# Computational Fluid Dynamics (CFD)

Area of study that uses high-speed computers and numerical methods to solve approximations to fluid-flow equations.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## Measuring and Improving Transfer of Neural Surrogates for Vehicle Aerodynamics

DevFeed: [Measuring and Improving Transfer of Neural Surrogates for Vehicle Aerodynamics](<https://devfeed.tech/articles/measuring-and-improving-transfer-of-neural-surrogates-for-vehicle-aerodynamics-143571.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24843>)

Author: Favour Nerrise, Julian Busch, Gurpreet Singh Hora, Martin Sipka, Danielle C. Maddix, Georgia Olympia Brikis

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [few-shot](<https://devfeed.tech/topics/few-shot.md>), [LLM evaluation / benchmarking](<https://devfeed.tech/topics/llm-evaluation-benchmarking.md>)

Tags: [benchmarking](<https://devfeed.tech/tags/benchmarking.md>), [cfd](<https://devfeed.tech/tags/cfd.md>), [computational-fluid-dynamics](<https://devfeed.tech/tags/computational-fluid-dynamics.md>), [computational-fluid-dynamics-cfd](<https://devfeed.tech/tags/computational-fluid-dynamics-cfd.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [domain](<https://devfeed.tech/tags/domain.md>), [error](<https://devfeed.tech/tags/error.md>), [external](<https://devfeed.tech/tags/external.md>), [fast](<https://devfeed.tech/tags/fast.md>), [fine-tuning](<https://devfeed.tech/tags/fine-tuning.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [independent](<https://devfeed.tech/tags/independent.md>), [model](<https://devfeed.tech/tags/model.md>), [models](<https://devfeed.tech/tags/models.md>), [neural](<https://devfeed.tech/tags/neural.md>), [physics](<https://devfeed.tech/tags/physics.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [pretraining](<https://devfeed.tech/tags/pretraining.md>), [results](<https://devfeed.tech/tags/results.md>), [risk](<https://devfeed.tech/tags/risk.md>), [vehicle](<https://devfeed.tech/tags/vehicle.md>)

### AI overview

Across four target datasets that vary in geometry, CFD solver, turbulence model, and flow conditions, fine-tuning five DrivAerML-pretrained vehicle-aerodynamics surrogates beat training from scratch in fewer than half the comparisons and reduced accuracy on the source domain. The authors introduce GeoReplay, which retains source simulations geometrically similar to the target and applies their updates only when aligned with the target gradient. It matches fine-tuning's target accuracy while reducing its source error in 77 of 80 settings. The study is based on these benchmark settings; the abstract does not report broader validation.

### Source excerpt

arXiv:2611.24843v1 Announce Type: new Abstract: Neural surrogates for external vehicle aerodynamics are fast alternatives to physics-based computational fluid dynamics (CFD) simulations. Their foundational limitation is a lack of zero-shot generalization across vehicle classes, which results in an expensive, custom training campaign for each new vehicle design. Similar to pretrained models in other fields, a pretrained surrogate should adapt to a new vehicle class using only a few simulations and stay accurate on its pretraining domain. No prior study has systematically tested this transfer behavior. We conduct a large-scale, controlled benchmarking study on five DrivAerML-pretrained surrogates across four independent target datasets. These datasets differ from the source in geometry, CFD solver, turbulence model, and flow conditions. We find that fine-tuning these surrogates outperforms training from scratch in fewer than half of the comparisons and lowers the accuracy of the surrogate on its source dataset. We introduce GeoReplay, a geometry- and gradient-aligned source replay that keeps source simulations resembling the target in training and applies their updates only when they agree with the target gradient. A risk bound separates geometric coverage from differences in the simulated physics, and we prove that every sufficiently small GeoReplay step decreases the target training objective. GeoReplay lowers the source error of fine-tuning in 77 of 80 settings and matches its target accuracy.

## Physically Admissible Generative Flow Matching for Sparse Flow Reconstruction

DevFeed: [Physically Admissible Generative Flow Matching for Sparse Flow Reconstruction](<https://devfeed.tech/articles/physically-admissible-generative-flow-matching-for-sparse-flow-reconstruction-141786.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22549>)

Author: Yuhao Wang, Yuanxi Peng

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [generative-ai-tools](<https://devfeed.tech/topics/generative-ai-tools.md>), [concept-learning](<https://devfeed.tech/topics/concept-learning.md>)

Tags: [boundary](<https://devfeed.tech/tags/boundary.md>), [bounded](<https://devfeed.tech/tags/bounded.md>), [consistency](<https://devfeed.tech/tags/consistency.md>), [continuous](<https://devfeed.tech/tags/continuous.md>), [coverage](<https://devfeed.tech/tags/coverage.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [energy](<https://devfeed.tech/tags/energy.md>), [error](<https://devfeed.tech/tags/error.md>), [fields](<https://devfeed.tech/tags/fields.md>), [flow](<https://devfeed.tech/tags/flow.md>), [flow-matching](<https://devfeed.tech/tags/flow-matching.md>), [fluid](<https://devfeed.tech/tags/fluid.md>), [generative](<https://devfeed.tech/tags/generative.md>), [generative-models](<https://devfeed.tech/tags/generative-models.md>), [global](<https://devfeed.tech/tags/global.md>), [intermediate](<https://devfeed.tech/tags/intermediate.md>), [invariant](<https://devfeed.tech/tags/invariant.md>), [laws](<https://devfeed.tech/tags/laws.md>), [loops](<https://devfeed.tech/tags/loops.md>), [matrix](<https://devfeed.tech/tags/matrix.md>), [metric](<https://devfeed.tech/tags/metric.md>)

### AI overview

Feasible Subspace Consistency (FSC) keeps the full generative ODE trajectory within the discrete conservation manifold: tangent-space matching enforces physical admissibility, while mass-orthonormal coordinates preserve the kinetic-energy metric. Sparse reconstruction uses amortized posterior transport in a frozen FSC prior's latent space, so samples inherit incompressibility and boundary conditions without per-sample projection. The authors evaluate exact spectral projection on periodic Kolmogorov flow and matrix-free iterative projection on a wall-bounded cylinder wake with immersed obstacles. They report that FSC avoids the mode-coverage collapse and high-wavenumber distortion associated with terminal projection, raising mode coverage from 0.000 to 0.790. Against the unconditioned prior, amortized FSC reduces global relative error by 50% and 46%, and sensor residuals by 63% and 55%, while providing calibrated, spatially heterogeneous uncertainty.

### Source excerpt

arXiv:2611.22549v1 Announce Type: new Abstract: Reconstructing full fluid velocity fields from sparse, localized observations is a severely ill-posed inverse problem: most of the spatial domain remains unobserved, discrete conservation laws must hold strictly, and predictive uncertainty must be reliable. Existing generative models typically enforce physical constraints only at the terminal output or via soft penalties, leaving intermediate generative trajectories unconstrained. We introduce Feasible Subspace Consistency (FSC), a framework that preserves physical admissibility as an invariant of continuous generative dynamics: tangent-space matching confines the entire ODE trajectory to the discrete conservation manifold, while mass-orthonormal coordinates preserve the kinetic-energy metric. Sparse reconstruction then reduces to amortized posterior transport in the latent space of a frozen FSC prior, ensuring every posterior sample inherits incompressibility and boundary conditions without per-sample projection loops. We validate FSC on a periodic Kolmogorov flow, where exact spectral projection isolates path tangency and metric isometry, and a wall-bounded cylinder wake with immersed obstacles, realized via matrix-free iterative projection. FSC prevents the mode-coverage collapse induced by terminal projection (increasing mode coverage from $0.000$ to $0.790$) while eliminating high-wavenumber spectral distortion. In sparse reconstruction, amortized FSC reduces global relative error by $50\%$ and $46\%$, and sensor residuals by $63\%$ and $55\%$, relative to the unconditioned prior, while delivering calibrated, spatially heterogeneous uncertainty.

## Where the Error Lives: Separating Lid-Edge and Body-Induced Error in Neural Pressure Prediction for 3D Cavity Flow

DevFeed: [Where the Error Lives: Separating Lid-Edge and Body-Induced Error in Neural Pressure Prediction for 3D Cavity Flow](<https://devfeed.tech/articles/where-the-error-lives-separating-lid-edge-and-body-induced-error-in-neural-pressure-prediction-for-3d-cavity-flow-138922.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19741>)

Author: Zirun Zhou, Zhi-Qian Zhang, Yixuan Hong, Guang Liu, Zhigang Liu, Lili Zhang

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [3D](<https://devfeed.tech/topics/3d.md>)

Tags: [3d](<https://devfeed.tech/tags/3d.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [boundary](<https://devfeed.tech/tags/boundary.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [error](<https://devfeed.tech/tags/error.md>), [errors](<https://devfeed.tech/tags/errors.md>), [fluid-dynamics](<https://devfeed.tech/tags/fluid-dynamics.md>), [grid](<https://devfeed.tech/tags/grid.md>), [neural](<https://devfeed.tech/tags/neural.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [prediction](<https://devfeed.tech/tags/prediction.md>)

### AI overview

In this 3D lid-driven cavity benchmark, a body-free, Reynolds-conditioned reference correction reduced mean full-volume relative L₂ error from 20.6% to 7.5% across 60 cases with 12 unseen geometries, but did little on average to improve near-body pressure prediction. The lid-edge junction occupied 1.5% of the fluid while accounting for 69% of truth pressure energy and 95% of the uncorrected model's squared error, so the global gain mainly reflects correction of a shared boundary field. Interior error changed by only +0.16% relative, and reference-relative body-response error remained about 25% in the interior near-body band. Grid-refinement, lid-regularization, and convergence checks delimit the numerical scope. The authors recommend reporting regional truth-energy and model-error shares alongside near-body and reference-relative body-response errors.

