures or textual synthesis.
|
||||
|
||||
**Observed pattern.** A closing display is optional, not conventional by default. It earns space when it performs genuine integration across a physical chronology or hierarchy. It fails when it merely repeats prior panels, suggests causal links not tested, or converts unlike metrics into a visual ranking.
|
||||
|
||||
### 3.6 Final bridge and closing cadence
|
||||
|
||||
- Deng et al. bridge from model-specific limitations to alternative state spaces and identification methods, ending prospectively.
|
||||
- Cornejo Maceda et al. move from the method hierarchy back to the fluidic pinball as a tractable benchmark for future development.
|
||||
- Li & Zhang move from local physics-informed simplification to broader coupling between fluid mechanics and learning control.
|
||||
- Rabault et al. move from proof-of-concept constraints to higher-fidelity simulation, experiment and complex actuation.
|
||||
|
||||
**Observed pattern.** The closing cadence typically has three beats: bounded result → unresolved boundary → prospective route or broader significance. The strongest ending returns to the paper's physical object or governing question; generic “more work is needed” language is less decisive.
|
||||
|
||||
## 4. Project recommendation
|
||||
|
||||
### 4.1 Separation and local function
|
||||
|
||||
Use two labelled units within the final manuscript section:
|
||||
|
||||
1. **Discussion** — four functional paragraphs that integrate evidence, classify non-equivalent support, state the envelope, and define decisive tests.
|
||||
2. **Conclusion** — three short paragraphs that answer the question once, state the ceiling once, and close on one claim-changing test plus the declared meaning of cloaking/illusion.
|
||||
|
||||
Trade-off: a combined “Conclusions and outlook” section would be shorter and has JFM precedent, but it would make this paper's several chains and estimands easier to collapse rhetorically. A separate Discussion provides room for boundary-aware synthesis; the short Conclusion prevents that room from becoming a second Results section.
|
||||
|
||||
### 4.2 Discussion paragraph jobs
|
||||
|
||||
1. **Re-establish the evidence ladder.** Begin with named capability under distinct task contracts, then narrow to the sole independently evaluated deep case. Import definitions only; do not replay values, acquisition details or panel narration. Job: establish that breadth and depth have different evidence ceilings.
|
||||
2. **Assemble the bounded interpretation.** Move from executable action-law evidence to mean/dynamic field organization in their declared order. Classify links explicitly as non-equivalent but compatible observations where appropriate. Job: show why a control-skeleton reading is supportable without calling it causal, unique or independently triangulated.
|
||||
3. **Define the envelope.** Group limitations by the inference they constrain: cross-task transfer, chain continuity, metric equivalence, theory closure and experimental status. Put the most dangerous qualification immediately after the affected synthesis sentence. Job: prevent the final Conclusion from carrying caveats that appeared nowhere earlier.
|
||||
4. **Name decisive tests.** Select only tests that would resolve an explicit open link: matched interventions/roles, independent parents where required, same-plant or same-chain comparisons, closed budgets, and synchronized calibrated measurements. State which ceiling each test could raise. Job: turn outlook into falsifiable research design rather than a wish list.
|
||||
|
||||
Local cadence for each Discussion paragraph:
|
||||
|
||||
`bounded synthesis → evidence relationship → strongest alternative/limit → consequence for the next inference`
|
||||
|
||||
### 4.3 Conclusion paragraph jobs
|
||||
|
||||
1. **Question and scope.** Restate the physical question and named scope without a method list. Answer first at capability level, then identify where stronger field evaluation exists. No citations, values or figure calls.
|
||||
2. **Strongest integrated answer.** State the bounded deep-case interpretation in evidence order. Use one concise reminder that the contributing analyses have distinct estimands and do not constitute causal or independent confirmation.
|
||||
3. **Ceiling and closing bridge.** State the strongest surviving boundary and one integrated testing programme capable of raising it. Close by returning from the test to cloaking/illusion as matching to declared references, not universal invisibility or transfer.
|
||||
|
||||
Recommended final cadence:
|
||||
|
||||
`what was established → exactly where it holds → what remains unresolved → what observation/intervention would decide it → return to the physical question`
|
||||
|
||||
Do not end on a package, algorithm name, generic application list, or unqualified promise.
