Files
MeltMind/references/2026-11-17/phase1-physics/physics_evidence_1.json
T

4 lines
12 KiB
JSON
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
{
"result": "[Boo23] Deep Learning‐Based Reduced‐Order Modeling of Darcy‐Flow Systems With Local Mass Conservation (2023)\nInternational Journal for Numerical Methods in Engineering\nDate: 2023-11-24\nBy W. Boon, N. R. Franco, A. Fumagalli, Paolo Zunino - 6 citations - 2.0 cit./yr\nDOI: 10.1002/nme.70188\nPDF ✓\n\nWe propose a new reduced‐order modeling (ROM) strategy for tackling parametrized Darcy‐flow systems in which the constraint is given by mass conservation. Our approach employs classical neural‐network architectures and supervised learning, but it is constructed in such a way that the resulting ROM is guaranteed to satisfy the linear constraints exactly. The procedure is based on a splitting of the solution into a particular solution satisfying the constraint and a homogenous solution. The homogeneous solution is approximated by mapping a suitable potential function, generated by a neural‐network model, onto the kernel of the constraint operator. For the particular solution, instead, we propose an efficient spanning‐tree algorithm. Starting from this paradigm, we present three approaches that follow this methodology, obtained by exploring different choices of the potential spaces: derived either via proper orthogonal decomposition (POD) or using the properties of a differential complex. To demonstrate the effectiveness of the proposed strategies and to emphasize their advantages over neural‐network regression approaches, we present a series of numerical experiments, ranging from mixed‐dimensional problems to nonlinear systems.\n\n[Zha26e] Structure-Preserving Neural Surrogates with Tractable Uncertainty Quantification (2026)\nArXiv\nDate: 2026-06-10\nBy Han-Di Zhang, Adrienne M. Propp, Brooks Kinch, H. Owhadi, Nathaniel Trask - 1 citations - 2.0 cit./yr\nDOI: 10.48550/arXiv.2606.11650\nPDF ✓\n\nRecent advances in scientific machine learning provide a means of near-real-time solution to partial differential equations (PDEs), but lack the theoretical underpinnings of conventional simulators that support contemporary verification and validation. In this work, we construct data-driven reduced-order models that serve as structure-preserving, real-time surrogates. Remarkably, the exterior calculus that imposes physical conservation structure also exposes topological structure that we use to build a Gaussian process (GP) representation of uncertainty in state-flux relationships, ultimately yielding a Dirichlet-to-Neumann map for quantities of interest with closed-form expressions for posterior uncertainty. We specifically propose structure-preserving $H(\\mathrm{div})$--$L^2$ subspaces of conventional Raviart--Thomas and $dgP_0$ elements prescribed by a lightweight transformer. Reduced-order dynamics consistent with this subspace are learned by posing a conservation law in which a GP describes the fluxes between volumes. This work hinges on a novel interface between mixed FEM spaces and GP regression; when training is posed as the optimal recovery problem (ORP), the resulting GP regression can be written as an optimization problem with equality constraints that impose a conservation structure, amenable to a fast Schur-complement training strategy. The trained model can then be solved in real time with closed-form estimators for boundary fluxes driven by prescribed Dirichlet data. The paper includes RKHS posterior error bounds for linear functionals to support uncertainty quantification, as well as numerical experiments demonstrating the accuracy of the posterior distribution as a surrogate for error estimation.\n\n[Pro25] Discovery of Probabilistic Dirichlet-to-Neumann Maps on Graphs (2025)\nArXiv\nDate: 2025-06-03\nBy Adrienne M. Propp, Jonas A. Actor, Elise Walker, H. Owhadi, Nathaniel Trask, D. Tartakovsky - 1 citations - 0.7 cit./yr\nDOI: 10.48550/arXiv.2506.02337\nPDF ✓\n\nDirichlet-to-Neumann maps enable the coupling of multiphysics simulations across computational subdomains by ensuring continuity of state variables and fluxes at artificial interfaces. We present a novel method for learning Dirichlet-to-Neumann maps on graphs using Gaussian processes, specifically for problems where the data obey a conservation constraint from an underlying partial differential equation. Our approach combines discrete exterior calculus and nonlinear optimal recovery to infer relationships between vertex and edge values. This framework yields data-driven predictions with uncertainty quantification across the entire graph, even when observations are limited to a subset of vertices and edges. By optimizing over the reproducing kernel Hilbert space norm while applying a maximum likelihood estimation penalty on kernel complexity, our method ensures that the resulting surrogate strictly enforces conservation laws without overfitting. We demonstrate our method on two representative applications: subsurface fracture networks and arterial blood flow. Our results show that the method maintains high accuracy and well-calibrated uncertainty estimates even under severe data scarcity, highlighting its potential for scientific applications where limited data and reliable uncertainty quantification are critical.