### Source excerpt

arXiv:2611.19741v1 Announce Type: cross Abstract: Can a pressure surrogate score much better globally without better resolving the pressure around immersed bodies? We test this in a three-dimensional lid-driven cavity with an unregularized lid and centered pressure at native $128^3$ resolution. On 60 cases from 12 unseen geometries, with predictions sealed before the CFD truth existed, adding a body-free, Reynolds-conditioned reference correction to the same frozen $64^3$ surrogate lifted to $128^3$ lowers the mean full-volume relative $L_2$ error from 20.6% to 7.5%. In the post hoc regional analysis, the lid-edge junction occupies 1.5% of the fluid but holds 69% of the truth pressure energy and 95% of the uncorrected surrogate's squared error. Interior error changes by just +0.16% relative, and object-local peak error and recall change little on average. Retrospective ablations locate the learned term's contribution in the interior and near the bodies; rebasing on the reference lowers global error further with little change in cohort-mean near-body metrics. Reference-relative body-response error remains about 25% in the interior near-body band. Grid-refinement, lid-regularization and convergence checks delimit the numerical scope. In this benchmark, a large global gain chiefly reflects correction of a shared boundary field, not better resolution of the body response. We recommend reporting regional truth-energy and model-error shares alongside near-body and reference-relative body-response errors.

## Learning Autoregressive Neural Operators on Solver-Defined Physical Constraint Manifolds

DevFeed: [Learning Autoregressive Neural Operators on Solver-Defined Physical Constraint Manifolds](<https://devfeed.tech/articles/learning-autoregressive-neural-operators-on-solver-defined-physical-constraint-manifolds-137492.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19753>)

Author: Nannan Zhou, Guanmeng Xian, Ning Yang

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Physics-guided deep learning](<https://devfeed.tech/topics/physics-guided-deep-learning.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>)

Tags: [autoregressive](<https://devfeed.tech/tags/autoregressive.md>), [boundary](<https://devfeed.tech/tags/boundary.md>), [constraint](<https://devfeed.tech/tags/constraint.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [flow](<https://devfeed.tech/tags/flow.md>), [learning](<https://devfeed.tech/tags/learning.md>), [neural](<https://devfeed.tech/tags/neural.md>), [operator-learning](<https://devfeed.tech/tags/operator-learning.md>), [operators](<https://devfeed.tech/tags/operators.md>), [pcm](<https://devfeed.tech/tags/pcm.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [projection](<https://devfeed.tech/tags/projection.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>)

### AI overview

PCM-NO keeps autoregressive incompressible-flow predictions close to solver-defined feasible states by projecting each predicted increment onto admissible directions and retracting the advanced state onto the prescribed constraint set. With exact affine constraint solves, transitions remain feasible; with approximate solves, per-step retraction resets feasibility to the residual achieved by that solve, while tangent-only residuals can accumulate under an upper bound. The abstract reports experiments on periodic Navier-Stokes and lid-driven cavity flows, where PCM-NO better preserved solver-level constraints over long rollouts and under parameter, boundary-value, and resolution shifts, with lower rollout errors and better flow-specific diagnostics than the compared approaches. The supplied text is an abstract, so implementation details and comparison baselines are not available.

### Source excerpt

arXiv:2611.19753v1 Announce Type: cross Abstract: Autoregressive neural operators approximate time-dependent PDE solvers by repeatedly composing a learned transition, although the resulting states need not remain feasible under the discrete constraints of the numerical problem. We study this issue for incompressible flow, where each velocity state must satisfy solver-level incompressibility together with prescribed boundary values. We formulate autoregressive operator learning as a map between solver-defined affine feasible sets and introduce PCM-NO, a projected-residual realization of this feasible-transition principle. The neural backbone predicts a candidate state increment, a fixed tangent projection extracts its admissible component, and a metric retraction maps the advanced state to the prescribed feasible set. With exact affine constraint solves, the resulting transition is closed on the feasible set, while an affine identity separates admissible-update parameterization from state-level closure. With approximate solves, tangent-only residuals admit a cumulative upper bound, whereas per-step retraction resets feasibility to the residual achieved by the current solve. Experiments on periodic Navier--Stokes flow and lid-driven cavity flow demonstrate the superiority of PCM-NO in preserving solver-level constraints over long autoregressive rollouts and under parameter, boundary-value, and resolution shifts, while achieving lower rollout errors and better flow-specific diagnostics across both in-distribution and stress settings.

## Sample-Efficient Control of Magnetoconvective Flows via Differentiable Simulation

DevFeed: [Sample-Efficient Control of Magnetoconvective Flows via Differentiable Simulation](<https://devfeed.tech/articles/sample-efficient-control-of-magnetoconvective-flows-via-differentiable-simulation-135856.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18008>)

Author: Jannis Becktepe, Dmitry Krasnov, Nils Thuerey, J\\"org Schumacher, Sebastian Peitz

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>)

Tags: [3](<https://devfeed.tech/tags/3.md>), [abstract](<https://devfeed.tech/tags/abstract.md>), [accelerated](<https://devfeed.tech/tags/accelerated.md>), [advanced](<https://devfeed.tech/tags/advanced.md>), [afc](<https://devfeed.tech/tags/afc.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [allows](<https://devfeed.tech/tags/allows.md>), [application](<https://devfeed.tech/tags/application.md>), [applications](<https://devfeed.tech/tags/applications.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [based](<https://devfeed.tech/tags/based.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [black](<https://devfeed.tech/tags/black.md>), [black-box](<https://devfeed.tech/tags/black-box.md>), [box](<https://devfeed.tech/tags/box.md>), [central](<https://devfeed.tech/tags/central.md>), [control](<https://devfeed.tech/tags/control.md>), [cooling](<https://devfeed.tech/tags/cooling.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [sample](<https://devfeed.tech/tags/sample.md>), [simulation](<https://devfeed.tech/tags/simulation.md>)

### AI overview

The paper presents φ-PICT, a differentiable, GPU-accelerated magnetohydrodynamic solver for active flow control, validated against MHD benchmarks. Integrated into FluidGym, it enables sample-efficient gradient-based methods that the authors report improve heat transfer by more than 25% and reach 10³ times higher accuracy in inverse design than black-box methods. The supplied abstract does not specify benchmark details or the conditions behind these results.

### Source excerpt

arXiv:2611.18008v1 Announce Type: new Abstract: Magnetohydrodynamic (MHD) fluid flows are central to demanding engineering applications, most prominently fusion blanket cooling, where flow control is a promising tool to optimize key properties such as heat transfer. While active flow control (AFC) has recently advanced through machine learning, its application to MHD flows remains largely unexplored. In this work, we present $\phi$-PICT, a novel differentiable GPU-accelerated MHD solver targeted at AFC. We validate $\phi$-PICT against MHD benchmarks and demonstrate its high computational efficiency. Integrated into FluidGym, $\phi$-PICT enables sample-efficient gradient-based algorithms to improve heat transfer by $>25$% and allows optimizers to achieve $10^3\times$ higher accuracy in inverse design than black-box methods. $\phi$-PICT is publicly available at https://github.com/upb-sciml/phiPICT.

## Wavelet Flow Matching for Multi-Scale Physics Emulation

DevFeed: [Wavelet Flow Matching for Multi-Scale Physics Emulation](<https://devfeed.tech/articles/wavelet-flow-matching-for-multi-scale-physics-emulation-134984.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.16573>)

Author: Gabriele Accarino, Juan Nathaniel, Carla Roesch, Pierre Gentine, Sara Shamekh, Duncan Watson-Parris, Viviana Acquaviva

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [emulation](<https://devfeed.tech/tags/emulation.md>), [flow-matching](<https://devfeed.tech/tags/flow-matching.md>), [fluid-dynamics](<https://devfeed.tech/tags/fluid-dynamics.md>), [generative](<https://devfeed.tech/tags/generative.md>), [optimal-transport](<https://devfeed.tech/tags/optimal-transport.md>), [physics](<https://devfeed.tech/tags/physics.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [scale](<https://devfeed.tech/tags/scale.md>)

### AI overview

Wavelet Flow Matching (WFM) emulates multi-scale physical dynamics by applying optimal-transport flow matching directly to a prescribed, lossless wavelet representation. A U-Net jointly predicts transport velocities across scales, avoiding learned latent compression and a separately pretrained autoencoder. The abstract reports that, on three chaotic fluid dynamics systems, WFM achieves superior spectral coherence and competitive or better long-horizon accuracy than state-of-the-art models. These findings are limited to the three systems described in the abstract.

### Source excerpt

arXiv:2605.16573v2 Announce Type: replace-cross Abstract: Accurate emulation of multi-scale physical systems governed by PDEs demands models that remain stable over long autoregressive rollouts while preserving fine-scale structures. Deterministic emulators produce overly smoothed predictions, while generative approaches better capture details but are costly. Latent-space generative models have emerged as a compromise but with the additional cost of separately pre-trained autoencoders. We propose Wavelet Flow Matching (WFM), a generative emulator that overcomes current trade-offs between cost and skill by performing optimal-transport flow matching directly in the multi-scale wavelet space. Rather than learning a latent compression, WFM leverages the hierarchical structure of a U-Net to jointly predict the transport velocities of a prescribed, lossless wavelet representation. On three challenging systems of chaotic fluid dynamics, WFM achieves superior spectral coherence and competitive or better long-horizon accuracy compared to state-of-the-art models. Our results position the wavelet space as an effective training-free representation for generative emulation of complex physical dynamics.