|
||||
|
||||
### 4.4 Evidence density, tense, and transitions
|
||||
|
||||
- **Discussion:** one evidence relationship per paragraph; source details stay with owning result sections. Use present tense for the inference and boundary, past for work performed, and conditional/future for proposed tests.
|
||||
- **Conclusion:** no new scientific detail. Use present tense for the bounded answer and conditional/future once for the decisive test.
|
||||
- Bridge from experiment to Discussion by separating qualitative deployability from quantitative/causal evidence.
|
||||
- Bridge from Discussion to Conclusion only after the envelope and tests are explicit; the Conclusion should feel like compression, not repair.
|
||||
- The final sentence should close the conceptual loop, not announce an undefined research horizon.
|
||||
|
||||
## 5. Ownership, imports, exports, and duplication deletions
|
||||
|
||||
### Imports
|
||||
|
||||
- From Introduction: exact three-level claim ladder and declared meanings of hydrodynamic cloaking/illusion.
|
||||
- From contracts: chain, role, comparator, metric/estimand, ROI/window and Legacy/V5 distinctions, imported only as short boundary phrases.
|
||||
- From capability and cross-case owners: named task capability and the non-transfer envelope.
|
||||
- From Kármán performance: sole detailed owner of independent field-matching evidence; Discussion imports only its result-level meaning.
|
||||
- From SR: executable-law and bounded-deletion ceiling, without formula or term-by-term replay.
|
||||
- From field/CCD: mean/dynamic organization, different-estimand boundary and exact noncausal terminology, without mode or operator detail.
|
||||
- From steady/theory: contextual recurrence/sign/failure boundary, without cross-chain amplitude comparison.
|
||||
- From experiment: qualitative, open-loop and provenance-gated status only; apparatus/provenance details remain with that section.
|
||||
|
||||
### Owned exports
|
||||
|
||||
- One integrated answer in the fixed order: named capability → deep-case independent field evaluation → bounded action/field interpretation.
|
||||
- One integrated envelope covering non-transfer, non-causality, non-equivalence, theory incompleteness and qualitative experiment status.
|
||||
- A prioritized set of decisive tests, each tied to an unresolved inferential link.
|
||||
- A closing bridge back to the concept as declared-reference matching.
|
||||
|
||||
### Duplication deletions
|
||||
|
||||
- Delete detailed values, ROI/window definitions and metric formulas already owned by the performance/contracts sections.
|
||||
- Delete the SR formula, coefficient list, deletion ledger and CCD mode/operator details.
|
||||
- Keep the full near-collision technical comparison with its owning field/interface material; retain only the classification `different estimand` here.
|
||||
- Do not repeat the cross-case case-by-case inventory; summarize only the asymmetric envelope.
|
||||
- Do not repeat the experiment provenance checklist; retain only its qualitative/open-loop/provenance gate.
|
||||
- State the full three-level ladder once in the Introduction and restate it once here in compressed evidence order; remove intermediate package-list versions.
|
||||
|
||||
**Opening dependency:** the experiment section must already have separated deployability from validation and resolved whether any author-supplied material passed provenance gates.
|
||||
|
||||
**Closing bridge:** from bounded computational interpretation and qualitative deployment to the exact intervention/measurement programme required for a higher ceiling, then back to declared-reference matching.
|
||||
|
||||
**Terminology ownership:** this unit may synthesize `control skeleton` and `action-correlated correction-field modes`, but may not redefine them or introduce synonyms implying mechanism, response, authority, uniqueness or transfer.
|
||||
|
||||
## 6. Claim ceiling and near-collision audit
|
||||
|
||||
### Green
|
||||
|
||||
- The Conclusion may state the three fixed levels in compressed evidence order: named cross-task capability under distinct contracts; independent field matching only for the named Legacy Kármán deep case; and a bounded Legacy Kármán/cloak-only control-skeleton interpretation.
|
||||
- It may state that a tested persistent action component ranks above smaller tested corrections and that the residual field organization is action-correlated, provided detail remains with its owner.
|
||||
|
||||
### Amber
|
||||
|
||||
- `supports`, `is compatible with`, `organizes`, `accounts for the measured difference`, `provides context`, and `motivates a bounded control-skeleton interpretation` are allowed only with the relevant chain/estimand boundary adjacent.