\n\n[Kie98] A free-front tracking algorithm for a control-volume-based Hele–Shaw method (1998)\nInternational Journal for Numerical Methods in Engineering, pp. 253-269\nDate: 1998-01-30\nBy C. Kietzmann, J. P. van der Walt, Y. Morsi - 14 citations - 0.5 cit./yr\nDOI: 10.1002/(SICI)1097-0207(19980130)41:2<253::AID-NME282>3.0.CO;2-7\nPDF X\n\nWith free surface flow problems the position of the free front at a particular time step needs to be predicted. When simulating the flow of hot molten polymer in injection moulds this position is essential for the prediction of the position of the weld lines on the final product. Weld lines are important in that they indicate positions of diminished strength and spoil the aesthetics of the product. A code using a finite volume approach on the thin gap Hele–Shaw flow equations to simulate the filling stage of the injection moulding process has been developed which requires a suitable algorithm for the free-front tracking. The Volume of Fluid (VOF) method was implemented in this study for the tracking of the free front. The discretization and various formulations of the VOF equation are presented as well as the results obtained. The impact on the existing equations is also investigated as well as the development of a packing routine that is used to eliminate the diffusive nature of the VOF equation. © 1998 John Wiley & Sons, Ltd.\n\n[Wan15] Fast Mold Filling Simulation Based on the GeodesicDistance Calculation Algorithm for Liquid CompositeMolding Processes (2015)\nCmes-computer Modeling in Engineering & Sciences, pp. 59-79\nDate: 2015-07-01\nBy J. Wang, P. Šimáček, S. Advani - 3 citations - 0.3 cit./yr\nDOI: 10.3970/CMES.2015.107.059\nPDF X\n\nIn Liquid Composite Molding (LCM) processes, resin is introduced into a stationary fiber reinforcement placed in the mold, until the reinforcement gets fully saturated with resin and all volatiles are vented out of the part. Finite element based software packages have been developed to simulate the mold filling process and eliminate expensive and tedious trial and error practices to arrive at a successful mold filling without any voids. However, the non-homogeneity of the fiber reinforcement material and its placement and layup in the mold creates a large degree of variability of flow patterns during the resin impregnation process. Executing simulations for every possible permutation of flow scenarios, which is required to devise a robust process design is computationally expensive. Therefore, it is necessary to find faster approximate mold filling simulation methods so that all simulations can be performed within a reasonable time frame. In this paper, a discretized one-dimensional flow model is developed to predict the fill time based on the distance resin travels. Combined with Dijkstra’s algorithm, this model is then implemented on spatial surface meshes to calculate fill time for each node and generate flow development pattern. The computational model developed can predict the mold filling pattern for complex parts even with variable permeability or thickness of the fiber preform, and can capture the disturbed flow behavior along any difficult geometric features at a fraction of the computational cost. Case studies are presented to demonstrate the efficiency and accuracy of the distance-based model.\n\n[Kim07] Homogenization of a Hele-Shaw type problem in periodic and random media (2007)\nUnknown Journal\nBy Inwon C. Kim, A. Mellet - 5 citations - 0.3 cit./yr\nNo DOI found. Link: https://www.semanticscholar.org/paper/d54c05f284ef7491fb2e780e4213071d2641395a\nPDF X\n\nWe investigate the homogenization limit of a free boundary problem with space-dependent free boundary velocities. The problem under consideration has a well-known obstacle problem transformation, formally obtained by integrating with respect to the time variable. By making rigorous the link between these two problems, we are able to derive an explicit formula for the homogenized free boundary velocity, and we establish the uniform convergence of the free boundaries.\n\n[Tra14] An L1 Penalty Method for General Obstacle Problems (2014)\nSIAM J. Appl. Math., pp. 1424-1444\nDate: 2014-04-04\nBy Giang Tran, Hayden Schaeffer, W. M. Feldman, S. Osher - 46 citations - 3.6 cit./yr\nDOI: 10.1137/140963303\nPDF ✓\n\nWe construct an efficient numerical scheme for solving obstacle problems in divergence form. The numerical method is based on a reformulation of the obstacle in terms of an $L^1$-like penalty on the variational problem. The reformulation is an exact regularizer in the sense that for a large (but finite) penalty parameter, we recover the exact solution. Our formulation is applied to classical elliptic obstacle problems as well as some related free boundary problems, for example, the two-phase membrane problem and the Hele--Shaw model. One advantage of the proposed method is that the free boundary inherent in the obstacle problem arises naturally in our energy minimization without any need for problem specific or complicated discretization. In addition, our scheme also works for nonlinear variational inequalities arising from convex minimization problems.\n\n[Dei13] Quasi-Variational Inequalities for Source-Expanding Hele-Shaw Problems (2013)\nUnknown Journal\nBy John DeIonno\nNo DOI found. Link: https://www.semanticscholar.org/paper/d93b975345b19646040679dfaa11ccb55fb569dd\nPDF X\n\nWe study nonlinear partial differential equations describing Hele-Shaw flows for which the source of the fluid, which classically is at a single point, instead expands as the moving boundary uncovers new sources. This problem is reformulated as a type of quasi- variational inequality, for which we show there exists a unique weak solution. Our procedure utilizes a Baiocchi integral transform. However, unlike in traditional applications, the inequality solved by the transformed variable continues to depend on the free boundary, and thus standard theory of variational inequalities cannot be immediately applied. Instead, we develop convergent time-stepping scheme. As a further application, we generalize and study a related problem with a growing density function obeying a hard constraint on the maximum density.\n\nPDF symbols - ✓ PDF available; X No PDF available; … PDF download attempt in progress\n\nWhen using cite keys in your reply to the user, format as [[cite_key]](url), where url is https://doi.org/<DOI> for the DOI shown above (or the exact Link shown for papers without one), separating multiple with commas. In workspace files, keep bare [cite_key] markers with no URLs."
}