## Learning Nonlinear Transient Dynamics in Fluid Flow Networks with the Neural Basis Method

DevFeed: [Learning Nonlinear Transient Dynamics in Fluid Flow Networks with the Neural Basis Method](<https://devfeed.tech/articles/learning-nonlinear-transient-dynamics-in-fluid-flow-networks-with-the-neural-basis-method-132879.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.14534>)

Author: Aijaz Nazir, Yuhe Wang, S. K. Rasel, Min Wang

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>)

Tags: [boundary](<https://devfeed.tech/tags/boundary.md>), [cs-ce](<https://devfeed.tech/tags/cs-ce.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [domain-decomposition](<https://devfeed.tech/tags/domain-decomposition.md>), [edge](<https://devfeed.tech/tags/edge.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [error](<https://devfeed.tech/tags/error.md>), [exploration](<https://devfeed.tech/tags/exploration.md>), [fixed](<https://devfeed.tech/tags/fixed.md>), [flow](<https://devfeed.tech/tags/flow.md>), [fluid](<https://devfeed.tech/tags/fluid.md>), [flux](<https://devfeed.tech/tags/flux.md>), [framework](<https://devfeed.tech/tags/framework.md>), [laws](<https://devfeed.tech/tags/laws.md>), [learning](<https://devfeed.tech/tags/learning.md>), [map](<https://devfeed.tech/tags/map.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [monitoring](<https://devfeed.tech/tags/monitoring.md>), [network](<https://devfeed.tech/tags/network.md>), [networks](<https://devfeed.tech/tags/networks.md>), [neural](<https://devfeed.tech/tags/neural.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [transient-dynamics](<https://devfeed.tech/tags/transient-dynamics.md>)

### AI overview

The neural basis method combines fixed neural bases on each network edge with projection of discretized equations to determine expansion coefficients, preserving a conservative formulation. Non-overlapping domain decomposition couples edge subdomains, and a neural operator learns the parametric map to those coefficients. On looped and branching networks, reported relative L² errors are below 0.02%; the physics-informed neural network baseline fails. For unseen boundary conditions, including interpolation and extrapolation, mean pressure errors stay below 1% and mean flux errors below 3.3%, compared with 8-18% pressure and 85-95% flux errors for the physics-informed DeepONet baseline. The results suggest potential for rapid repeated evaluation in monitoring, design exploration and uncertainty quantification.

### Source excerpt

arXiv:2611.14534v1 Announce Type: cross Abstract: Fluid flow networks, from pipeline infrastructure to biological vasculature, are governed by nonlinear edgewise hyperbolic laws coupled through continuity and conservation at junctions. Many-query tasks require rapid evaluation across unseen parametric settings, but conventional methods re-solve the system for each query at prohibitive cost. Existing physics-informed neural approaches impose the governing equations through penalty losses over generic neural approximators; these losses act only as optimization surrogates, with no guarantee that their reduction reduces solution error, and the resulting optimization is prone to ill-conditioning, particularly for hyperbolic systems. Building on the "representation-then-projection" principle of the neural basis method, we introduce a new approach for modeling fluid flow networks. Fixed neural bases define the representation space on each edge, while projection of the discretized equations determines the expansion coefficients, embedding edgewise laws into a conservative formulation. Edge subdomains are coupled through non-overlapping domain decomposition, and a neural operator learns, directly in this reduced space, the parametric map to the expansion coefficients. On looped and branching network topologies, we reproduce references with relative $L^2$ errors below 0.02%, whereas the physics-informed neural network baseline fails. Across unseen boundary conditions in both interpolation and extrapolation, we maintain mean pressure errors below 1% and mean flux errors below 3.3%, while the physics-informed DeepONet baseline reaches 8-18% pressure errors and 85-95% flux errors. These results highlight the framework's potential for real-time monitoring, design exploration, and uncertainty quantification of fluid flow networks.

## EasyFoam: Skill-Oriented OpenFOAM Agents with Evidence-Driven Evolution

DevFeed: [EasyFoam: Skill-Oriented OpenFOAM Agents with Evidence-Driven Evolution](<https://devfeed.tech/articles/easyfoam-skill-oriented-openfoam-agents-with-evidence-driven-evolution-132016.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.14525>)

Author: Peng Yin, Zhiwei Chen, Tianyi Liu, Chen Li, Kai Li, Jian Cheng

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [Agent Skills](<https://devfeed.tech/topics/agent-skills.md>), [knowledge-engineering](<https://devfeed.tech/topics/knowledge-engineering.md>), [Loop Engineering](<https://devfeed.tech/topics/loop-engineering.md>)

Tags: [agent-framework](<https://devfeed.tech/tags/agent-framework.md>), [agents](<https://devfeed.tech/tags/agents.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [cfd](<https://devfeed.tech/tags/cfd.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ce](<https://devfeed.tech/tags/cs-ce.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [evolution](<https://devfeed.tech/tags/evolution.md>), [execution](<https://devfeed.tech/tags/execution.md>), [framework](<https://devfeed.tech/tags/framework.md>), [requirements](<https://devfeed.tech/tags/requirements.md>), [reuse](<https://devfeed.tech/tags/reuse.md>), [skill](<https://devfeed.tech/tags/skill.md>), [workflow](<https://devfeed.tech/tags/workflow.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

The abstract identifies a gap in OpenFOAM agent evaluation: simulations may run successfully yet fail numerical or quantity-of-interest criteria. It introduces ExtFoamBench, which varies workflows across five axes and assesses execution completion, numerical soundness, and QoI satisfaction separately. EasyFoam externalizes CFD expertise into reusable skills. Its execution loop builds workflows from task requirements and uses structured simulation evidence to diagnose failures and guide repairs; its evolution loop turns task experience into skill updates chosen through paired evaluation. The abstract reports substantially higher task acceptance rates than baseline agents, without giving figures or further evaluation details.

### Source excerpt

arXiv:2611.14525v1 Announce Type: cross Abstract: Large language model (LLM) agents are increasingly capable of automating end-to-end OpenFOAM workflows for computational fluid dynamics (CFD). However, current progress is bottlenecked by two critical limitations. First, existing evaluations are dominated by tutorial-level cases and execution-oriented metrics, masking a substantial executability--acceptability gap: successfully executed simulations often fail to satisfy prescribed numerical or quantity-of-interest (QoI) criteria. Second, domain expertise in existing systems is tightly coupled with bespoke agent architectures, limiting its reuse and continual refinement across agent implementations. To address these challenges, we introduce ExtFoamBench, a challenging benchmark for OpenFOAM agents that spans five axes of workflow variation, and separately assesses execution completion, numerical soundness, and QoI satisfaction. Complementing this, we propose EasyFoam, an agent framework that decouples CFD expertise into modular, externalized skills driven by a dual-loop mechanism. An execution loop constructs workflows from explicit task requirements and uses structured simulation evidence to diagnose failures and guide targeted repairs. Meanwhile, an evolution loop distills task experience into reusable skill updates selected through paired evaluation. Empirical evaluations demonstrate that EasyFoam achieves substantially higher task acceptance rates than baseline agents, establishing this work as a starting point for future research toward self-improving CFD agents.

## AeroZoo: A Benchmark of Well-Posed Flow Problems for Foundation Models of Aerodynamics

DevFeed: [AeroZoo: A Benchmark of Well-Posed Flow Problems for Foundation Models of Aerodynamics](<https://devfeed.tech/articles/aerozoo-a-benchmark-of-well-posed-flow-problems-for-foundation-models-of-aerodynamics-133184.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17152>)

Author: Peter Sharpe, Carmelo Gonzales, Corey Adams, Mehdi Ataei, Sanjay Choudhry

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [AI Foundation Models](<https://devfeed.tech/topics/ai-foundation-models.md>), [Benchmark](<https://devfeed.tech/topics/benchmark.md>), [Model Development](<https://devfeed.tech/topics/model-development.md>)

Tags: [benchmark](<https://devfeed.tech/tags/benchmark.md>), [cfd](<https://devfeed.tech/tags/cfd.md>), [code](<https://devfeed.tech/tags/code.md>), [cs-ce](<https://devfeed.tech/tags/cs-ce.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [flow](<https://devfeed.tech/tags/flow.md>), [foundation-models](<https://devfeed.tech/tags/foundation-models.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [rans](<https://devfeed.tech/tags/rans.md>)

### AI overview

AeroZoo introduces a benchmark and training corpus for testing aerodynamic models on flow problems that vary across shapes, enclosures, boundary conditions and Reynolds numbers. It contains 3,815 time-mean RANS solutions; 2,816 with stable means form the benchmark, with two tasks and 22 geometry-disjoint splits. On a random split, boundary-condition tokens reduce GeoTransolver's median velocity error to 8.4% of inlet speed, versus 11.0% with uniform inlet flow. ISLA reaches 5.3% across four seeds. For ISLA, holding out a geometry family has little effect relative to uniform flow, while holding out an enclosure or boundary-condition layout raises error 2.4 to 9.2 times. The abstract says the data and scoring code will be released.

### Source excerpt

arXiv:2611.17152v1 Announce Type: new Abstract: Neural surrogates for computational fluid dynamics (CFD) are trained and ranked on datasets that each vary one design family in one flow domain. Such datasets cannot test the least that an aerodynamic foundation model must do: predict CFD-quality incompressible flow in seconds for new shapes in new enclosures. On DrivAerML, larger models and longer training barely lower Transolver's error; doubling the training cases cuts it by about 12% up to 400 cases, exhausting the dataset. We introduce AeroZoo, a benchmark and training corpus of 3,815 time-mean solutions to the unsteady Reynolds-averaged Navier--Stokes (RANS) equations with a transition model, across seven geometry families, eight enclosure configurations and Reynolds numbers from $10^3$ to $10^8$. Each sample poses a well-posed PDE problem in the form a solver receives it: the boundary meshes grouped by condition type, the prescribed values and the fluid properties. The 2,816 samples with a stable time mean form the canonical benchmark: two tasks, engineering metrics and 22 frozen splits, disjoint in geometry, that hold out geometry families, enclosures, boundary-condition layouts and Reynolds ranges. On the random split, boundary-condition tokens cut GeoTransolver's median velocity error from 76.9% to 8.4% of the inlet speed in single-seed runs (uniform inlet flow: 11.0%). ISLA, a transformer for boundary-driven PDEs, reaches 5.3% over four seeds, about 4 times the change from a finer mesh. On the same test samples, a held-out geometry family barely changes ISLA's error relative to uniform flow, and a held-out enclosure or boundary-condition layout raises it 2.4 to 9.2 times. The data and the scoring code will be released.