|
||||
- Steady evidence may provide separate-chain context for recurrence/sign; it cannot become proof of the periodic interpretation.
|
||||
|
||||
### Prohibited red without new authority and dedicated tests
|
||||
|
||||
`causes`, `mechanism`, `necessary`, `sufficient`, `unique`, `optimal`, `stable`, `robust`, `general`, `universal`, `transferable`, `restores momentum`, `independently validates`, `triangulates`, `experimentally validates`, or any implication that Illusion SR is positive, OID carries a result, theory closes finite-Re physics, or qualitative media establish quantitative matching.
|
||||
|
||||
### Near-collision findings
|
||||
|
||||
| Pair | Acquisition/roles/chain/ROI-mask-weight-window finding | Classification | Discussion/Conclusion instruction |
|
||||
|---|---|---|---|
|
||||
| Standardized PPO performance mean and corrected four-role CCD mean | Distinct role acquisitions and role sets; different retained windows, masks and weighting/normalization contracts | `different estimand` | No arithmetic, agreement, replication, validation or triangulation wording; mention classification once |
|
||||
| SR action magnitude and V5 steady endpoint | Separate Legacy periodic and V5 steady chains, different action/metric contracts and evaluation semantics | `related diagnostic` | Context only; no amplitude agreement, continuation or same-law inference |
|
||||
| Periodic field L2 and DTW | Spatial registered-field and signal-alignment diagnostics with different objects and semantics | `related diagnostic` | L2 carries field matching; DTW remains complementary and case-specific |
|
||||
| Cross-case periodic, event-relative and steady evaluations | Different temporal coordinates, references, roles and windows | `different estimand` | No pooled breadth score or common efficacy ranking |
|
||||
|
||||
No pair above is licensed as `same estimand` by the controlling inputs.
|
||||
|
||||
## 7. Figure, table, and equation contracts
|
||||
|
||||
| Item | Owner and argumentative job | Source/roles and caption ceiling | Readiness | Fallback |
|
||||
|---|---|---|---|---|
|
||||
| Compact limitations table | Conclusion; align each claim level with chain/task, evidence type, strongest non-claim and decisive test | Text contracts imported from R3 owners; caption may say it summarizes boundaries, not new results or negative findings | `required`; text-ready, formatting/placement not production-ready; no underlying artifact audit implied | Integrate the same four fields into Discussion paragraph 3–4 if space forces omission |
|
||||
| Integrated evidence map | Conclusion; visualize evidence order and distinguish same-chain arrows from contextual/non-equivalent links | Only authority-bound labels; dashed/context links must not imply causality, closure, validation or pooled metrics | `optional`; text concept ready, panel not created or artifact-audited | Omit; use the Discussion sequence and limitations table |
|
||||
| Reuse of earlier figures | Earlier result owners; only if a brief backward reference materially anchors synthesis | Retain original source, roles, caption ceilings and readiness; Conclusion adds no reinterpretive panel labels | `optional`; depends on each owning panel's readiness | Refer generically to the evidence class or omit figure call |
|
||||
| New scientific equation | None; no argumentative need in final synthesis | Not applicable | `blocked` by section ownership; Conclusion introduces no equation | Use fixed terms in prose |
|
||||
| New numerical result figure | None; would constitute new/replayed evidence | Not applicable | `blocked` | No replacement needed |
|
||||
|
||||
A late summary display is recommended only if it reorganizes relationships that prose cannot show compactly. The safe default is the limitations table without an evidence map. Text readiness does not imply any panel or production readiness.
|
||||
|
||||
## 8. Author decisions, conflicts, deferrals, and blockers
|
||||
|
||||
### Evidence-governed decisions
|
||||
|
||||
- Separate Discussion and Conclusion because the paper must integrate several non-equivalent chains/estimands before compressing its answer.
|
||||
- Preserve evidence order rather than manuscript-section inventory order.
|
||||
- Put limitations beside their corresponding synthesis claims; do not use the Conclusion to repair prior overstatement.
|
||||
- Allow no new number, citation, equation, scientific result or panel narration in the Conclusion.
|
||||
- Tie every future test to the exact claim ceiling it could change.