## SUMNO: Spatially Unrolled Multigrid Neural Operator for Fluid Flow Surrogates

DevFeed: [SUMNO: Spatially Unrolled Multigrid Neural Operator for Fluid Flow Surrogates](<https://devfeed.tech/articles/sumno-spatially-unrolled-multigrid-neural-operator-for-fluid-flow-surrogates-132222.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.16341>)

Author: Yingqi Hu, Nan Deng, Changxiao Shao, Xiaozhou He

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>)

Tags: [architecture](<https://devfeed.tech/tags/architecture.md>), [attention](<https://devfeed.tech/tags/attention.md>), [computational-fluid-dynamics](<https://devfeed.tech/tags/computational-fluid-dynamics.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [fluid-dynamics](<https://devfeed.tech/tags/fluid-dynamics.md>), [operator](<https://devfeed.tech/tags/operator.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [surrogate-modeling](<https://devfeed.tech/tags/surrogate-modeling.md>)

### AI overview

SUMNO maps a geometric multigrid V-cycle onto a feed-forward U-shaped neural operator for fluid flow on regular grids. Parameter-free downsampling and upsampling represent restriction and prolongation, while additive skip connections provide coarse-grid correction through normalization-free pathways. The network uses convolutions at fine and intermediate scales, then shifted-window attention at the coarsest scale, with window partitioning adapted for independent periodicity along each axis. Across unseen configurations in four steady and unsteady flow benchmarks, it ranks first in five of seven settings and second in the other two. Its 10-step rollout MSE is about 30% lower than the strongest baseline; it also outperforms three baselines with more than 130 million parameters while using 9.4 million. These findings are limited to the reported benchmarks.

### Source excerpt

arXiv:2611.16341v1 Announce Type: cross Abstract: Data-driven surrogate modeling has emerged as a powerful paradigm for accelerating computational fluid dynamics, and hierarchical U-shaped networks, which resemble the structure of classical Geometric Multigrid (GMG) methods, have become prevalent designs. Modern surrogates built on these designs achieve competitive accuracy at a fraction of the cost of numerical solvers; however, accurate prediction of complex flows remains challenging, and no existing architecture leads across them. In such networks, the multigrid analogy mainly concerns the multi-resolution topology, while their components largely follow common vision practice. To address this, this study introduces SUMNO, a multigrid neural operator for fluid flow fields on regular grids that spatially unrolls a V-cycle into a feed-forward U-shaped network. Parameter-free down- and upsampling stand in for restriction and prolongation, and additive skip connections for coarse-grid correction, with normalization-free transfer and skip pathways. Within this macro-architecture, convolutions are used at the fine and intermediate resolutions and shifted-window attention at the coarsest, whose window partitioning is generalized to handle periodicity independently per axis. Evaluated on unseen test configurations across four unsteady and steady flow benchmarks against five representative baselines, SUMNO is best in 5 of 7 settings and ranks second in the remaining two. It reduces the 10-step rollout MSE by about 30% relative to the strongest baseline, and outperforms three baselines with over 130 million parameters using 9.4 million parameters.

## Interpretable Multifidelity Learning with Kolmogorov-Arnold Networks for Spatiotemporal Nuclear Reactor Flow Modeling

DevFeed: [Interpretable Multifidelity Learning with Kolmogorov-Arnold Networks for Spatiotemporal Nuclear Reactor Flow Modeling](<https://devfeed.tech/articles/interpretable-multifidelity-learning-with-kolmogorov-arnold-networks-for-spatiotemporal-nuclear-reactor-flow-modeling-123271.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.08512>)

Author: Meredith Eaheart, Majdi I. Radaideh

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>)

Tags: [cfd](<https://devfeed.tech/tags/cfd.md>), [correlation](<https://devfeed.tech/tags/correlation.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [flow](<https://devfeed.tech/tags/flow.md>), [forecasting](<https://devfeed.tech/tags/forecasting.md>), [kan](<https://devfeed.tech/tags/kan.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [networks](<https://devfeed.tech/tags/networks.md>), [neural-networks](<https://devfeed.tech/tags/neural-networks.md>), [nuclear](<https://devfeed.tech/tags/nuclear.md>), [surrogate-modeling](<https://devfeed.tech/tags/surrogate-modeling.md>), [synthetic](<https://devfeed.tech/tags/synthetic.md>)

### AI overview

The study finds that multifidelity Kolmogorov-Arnold Networks (MFKANs) perform comparably to feedforward neural networks for one-step-ahead prediction of mass flow across 193 PWR fuel assemblies, while also yielding closed-form symbolic equations. Absolute transient prediction remains limited, and true cross-fidelity forecasting is poor: correlations are 0.03-0.06, despite self-evaluation correlations of roughly 0.73-0.81. With only one CFD transient available per fidelity, the authors augmented forecasting with TabDDPM-generated data. Assembly-level results indicate that forecast accuracy tracks the underlying correlation between fidelity levels, suggesting weak cross-fidelity correlation is a central limitation. The abstract also reports that KAN symbolic equations match or exceed spline representation accuracy.

### Source excerpt

arXiv:2611.08512v1 Announce Type: new Abstract: High-fidelity computational fluid dynamics (CFD) simulations of nuclear reactor systems are computationally expensive, motivating machine learning methods that reduce cost while preserving accuracy. Multifidelity surrogate modeling addresses this by combining inexpensive low-fidelity data with limited high-fidelity data. This work investigates whether the high-fidelity data required to predict the spatiotemporal mass flow rate across all 193 fuel assemblies of a pressurized water reactor (PWR) core can be reduced, comparing black-box multifidelity feedforward neural networks (MFNNs) with interpretable multifidelity Kolmogorov-Arnold Networks (MFKANs). Cross-fidelity correlation across the core is weak, making this a challenging multifidelity problem, and neither architecture predicts the transient particularly well in absolute terms. MFKAN nonetheless performs comparably to MFNN for one-time-step-forward prediction while also providing closed-form symbolic equations unavailable with MFNN. To address the scarcity of CFD transients, with only one available per fidelity level, TabDDPM-generated synthetic data was used to augment cross-fidelity forecasting. KAN and FNN achieve comparable self-evaluation accuracy (Pearson correlation of roughly 0.73--0.81), but both perform poorly at true cross-fidelity forecasting (correlations of 0.03--0.06). Assembly-level analysis shows that each assembly's forecast correlation tracks its real cross-fidelity correlation for both architectures ($r=0.840$--$0.896$ for KAN, $r=0.804$--$0.903$ for FNN), confirming that forecasting accuracy is fundamentally limited by how correlated the fidelity levels already are. Overall, KAN-based multifidelity modeling achieves accuracy consistently comparable to MFNN across all settings, while additionally providing symbolic equations that match or exceed the spline representation's accuracy.

## Task-Contribution Dynamics of Adam in Physics-Informed Neural Networks: Slow-Manifold Analysis and Gradient Modulation

DevFeed: [Task-Contribution Dynamics of Adam in Physics-Informed Neural Networks: Slow-Manifold Analysis and Gradient Modulation](<https://devfeed.tech/articles/task-contribution-dynamics-of-adam-in-physics-informed-neural-networks-slow-manifold-analysis-and-gradient-modulation-123324.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.08910>)

Author: Enzhan Zheng, Xiaoliang Song

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Deep learning](<https://devfeed.tech/topics/deep-learning.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>)

Tags: [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-na](<https://devfeed.tech/tags/cs-na.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [gradient](<https://devfeed.tech/tags/gradient.md>), [math-na](<https://devfeed.tech/tags/math-na.md>), [neural-networks](<https://devfeed.tech/tags/neural-networks.md>), [numerical-analysis](<https://devfeed.tech/tags/numerical-analysis.md>), [physics](<https://devfeed.tech/tags/physics.md>), [physics-informed](<https://devfeed.tech/tags/physics-informed.md>)

### AI overview

The paper analyzes how individual loss-task contributions to Adam's parameter updates evolve in PINNs. It identifies smooth variation, intermittent spikes, and persistent oscillations, then focuses on the early smooth regime under full-batch training and a stated parameter condition. A first-order slow-manifold approximation attributes step-to-step redistribution to curvature-related alignment disparities and differences in coordinatewise preconditioned task-gradient distributions; both involve Adam's shared second-moment estimate. The proposed Gradient-Modulated Adam (GMAdam), inspired by AdaTask, uses task-specific second-moment estimates to modulate incoming task gradients. Experiments on high-frequency Poisson and wave equations, Allen-Cahn, and lid-driven cavity flow at Reynolds numbers 100 and 400 report better predictive accuracy than AdaTask at comparable computational cost.

### Source excerpt

arXiv:2611.08910v1 Announce Type: cross Abstract: Physics-informed neural networks (PINNs) often suffer from imbalanced optimization of competing loss components, particularly when approximating solutions with high-frequency or multiscale features. Although existing balancing strategies have proven effective in alleviating this difficulty, the mechanisms governing the evolution of individual task contributions to Adam's parameter updates remain insufficiently understood. We introduce task-wise indicators to characterize this evolution and identify three distinct training regimes: smooth variation, intermittent spikes, and persistent oscillations. Focusing on the early smooth regime, we combine discrete-time analysis with a continuous-time singular perturbation framework under the full-batch setting and the parameter regime $(1-\beta_2)/(1-\beta_1)\ll1$. A first-order approximation of the locally attracting slow manifold reveals two factors governing the step-to-step redistribution of task contributions: curvature-related alignment disparities and differences in the coordinatewise distributions of preconditioned task gradients. Both factors involve Adam's shared second-moment estimate. Inspired by AdaTask, we propose Gradient-Modulated Adam (GMAdam), which modulates incoming task gradients using task-specific second-moment estimates. Numerical experiments on high-frequency Poisson and wave equations, the Allen-Cahn equation, and two-dimensional lid-driven cavity flow ($Re=100,400$) demonstrate consistent improvements in predictive accuracy over AdaTask at comparable computational cost.

## Project and Generate: Divergence-Free Neural Operators for Incompressible Flows

DevFeed: [Project and Generate: Divergence-Free Neural Operators for Incompressible Flows](<https://devfeed.tech/articles/project-and-generate-divergence-free-neural-operators-for-incompressible-flows-123109.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2603.24500>)

Author: Xigui Li, Hongwei Zhang, Ruoxi Jiang, Deshu Chen, Chensen Lin, Limei Han, Yuan Qi, Xin Guo, Yuan Cheng

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>)

Tags: [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [curl](<https://devfeed.tech/tags/curl.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [flow](<https://devfeed.tech/tags/flow.md>), [flow-matching](<https://devfeed.tech/tags/flow-matching.md>), [fluid](<https://devfeed.tech/tags/fluid.md>), [fluid-dynamics](<https://devfeed.tech/tags/fluid-dynamics.md>), [framework](<https://devfeed.tech/tags/framework.md>), [function](<https://devfeed.tech/tags/function.md>), [gaussian](<https://devfeed.tech/tags/gaussian.md>), [generative](<https://devfeed.tech/tags/generative.md>), [integrate](<https://devfeed.tech/tags/integrate.md>), [intrinsic](<https://devfeed.tech/tags/intrinsic.md>), [learning](<https://devfeed.tech/tags/learning.md>), [matching](<https://devfeed.tech/tags/matching.md>), [modeling](<https://devfeed.tech/tags/modeling.md>), [models](<https://devfeed.tech/tags/models.md>), [navier-stokes-equations](<https://devfeed.tech/tags/navier-stokes-equations.md>), [neural](<https://devfeed.tech/tags/neural.md>), [operators](<https://devfeed.tech/tags/operators.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [projection](<https://devfeed.tech/tags/projection.md>)

### AI overview

The framework enforces incompressibility as a hard constraint in deterministic and generative neural operators. It applies a differentiable spectral Leray projection; for flow matching, it builds a divergence-free Gaussian reference measure through a curl-based pushforward, aligning reference samples, interpolation paths, and velocity targets with the projected dynamics. In 2D Navier-Stokes experiments, the authors report incompressibility up to discretization error and improved stability and physical consistency. The supplied text is an abstract, so it does not provide further experimental details.

### Source excerpt

arXiv:2603.24500v2 Announce Type: replace Abstract: Learning-based models for fluid dynamics often operate in unconstrained function spaces, leading to physically inadmissible, unstable simulations. Penalty-based methods can reduce constraint violations but do not guarantee exact incompressibility. In this work, we introduce a unified framework that enforces the incompressible continuity equation as a hard, intrinsic constraint for both deterministic and generative modeling. We integrate a differentiable spectral Leray projection into neural operators to enforce incompressibility. For flow matching, we construct a divergence-free Gaussian reference measure via a curl-based pushforward, so that the reference samples, interpolation paths, and velocity targets are compatible with the projected dynamics. Experiments on 2D Navier-Stokes equations demonstrate exact incompressibility up to discretization error and substantially improved stability and physical consistency.

## Readable Is Not Steerable: Pre-Registered Tests of Physical Structure in Learned Fluid Representations

DevFeed: [Readable Is Not Steerable: Pre-Registered Tests of Physical Structure in Learned Fluid Representations](<https://devfeed.tech/articles/readable-is-not-steerable-pre-registered-tests-of-physical-structure-in-learned-fluid-representations-118346.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.05577>)

Author: Manvendra Modgil

Published: 2026-11-05T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Representation learning](<https://devfeed.tech/topics/representation-learning.md>), [World models](<https://devfeed.tech/topics/world-models.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>)

Tags: [autoencoders](<https://devfeed.tech/tags/autoencoders.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data](<https://devfeed.tech/tags/data.md>), [decoder](<https://devfeed.tech/tags/decoder.md>), [field](<https://devfeed.tech/tags/field.md>), [flow](<https://devfeed.tech/tags/flow.md>), [fluid](<https://devfeed.tech/tags/fluid.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [linear](<https://devfeed.tech/tags/linear.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [routing](<https://devfeed.tech/tags/routing.md>), [rules](<https://devfeed.tech/tags/rules.md>), [scale](<https://devfeed.tech/tags/scale.md>), [sealed](<https://devfeed.tech/tags/sealed.md>), [shape](<https://devfeed.tech/tags/shape.md>), [static](<https://devfeed.tech/tags/static.md>), [study](<https://devfeed.tech/tags/study.md>), [tests](<https://devfeed.tech/tags/tests.md>), [time](<https://devfeed.tech/tags/time.md>), [variance](<https://devfeed.tech/tags/variance.md>), [world-models](<https://devfeed.tech/tags/world-models.md>)

### AI overview

On held-out 2D decaying Navier-Stokes data, learned representations did not consistently organize around physical quantities or make those quantities steerable. A flow-predicting JEPA did not outperform an untrained network by the pre-registered criteria; a JEPA trained on non-physical dynamics and a static invariance model did as well or better. Plain autoencoders made enstrophy prominent in slow latent directions (best r 0.80-0.86 versus 0.38 untrained), while adding VICReg reduced that prominence but increased linear decodability. In a routed decoder, label-free rules found a loudness dial in all eight trained models, but a shape dial in only three of five fresh seeds, below the pre-registered threshold of four. A direction that read spectral shape almost perfectly during exploration had almost no steering effect. Thus, latent information and output control diverged in these experiments; the abstract reports partially supported results and notes that testing was once per study on held-out data.

### Source excerpt

arXiv:2611.05577v1 Announce Type: cross Abstract: Self-supervised world models are often said to organize their latents around physical quantities. We test this on 2D decaying Navier-Stokes flow in three pre-registered studies, each tested once on held-out data. In Study 1, a joint-embedding predictive architecture (JEPA) trained to predict the flow fails the pre-registered criteria for organizing its latent around fluid quantities better than an untrained network of the same architecture (not supported); a JEPA trained on non-physical linear dynamics and a static invariance model that never sees time do as well or better. In Study 2, plain autoencoders that render the field make enstrophy prominent in their slowest unsupervised latent directions (best r 0.80-0.86 vs 0.38 untrained; fresh seeds, sealed data). Adding VICReg lowers this prominence in every comparison, though not by the pre-registered margin for vorticity, while raising the autoencoders' linear decodability (partially supported). In Study 3, we build steerable directions ("dials") by routing: the decoder gets overall scale only through a small conditioning code, while its other pathway carries a degraded copy of spectral shape. Label-free rules find a loudness dial in all 8 trained models and a shape dial in 3 of 5 fresh seeds on sealed data, against a pre-registered bar of 4; neither control shows a shape dial (partially supported). Across the studies, what a latent contains and what can be controlled through it come apart. In exploration, a supervised direction reading spectral shape almost perfectly moved nothing when steered. During the build, the loudness dial had to be found by its effect on the output (closest to a pure rescaling), not by data variance; for shape, a variance rule and an effect-constrained rule each worked in 3 of 5 seeds. The program used about 70 logged hours on one consumer GPU; all pre-registrations, deviations, code and outputs are public.

## Deep Neural Operator Learning for stochastic heat equations and applications

DevFeed: [Deep Neural Operator Learning for stochastic heat equations and applications](<https://devfeed.tech/articles/deep-neural-operator-learning-for-stochastic-heat-equations-and-applications-116998.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.04955>)

Author: Hengrong Du, Qi Feng, Weinan Wang

Published: 2026-11-04T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Deep learning](<https://devfeed.tech/topics/deep-learning.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>)

Tags: [applications](<https://devfeed.tech/tags/applications.md>), [approximation](<https://devfeed.tech/tags/approximation.md>), [chaos](<https://devfeed.tech/tags/chaos.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [deep](<https://devfeed.tech/tags/deep.md>), [gaussian](<https://devfeed.tech/tags/gaussian.md>), [grid](<https://devfeed.tech/tags/grid.md>), [heat](<https://devfeed.tech/tags/heat.md>), [learning](<https://devfeed.tech/tags/learning.md>), [linear](<https://devfeed.tech/tags/linear.md>), [local](<https://devfeed.tech/tags/local.md>), [math-ap](<https://devfeed.tech/tags/math-ap.md>), [math-pr](<https://devfeed.tech/tags/math-pr.md>), [measurements](<https://devfeed.tech/tags/measurements.md>), [navier-stokes-equations](<https://devfeed.tech/tags/navier-stokes-equations.md>), [networks](<https://devfeed.tech/tags/networks.md>), [neural](<https://devfeed.tech/tags/neural.md>), [operator](<https://devfeed.tech/tags/operator.md>), [operator-learning](<https://devfeed.tech/tags/operator-learning.md>), [reduction](<https://devfeed.tech/tags/reduction.md>), [relu](<https://devfeed.tech/tags/relu.md>)

### AI overview

The paper proves universal approximation results for solution operators of semilinear stochastic heat equations in spatial dimensions d>=2, including equations driven by distribution-valued Gaussian forcing. A smooth Fourier cutoff reduces the forcing to finitely many measurements; Besov and Schauder estimates bound the resulting solution error. ReLU branch networks and grid-based trunk functions then approximate the finite-dimensional solution map, with explicit bounds on measurement and trunk-function counts and convergence in every finite stochastic moment. For linear equations, the reduction is equivalent to truncating the first Wiener chaos. The paper also establishes high-probability local-in-time approximation for stochastic Navier-Stokes equations using finite measurements of both the initial condition and forcing, with an explicit sensing rate.

### Source excerpt

arXiv:2611.04955v1 Announce Type: new Abstract: In this paper, we establish a universal approximation theorem for solution operators of semilinear stochastic heat equations in spatial dimension \(d\ge2\), extending deep neural operator approximation results to SPDEs with distribution-valued Gaussian forcing. A smooth Fourier cutoff reduces the forcing to finitely many measurements, while Besov and Schauder estimates control the resulting solution error. The finite-dimensional solution map is then approximated by ReLU branch networks together with grid-based trunk functions, yielding explicit bounds on the numbers of forcing measurements and trunk functions and approximation in every finite stochastic moment. For linear equations, the reduction coincides with truncation of the first Wiener chaos. Lastly, as an application, we prove a high-probability local-in-time approximation theorem for stochastic Navier--Stokes equations, with finite measurements of both the initial condition and forcing and an explicit sensing rate.