|
||||
|
||||
### Author choice
|
||||
|
||||
- **Optional integrated evidence map.** Recommendation: omit unless a draft proves that dashed/non-equivalent links are immediately legible and the map replaces rather than duplicates prose. Consequence of inclusion: consumes space and creates a high risk of causal or pooled reading. Safe default: retain only the compact limitations table. Status remains `discuss` until the author/coordinator decides during display compression.
|
||||
|
||||
### Conflicts resolved
|
||||
|
||||
- The four-paper sample does not prescribe one universal section label: both split and combined architectures occur. Project complexity, not frequency in the sample, supports the split recommendation.
|
||||
- Late synthesis figures have JFM precedent, but two of four closest papers close without them. Therefore an evidence map is optional, not required by style.
|
||||
- Broad future-facing closing sentences occur in the sample, but this project's fixed ceiling requires narrower, claim-linked decisive tests.
|
||||
|
||||
### Deferrals and blockers
|
||||
|
||||
- No style blocker remains: four closest JFM PDFs were available and read at full-PDF level, and all controlling reports were read.
|
||||
- The optional evidence map is panel-production deferred and nonblocking.
|
||||
- The experiment's final author-supplied prose/media and provenance remain externally gated; this does not block the style contract but affects which experiment boundary the later final section may state.
|
||||
- Any retained statement that depends on hidden project artifacts remains authority-led and artifact-audit deferred; this memo does not raise or verify those claims.
|
||||
|
||||
## 9. Anti-patterns and validation checklist
|
||||
|
||||
### Anti-patterns
|
||||
|
||||
- Abstract replay or one bullet/paragraph per manuscript section.
|
||||
- Opening Discussion with the control-skeleton interpretation before capability and independent evaluation have been restated.
|
||||
- Treating compatible observations as independent proof or collapsing different estimands into “agreement”.
|
||||
- Introducing a limitation for the first time in the final Conclusion.
|
||||
- Converting every limitation into future work without identifying a decisive observable, intervention or decision criterion.
|
||||
- Replaying values, equations, formulae, ROI/window details, mode galleries or captions.
|
||||
- Using a summary figure as a causal graph, pooled score or metric hierarchy across incompatible contracts.
|
||||
- Ending with unsupported applications, generic “opens new avenues”, or universal claims.
|
||||
- Allowing qualitative experiment language to inherit computational evaluation or validation wording.
|
||||
- Renaming action-correlated correction-field modes for variety or using mechanism/response/control-authority synonyms.
|
||||
|
||||
### Validation checklist for the later section contract
|
||||
|
||||
- [ ] Discussion and Conclusion have distinct jobs and no duplicate paragraph.
|
||||
- [ ] Synthesis follows capability → independent evaluation → bounded interpretation → envelope → decisive tests.
|
||||
- [ ] Every imported inference remains within its owning section's ceiling.
|
||||
- [ ] Different-estimand and related-diagnostic pairs are classified, not blended.
|
||||
- [ ] The strongest local limitation is adjacent to the claim it constrains.
|
||||
- [ ] Future tests name the unresolved link and the ceiling they could raise.
|
||||
- [ ] Conclusion contains no new number, citation, equation, result, figure narration or artifact claim.
|
||||
- [ ] Any limitations table or map states owner, job, source, readiness, caption ceiling and fallback.
|
||||
- [ ] Text readiness remains separate from panel/artifact readiness.
|
||||
- [ ] The final sentence returns to the physical question under declared-reference matching.
|
||||
- [ ] A coordinator second audit occurs before `frozen`; this first-pass memo remains `researching`.
|
||||
|
||||
## 10. Validation result
|
||||
|
||||
**Style gate: PASS.** Exact identities, DOIs and read levels are recorded; four closest JFM full PDFs and the required repository controls were read; observed practice is separated from recommendation; the memo covers separation, evidence order, limitations, decisive tests, claim ceiling, displays, cadence, ownership, near-collisions, anti-patterns and coverage states. No manuscript prose, project result value, new scientific claim, new figure, equation, CFD/training/analysis, or artifact-readiness assertion was created.
|
||||
|
||||
**Round-4 disposition: `researching`.** This is a completed first-pass style memo awaiting coordinator use and second audit; it is not a frozen section contract.
|
||||
Reference in New Issue
Block a user