## Convergent Science Advances Engineering Simulation on CoreWeave

DevFeed: [Convergent Science Advances Engineering Simulation on CoreWeave](<https://devfeed.tech/articles/convergent-science-advances-engineering-simulation-on-coreweave-114287.md>)

Original publisher: [Read original article](<https://wf.coreweave.com/blog/convergent-science-advances-engineering-simulation-on-coreweave>)

Author: CoreWeave

Published: 2026-11-03T15:33:33Z

Content type: article

Language: en

Sources: [Coreweave Blog Posts](<https://devfeed.tech/sources/coreweave-blog-posts.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>), [gpudirect rdma](<https://devfeed.tech/topics/gpudirect-rdma.md>), [NVLink](<https://devfeed.tech/topics/nvlink.md>)

Tags: [blog](<https://devfeed.tech/tags/blog.md>), [cfd](<https://devfeed.tech/tags/cfd.md>), [coreweave](<https://devfeed.tech/tags/coreweave.md>), [dns](<https://devfeed.tech/tags/dns.md>), [engineering-simulation](<https://devfeed.tech/tags/engineering-simulation.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [scale](<https://devfeed.tech/tags/scale.md>), [simulations](<https://devfeed.tech/tags/simulations.md>), [size](<https://devfeed.tech/tags/size.md>)

### AI overview

Convergent Science ran a 536-million-cell CONVERGE CFD direct numerical simulation to completion in 3.62 hours on two NVIDIA HGX B200 instances (16 Blackwell GPUs). In its scaling study, it also compared NVIDIA RTX PRO 6000 Blackwell Server Edition and HGX B200 using a validated turbulent-flow case. The excerpt distinguishes weak scaling (larger meshes as GPU resources grow) from strong scaling (same mesh completed faster), and explains that memory determines whether a job fits while communication affects speed as GPUs are added. RTX PRO systems target jobs within one server, using PCIe between GPUs; HGX B200 supports larger, multi-node cases with more memory per GPU, NVLink within nodes, and 400 Gb/s InfiniBand between them. Results depend on workload and topology. The tested DNS case was lean; production simulations with additional physics may require more memory at the same cell count, so its figures may underestimate those needs.

### Source excerpt

How far can GPU CFD scale? Convergent Science ran 536 million cells on two NVIDIA HGX B200 instances in 3.62 hours. Here is how to size and scale your own simulations.

## DynaSolver: A Neural Operator for Autoregressive 4D Flow Prediction with Dynamic Boundaries

DevFeed: [DynaSolver: A Neural Operator for Autoregressive 4D Flow Prediction with Dynamic Boundaries](<https://devfeed.tech/articles/dynasolver-a-neural-operator-for-autoregressive-4d-flow-prediction-with-dynamic-boundaries-114814.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.03595>)

Author: Rui Qian, Shibo Dai, Mengjun Lin, Yuk-Kit Hou, Saining Zhang, Bohan Li, Junhao Chen, Hao Zhao, Chaojian Li

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>)

Tags: [autoregressive](<https://devfeed.tech/tags/autoregressive.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [boundary](<https://devfeed.tech/tags/boundary.md>), [cfd](<https://devfeed.tech/tags/cfd.md>), [computational-fluid-dynamics](<https://devfeed.tech/tags/computational-fluid-dynamics.md>), [computational-fluid-dynamics-cfd](<https://devfeed.tech/tags/computational-fluid-dynamics-cfd.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [evolution](<https://devfeed.tech/tags/evolution.md>), [flow](<https://devfeed.tech/tags/flow.md>), [fluid](<https://devfeed.tech/tags/fluid.md>), [fluid-dynamics](<https://devfeed.tech/tags/fluid-dynamics.md>), [history](<https://devfeed.tech/tags/history.md>), [model](<https://devfeed.tech/tags/model.md>), [moving](<https://devfeed.tech/tags/moving.md>), [neural](<https://devfeed.tech/tags/neural.md>), [operator](<https://devfeed.tech/tags/operator.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [point](<https://devfeed.tech/tags/point.md>), [prediction](<https://devfeed.tech/tags/prediction.md>), [projection](<https://devfeed.tech/tags/projection.md>), [resolved](<https://devfeed.tech/tags/resolved.md>)

### AI overview

DynaSolver targets 4D CFD prediction for articulated geometries with continuously moving boundaries, a setting the authors say static-geometry neural operators are not designed for. The work introduces RoboCFD-4D, a temporally resolved benchmark spanning diverse morphologies and flow conditions. DynaSolver conditions on boundary evolution and causal flow history; its physical-token formulation amortizes point-to-token projection across the operator stack to reduce repeated point-space computation while retaining point-level reconstruction. Replay-Augmented Parallel Training adds autoregressively generated histories while preserving parallel dense supervision. On RoboCFD-4D, the authors report the best full-rollout accuracy among a broad range of learned CFD surrogates, with a growing advantage over longer rollouts, indicating reduced autoregressive error accumulation.

### Source excerpt

arXiv:2611.03595v1 Announce Type: new Abstract: Neural operators provide efficient surrogates for computational fluid dynamics (CFD), but most are designed for static geometries and are not equipped to model unsteady flows induced by continuously moving boundaries. We study dynamic-boundary 4D CFD prediction for articulated moving geometries and introduce RoboCFD-4D, a temporally resolved robotic CFD benchmark spanning diverse morphologies and flow conditions. To address the computational and autoregressive challenges of this setting, we propose DynaSolver, a neural operator that explicitly conditions on boundary evolution and causal flow history. DynaSolver employs an efficient physical-token formulation that amortizes point-to-token projection across the operator stack, substantially reducing repeated point-space spatio-temporal computation while retaining accurate point-level reconstruction. We further introduce Replay-Augmented Parallel Training, which exposes the model to autoregressively generated histories while preserving parallel dense supervision, narrowing the gap between ground-truth-conditioned training and autoregressive rollout. Experiments on RoboCFD-4D show that DynaSolver achieves the best full-rollout accuracy among a broad range of learned CFD surrogates, with its advantage increasing as rollout proceeds, indicating substantially reduced autoregressive error accumulation. Project page: https://made-gpt.github.io/DynaSolver-ProjectPage.

## FastAgent: A Physics-Aware FastFluent-to-Fluent Workflow for Rapid CFD Screening and Industrial Steam Dewaxing Design

DevFeed: [FastAgent: A Physics-Aware FastFluent-to-Fluent Workflow for Rapid CFD Screening and Industrial Steam Dewaxing Design](<https://devfeed.tech/articles/fastagent-a-physics-aware-fastfluent-to-fluent-workflow-for-rapid-cfd-screening-and-industrial-steam-dewaxing-design-115074.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.00645>)

Author: Chen YanKai, Yu QiuHeng, Yuan Man, Zou Ting, Wu Lin, He Jian, Zhang QiongYuan, Jing Tao, Xu QingYan

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Simulation and Design](<https://devfeed.tech/topics/simulation-and-design.md>)

Tags: [adaptive](<https://devfeed.tech/tags/adaptive.md>), [agent](<https://devfeed.tech/tags/agent.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [cfd](<https://devfeed.tech/tags/cfd.md>), [crack](<https://devfeed.tech/tags/crack.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [edge](<https://devfeed.tech/tags/edge.md>), [industrial](<https://devfeed.tech/tags/industrial.md>), [investment](<https://devfeed.tech/tags/investment.md>), [layouts](<https://devfeed.tech/tags/layouts.md>), [mechanism](<https://devfeed.tech/tags/mechanism.md>), [models](<https://devfeed.tech/tags/models.md>), [outlet](<https://devfeed.tech/tags/outlet.md>), [physics](<https://devfeed.tech/tags/physics.md>), [rapid](<https://devfeed.tech/tags/rapid.md>), [search](<https://devfeed.tech/tags/search.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [steam](<https://devfeed.tech/tags/steam.md>), [workflow](<https://devfeed.tech/tags/workflow.md>)

### AI overview

FastAgent uses FastFluent for rapid, stage-specific CFD screening, then adapts physical models, resolution, and simulation horizons to active quantities of interest; discrepancies from Fluent inform later choices. Across six public benchmarks, it reduced normalized worst-QoI error by 5.68-99.14% and sped decision-relevant computation 1.53-15.36 times versus a matched Generic Agent control. In high-pressure steam dewaxing, it identified nonuniform condensation heating as the dominant startup mechanism and separated rapid risk-window localization from high-fidelity melt-completion confirmation; the rapid route used 81.39% fewer cell-time updates than a fine-grid, small-time-step reference. Screening 131,072 outlet layouts with selective Fluent and Abaqus evaluation identified Sparse-6, which reduced simulated crack opening by 96.0% and connected crack extent by 66.7% versus the Original design. These are abstract-level reported results.

### Source excerpt

arXiv:2611.00645v1 Announce Type: new Abstract: Industrial computational fluid dynamics (CFD) becomes costly when complex geometries, long transients and coupled physics must be evaluated across large design spaces. We introduce FastAgent, a physics-aware campaign controller built around FastFluent, a lightweight numerical engine for rapid physical screening. FastFluent supplies stage-specific fields and diagnostics; FastAgent allocates physical models, numerical resolution and simulation horizons by the active quantities of interest (QoIs), while returned Fluent discrepancies update subsequent decisions. Across six public CFD benchmark campaigns, the framework reduces normalised worst-QoI error by 5.68-99.14% and accelerates decision-relevant computation by factors of 1.53-15.36 relative to a matched Generic Agent control. Applied to high-pressure steam dewaxing in investment casting, FastAgent identifies nonuniform condensation heating as the dominant startup mechanism and separates rapid risk-window localisation from high-fidelity melt-completion confirmation. The rapid route requires 81.39% fewer cell-time updates than a fine-grid, small-time-step reference. Structural analysis localises fracture at the observed blade edge. Screening 131,072 outlet layouts and selectively evaluating candidates with Fluent and Abaqus identifies Sparse-6, which reduces simulated crack opening by 96.0% and connected crack extent by 66.7% relative to the Original design. These results demonstrate how rapid numerical evidence and adaptive fidelity control connect large CFD search spaces to mechanism-guided industrial design.

## Hard Constraint Projection in a Physics Informed Neural Network (Burgers)

DevFeed: [Hard Constraint Projection in a Physics Informed Neural Network (Burgers)](<https://devfeed.tech/articles/hard-constraint-projection-in-a-physics-informed-neural-network-burgers-111668.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.42106>)

Author: Miranda J S Horne, Peter K Jimack, Amirul Khan, He Wang

Published: 2026-11-02T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Physics-guided deep learning](<https://devfeed.tech/topics/physics-guided-deep-learning.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [generalization in machine learning](<https://devfeed.tech/topics/generalization-in-machine-learning.md>)

Tags: [constraint](<https://devfeed.tech/tags/constraint.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [fluid-dynamics](<https://devfeed.tech/tags/fluid-dynamics.md>), [linear-algebra](<https://devfeed.tech/tags/linear-algebra.md>), [neural-network](<https://devfeed.tech/tags/neural-network.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [physics-informed](<https://devfeed.tech/tags/physics-informed.md>), [physics-informed-machine-learning](<https://devfeed.tech/tags/physics-informed-machine-learning.md>), [projection](<https://devfeed.tech/tags/projection.md>)

### AI overview

The study extends hard constraint projection (HCP) to a physics informed neural network for the nonlinear 1D Burgers' equation. It locally discretizes the equation with a finite difference stencil, uses the resulting constraint matrix to build a projection, and adds an unlearnable final layer that projects the network's velocity prediction onto the exact-solution hyperplane of the discrete equation. In initial results, HCP-PINN trains similarly to a comparable PINN at first, but continues improving after the PINN loss plateaus. On a previously unseen uniform grid, it achieves roughly an order of magnitude lower data error and better adherence to the governing PDE. The authors suggest it may be more robust to collocation point selection and less dependent on the number of those points.

### Source excerpt

arXiv:2610.42106v1 Announce Type: cross Abstract: Physics informed machine learning provides a framework to study and simulate fluid dynamics, potentially at a fraction of the computational cost of traditional numerical methods, with the goal of improving the explainability, accuracy and robustness of purely data driven machine learning methods. In this work, we extend the hard constraint projection (HCP) method developed by Chen et al. [Journal of Computational Physics, 445, 110624, 2021] from a linear partial differential equation (PDE) to a nonlinear PDE: the 1D Burgers' equation. To strictly impose a discrete approximation of the governing equation, the Burgers' equation is first locally discretised (using a finite difference stencil) then decomposed into a constraint matrix and a prediction matrix. Using tools from linear algebra, the constraint matrix is used to construct a projection matrix. A physics informed neural network (PINN) [Raissi et al. Journal of Computational Physics, 378, 686, 2019] is used to predict the velocity, and a final unlearnable layer projects the velocity prediction to the hyperplane of exact solutions to a discretised form of the Burgers' equation. Our initial results indicate that the HCP-PINN initially trains similarly to an equivalent PINN, and as the PINN loss plateaus, the HCP-PINN loss continues to decline. When tested on a previously unseen uniform grid of points covering the domain, the HCP-PINN exhibits higher accuracy, converging to a data error approximately an order of magnitude lower than the comparable PINN. We also observe that the inclusion of the HCP layer results in better observance of the governing PDE. Our findings suggest that the HCP-PINN can be more robust to the selection of physics collocation points, and less dependent on collocation point quantity.

## Rank-Aware CPU Reservation and In-Place Adjustment for MPI CFD on Kubernetes

DevFeed: [Rank-Aware CPU Reservation and In-Place Adjustment for MPI CFD on Kubernetes](<https://devfeed.tech/articles/rank-aware-cpu-reservation-and-in-place-adjustment-for-mpi-cfd-on-kubernetes-107553.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2603.22691>)

Author: Tianfang Xie

Published: 2026-10-30T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>)

Tags: [adjustment](<https://devfeed.tech/tags/adjustment.md>), [cfd](<https://devfeed.tech/tags/cfd.md>), [controller](<https://devfeed.tech/tags/controller.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [job](<https://devfeed.tech/tags/job.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [mesh](<https://devfeed.tech/tags/mesh.md>), [rank](<https://devfeed.tech/tags/rank.md>), [throughput](<https://devfeed.tech/tags/throughput.md>), [waiting](<https://devfeed.tech/tags/waiting.md>)

### AI overview

The framework uses MPI profiles to reserve CPU according to rank-level demand and adjust requests while OpenFOAM runs on Kubernetes, without hard CPU caps or changes to OpenFOAM, MPI, or the scheduler. It cut per-job reservations from 16.0 to 5.9 vCPU with at most a 6% increase in single-job wall time; six simulations on twelve nodes achieved 4.09x sequential throughput. Proportional request shaping showed no consistent benefit, and oversubscribing nominal CPU capacity caused throughput to collapse with default busy-waiting, though yielding waits softened the drop. In one run, 120 in-place resizes caused no pod restarts, and the pipeline reached 2.5x sequential throughput. Separately, reallocating ranks from far-field to near-wall mesh zones reduced fixed-iteration time by 2.0-4.2x versus the original concentric allocation and outperformed balanced Scotch partitions; results varied by mesh and iteration horizon.

### Source excerpt

arXiv:2603.22691v2 Announce Type: replace Abstract: Non-uniform meshes in computational fluid dynamics (CFD) can leave some Message Passing Interface (MPI) ranks waiting while others determine execution time. On Kubernetes, reserving a full CPU for every rank can then restrict the admission of further simulations. We investigate rank-aware CPU reservation and in-place request adjustment using OpenFOAM on three Kubernetes testbeds. The framework lowers reservations without hard CPU caps and uses online MPI profiles to adjust each rank's CPU request while the solver runs. We reduced the per-job CPU reservation from 16.0 to 5.9 vCPU, with single-job wall-clock time increasing by at most 6%. With these reservations on twelve nodes, six co-located simulations reached 4.09 times the throughput of sequential execution. Matched controls showed no consistent benefit from proportional request shaping. Beyond the nominal CPU capacity, throughput collapsed with default busy-waiting; yielding waits softened the collapse. In one six-simulation run, the online controller performed 120 in-place resizes of running solvers with zero pod restarts; the pipeline reached 2.5 times the throughput of sequential execution. In a separate comparison on 16 physical cores, reallocating ranks from the far field to the near-wall zone reduced fixed-iteration wall-clock time by factors of 2.0 to 4.2 relative to the original concentric allocation, depending on mesh and iteration horizon. The reallocation also outperformed balanced Scotch partitions. The implementation requires no changes to OpenFOAM, the MPI library, or the Kubernetes scheduler.

## CraftSPH: A composable framework for differentiable smoothed particle hydrodynamics implemented in PyTorch

DevFeed: [CraftSPH: A composable framework for differentiable smoothed particle hydrodynamics implemented in PyTorch](<https://devfeed.tech/articles/craftsph-a-composable-framework-for-differentiable-smoothed-particle-hydrodynamics-implemented-in-pytorch-103827.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2609.38208>)

Author: Gen Matono, Shujiro Fujioka, Mayuko Nishio

Published: 2026-10-28T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [PyTorch](<https://devfeed.tech/topics/pytorch.md>)

Tags: [automatic](<https://devfeed.tech/tags/automatic.md>), [coupling](<https://devfeed.tech/tags/coupling.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-na](<https://devfeed.tech/tags/cs-na.md>), [deep-learning](<https://devfeed.tech/tags/deep-learning.md>), [flow](<https://devfeed.tech/tags/flow.md>), [fluid](<https://devfeed.tech/tags/fluid.md>), [fluid-dynamics](<https://devfeed.tech/tags/fluid-dynamics.md>), [framework](<https://devfeed.tech/tags/framework.md>), [math-na](<https://devfeed.tech/tags/math-na.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [pytorch](<https://devfeed.tech/tags/pytorch.md>)

### AI overview

CraftSPH is a modular, differentiable smoothed particle hydrodynamics framework implemented in PyTorch. Users combine independent computational modules to build problem-specific solvers, with high-accuracy discretization and both explicit and implicit schemes available in a common framework. Selected operations support automatic differentiation for parameter estimation and possible coupling with deep learning models. Forward tests using an implicit scheme agreed well with analytical or reference solutions for Poiseuille flow and a rising bubble, and captured key experimental trends in 2D and 3D dam-break problems. In inverse tests, explicit and implicit schemes estimated kinematic viscosity for Taylor-Green vortex flow and Reynolds number for lid-driven cavity flow. The authors also found that which operations are included in the computational graph affects the resulting gradients.

### Source excerpt

arXiv:2609.38208v2 Announce Type: replace-cross Abstract: Smoothed particle hydrodynamics (SPH) is well suited to large-deformation fluid problems, and the growing use of automatic differentiation in engineering applications has motivated interest in differentiable SPH solvers. In this study, a composable differentiable SPH framework, CraftSPH, has been developed. In CraftSPH, major computational operations are implemented as independent modules that users can combine to construct solvers for different purposes. This design supports high-accuracy discretization and both explicit and implicit computation within a common framework, while allowing individual components to be maintained and extended. Selected numerical operations also support automatic differentiation (AD) for physical-parameter estimation and potential coupling with deep learning models. These capabilities were demonstrated through forward and inverse analyses. Forward analyses using an implicit SPH scheme with high-accuracy discretization showed good agreement with analytical and reference solutions for Poiseuille flow and a rising bubble, while capturing the key experimental trends in two- and three-dimensional dam-break problems. In inverse analyses, explicit and implicit SPH schemes were used to estimate kinematic viscosity in Taylor-Green vortex flow and Reynolds number in lid-driven cavity flow. The results also showed that the choice of operations included in the computational graph affects the computed gradients. Together, these findings demonstrate that CraftSPH enables users to assemble problem-specific solvers for forward and inverse analyses within a common framework.

## Solving and Learning Navier-Stokes Dynamics with the Neural Basis Method

DevFeed: [Solving and Learning Navier-Stokes Dynamics with the Neural Basis Method](<https://devfeed.tech/articles/solving-and-learning-navier-stokes-dynamics-with-the-neural-basis-method-99569.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.32167>)

Author: Yuhe Wang, Zhu Wang, Min Wang

Published: 2026-10-26T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [approach](<https://devfeed.tech/tags/approach.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-na](<https://devfeed.tech/tags/cs-na.md>), [data](<https://devfeed.tech/tags/data.md>), [define](<https://devfeed.tech/tags/define.md>), [effectively](<https://devfeed.tech/tags/effectively.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [error](<https://devfeed.tech/tags/error.md>), [errors](<https://devfeed.tech/tags/errors.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [fixed](<https://devfeed.tech/tags/fixed.md>), [flow](<https://devfeed.tech/tags/flow.md>), [fluid](<https://devfeed.tech/tags/fluid.md>), [fluid-dynamics](<https://devfeed.tech/tags/fluid-dynamics.md>), [free](<https://devfeed.tech/tags/free.md>), [here](<https://devfeed.tech/tags/here.md>), [learning](<https://devfeed.tech/tags/learning.md>), [math-na](<https://devfeed.tech/tags/math-na.md>), [neural](<https://devfeed.tech/tags/neural.md>), [physics-flu-dyn](<https://devfeed.tech/tags/physics-flu-dyn.md>), [with](<https://devfeed.tech/tags/with.md>)

### AI overview

The neural basis method (NBM) represents Navier-Stokes solutions with fixed neural bases, then projects the governing equations to determine their expansion coefficients. Reusing those bases as shared coordinates turns operator learning into a parameter-to-coefficient mapping, removing explicit spatial dependence. For incompressible flow, NBM uses a velocity-only formulation whose residual gives an a posteriori velocity-error estimate, supporting a certified, data-free training objective for coefficient learning. The abstract reports theoretical analysis and numerical experiments showing accuracy, efficiency, generalization, scalability, and substantial improvements over baselines; it does not provide specific benchmark results or comparison details.

### Source excerpt

arXiv:2610.32167v1 Announce Type: cross Abstract: Physics-informed machine learning has become a major direction in predictive modeling of fluid dynamics, combining governing physics with learned representations. Yet prevailing formulations neither guarantee that small PDE-residual losses imply small solution errors nor scale effectively with increasing problem size. Here we present the neural basis method (NBM) as a new approach for Navier--Stokes dynamics. NBM is built on a ``representation-then-projection'' principle: fixed neural bases define the representation space, while projection of the governing equations determines the expansion coefficients. The same bases define common coordinates for the parametric solution manifold, recasting operator learning as a parameter-to-coefficient mapping and enabling scalability by removing explicit spatial dependence. For incompressible flow, NBM yields a velocity-only formulation whose residual provides an a posteriori velocity-error estimate, establishing a certified, data-free training objective for coefficient learning. Theoretical analysis and numerical experiments demonstrate the accuracy, efficiency, generalization, and scalability of NBM, with substantial improvements over baselines.

## RheOPT: Few-Shot Non-Newtonian Flow Prediction via Newtonian Pretraining of RheOFormer

DevFeed: [RheOPT: Few-Shot Non-Newtonian Flow Prediction via Newtonian Pretraining of RheOFormer](<https://devfeed.tech/articles/rheopt-few-shot-non-newtonian-flow-prediction-via-newtonian-pretraining-of-rheoformer-97068.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.31465>)

Author: Maedeh Saberi, Amir Barati Farimani, Safa Jamali

Published: 2026-10-23T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [few-shot](<https://devfeed.tech/topics/few-shot.md>), [few-shot-learning](<https://devfeed.tech/topics/few-shot-learning.md>), [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>), [Inference](<https://devfeed.tech/topics/inference.md>)

Tags: [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [few-shot](<https://devfeed.tech/tags/few-shot.md>), [flow](<https://devfeed.tech/tags/flow.md>), [neural](<https://devfeed.tech/tags/neural.md>), [operator-learning](<https://devfeed.tech/tags/operator-learning.md>), [pretraining](<https://devfeed.tech/tags/pretraining.md>), [transformer](<https://devfeed.tech/tags/transformer.md>)

### AI overview

RheOPT pretrains a transformer neural operator on inexpensive Newtonian simulations, then fine-tunes it on a few non-Newtonian samples, reusing geometry-conditioned transport structure rather than Newtonian constitutive behavior. In a 4:1 contraction across Giesekus, Oldroyd-B, and Bingham fluids, four target samples matched or exceeded a from-scratch model trained on four times as many; the gain was reliable with reversible instance normalization (RevIN) but dataset-dependent with Global normalization, which the authors link to differences in velocity and stress dynamic ranges. For Oldroyd-B flow past a triangular obstacle, phase-insensitive diagnostics showed pretraining recovered wake shedding frequency and fluctuation amplitude suppressed by few-shot training from scratch. The evidence is limited to these two geometry-matched settings.

### Source excerpt

arXiv:2610.31465v1 Announce Type: new Abstract: Despite advances in numerical methods, non-Newtonian flow simulations remain computationally expensive because velocity and stress must be advanced together through rate- and history-dependent constitutive dynamics. Neural-operator surrogates reduce inference cost after training, but their training data require the same expensive simulations they aim to replace, making data-efficient learning essential when only a few non-Newtonian simulations can be afforded. Here, we introduce the Rheological Operator Pretrained Transformer (RheOPT), which pretrains a transformer-based neural operator on inexpensive Newtonian simulations and fine-tunes it on a small number of non-Newtonian samples. Rather than transferring the Newtonian constitutive response itself, RheOPT reuses geometry-conditioned transport structure that remains useful after the stress model changes. We evaluate RheOPT in two geometry-matched settings: a $4{:}1$ contraction channel for Giesekus, Oldroyd-B, and Bingham fluids, and the oscillatory wake behind a triangular obstacle for Oldroyd-B flow. In the contraction, Newtonian pretraining substantially improves few-shot prediction, and a model fine-tuned on four target samples can match or exceed a from-scratch model trained on four times as many. This benefit is reliable under reversible instance normalization (RevIN) but dataset dependent under Global normalization, a difference we trace to the dynamic range of each fluid's velocity and stress fields. In the wake, where relative $L_2$ error alone cannot verify that vortex shedding is recovered, phase-insensitive diagnostics show that pretraining recovers the shedding frequency and wake fluctuation amplitude that are suppressed by the from-scratch model in the few-shot regime. Overall, RheOPT establishes Newtonian simulation as a physically grounded pretraining source for data-efficient non-Newtonian operator learning.

## Structure-preserving reduced-order models of thin panels in supersonic and hypersonic flow

DevFeed: [Structure-preserving reduced-order models of thin panels in supersonic and hypersonic flow](<https://devfeed.tech/articles/structure-preserving-reduced-order-models-of-thin-panels-in-supersonic-and-hypersonic-flow-97073.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.31504>)

Author: Miguel A. Aguilo

Published: 2026-10-23T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Computational Fluid Dynamics (CFD)](<https://devfeed.tech/topics/cfd.md>), [structure](<https://devfeed.tech/topics/structure.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [all](<https://devfeed.tech/tags/all.md>), [also](<https://devfeed.tech/tags/also.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [behavior](<https://devfeed.tech/tags/behavior.md>), [build](<https://devfeed.tech/tags/build.md>), [can](<https://devfeed.tech/tags/can.md>), [challenge](<https://devfeed.tech/tags/challenge.md>), [condition](<https://devfeed.tech/tags/condition.md>), [cs-ce](<https://devfeed.tech/tags/cs-ce.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-na](<https://devfeed.tech/tags/cs-na.md>), [data](<https://devfeed.tech/tags/data.md>), [edge](<https://devfeed.tech/tags/edge.md>), [energy](<https://devfeed.tech/tags/energy.md>), [error](<https://devfeed.tech/tags/error.md>), [experiment](<https://devfeed.tech/tags/experiment.md>), [flow](<https://devfeed.tech/tags/flow.md>), [flutter](<https://devfeed.tech/tags/flutter.md>), [form](<https://devfeed.tech/tags/form.md>), [math-na](<https://devfeed.tech/tags/math-na.md>), [models](<https://devfeed.tech/tags/models.md>), [order](<https://devfeed.tech/tags/order.md>), [structure](<https://devfeed.tech/tags/structure.md>)

### AI overview

The study builds physics-structured reduced-order models for a Mach 1.92 clamped panel (RC-19) and a Mach 5.85 cantilever plate (HyMAX), despite published data lacking time-resolved responses and repeated runs. Each case has a Ritz-Galerkin reference model and a learned model that preserves its exact potential, fits remaining forces, and trains only dissipation constrained not to add energy; the energy balance holds for any weights. For RC-19, the model's mean dominant frequency across 30 runs is within 1% of the measured 236 Hz, and it reproduces periodic and chaotic response regimes in all 16 reported runs without a 113 Hz peak absent from measurements. A displacement-based force correction, tested on synthetic 5 kHz data, yields state errors of 1.25% of panel thickness without noise and 3.5% with 10 μm noise. For HyMAX, the learned model tracks reference trailing-edge displacement over 65 ms with 3.7% relative error. These reported results suggest the structural constraints help match flutter behavior under sparse data; the correction test uses synthetic data.

### Source excerpt

arXiv:2610.31504v1 Announce Type: cross Abstract: We build structure-preserving reduced-order models for both challenge problems of the AePW-4 High-Speed Working Group. They are a clamped thin panel at Mach 1.92 (RC-19) and a cantilever plate at Mach 5.85 (HyMAX). The published data for each case hold no time-resolved response and no repeated runs. The physics must therefore be carried by the form of the model. One construction gives two models per problem. A Ritz-Galerkin reference model describes the plate. A learned model keeps the reference model's exact potential, fits the remaining generalized forces, and trains only a dissipation that cannot add energy. Its energy balance holds for any weights by construction. On the trained model the worst relative residual is 2.2e-15. On RC-19, shock relations predict a pressure ratio of 1.521 across the panel. Pressure-sensitive paint measured the same value. Thirty runs of the learned model place the mean dominant frequency within 1.0% of the measured 236 Hz. At one periodic and one chaotic condition, the learned model reproduces the response regime of a published reduced-order model in all sixteen runs. It does so without that model's 113 Hz spectral peak, which the measurements do not show. Measured displacements can also correct the chaotic response through one more force that respects the energy balance. We tested this on synthetic data at the experiment's 5 kHz sampling rate. The state error is 1.25% of the panel thickness without noise and 3.5% with 10 micrometers of noise. On HyMAX the learned model follows the reference model's trailing-edge displacement for 65 ms with a relative error of 3.7%. These results show that a model whose structure carries the physics can match measured flutter behavior when the available data cannot train it.

[Next page](<https://devfeed.tech/topics/cfd.md?cursor=WyIyMDI2LTEwLTIzVDA0OjAwOjAwKzAwOjAwIiwgIjkxNmY1YmFiLTc5MzAtNGZjYy1iYWQyLWI0N2U2YTM4NGMxNyJd>)