mlsanityData to 7 Oct 2026

Latent Generative Solvers for Generalizable Long-Term Physics Simulation

arXiv 2602.11229 ↗First posted 11 Feb 2026cs.AI cs.LGcited by 2

Not on the map. Why?

Abstract (arXiv)

Reliable physics simulation demands two capabilities that today's neural PDE solvers do not deliver together: generalization across heterogeneous PDE families, and stability under long autoregressive rollouts. Deterministic operators accumulate error geometrically, while existing probabilistic solvers are confined to a single PDE family or short horizons. We close this gap with the Latent Generative Solver (LGS), three coupled components: (i) a Physics VAE (PhyVAE) compressing twelve PDE families into a shared latent manifold; (ii) a Pyramidal Flow-Forcing Transformer (PFlowFT) that generates the next latent by flow matching, conditioned on a per-trajectory context updated on the model's own predictions; and (iii) input noising during training, for which we derive a sufficient-condition contraction bound explaining the observed long-horizon stability. Pretrained on a 2.5,M-trajectory, 16-system corpus at 1282, LGS matches the strongest deterministic baseline at one step, wins on 15/16 systems at both 5- and 10-step rollout, cuts 20-step L2RE from 56.1% to 30.2%, and uses 13--77× less recurrent dynamics-step compute. It also adapts efficiently to a 2562 Kolmogorov flow held out from the pretraining corpus, dropping 1-step L2RE from 0.398 to 0.129 in five finetune epochs against U-AFNO's 0.653→0.343. These empirical claims are limited to the evaluated 2D regular-grid systems and discrete horizons.

Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 47 of its 70 references that are in the map. First posted on arXiv 11 Feb 2026.

Builds on

47 of 70 references.

References · 47

  1. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  2. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  3. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  4. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  5. 2020Scaling Laws for Neural Language Modelscited by 6,411
  6. 2022Scalable Diffusion Models with Transformerscited by 5,931
  7. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,067
  8. 2020Fourier Neural Operator for Parametric Partial Differential Equationscited by 2,806
  9. 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,551
  10. 2019DeepONet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operatorscited by 1,965
  11. 2021Perceiver: General Perception with Iterative Attentioncited by 1,089
  12. 2024SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformerscited by 707
35 more
  1. 2021Maximum Likelihood Training of Score-Based Diffusion Modelscited by 670
  2. 2024Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusioncited by 634
  3. 2025Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusioncited by 569
  4. 2021Physics-Informed Neural Operator for Learning Partial Differential Equationscited by 516
  5. 2024Pyramidal Flow Matching for Efficient Video Generative Modelingcited by 354
  6. 2023Neural Operators for Accelerating Scientific Simulations and Designcited by 218
  7. 2023PDE-Refiner: Achieving Accurate Long Rollouts with Neural PDE Solverscited by 205
  8. 2022Towards Multi-spatiotemporal-scale Generalized PDE Modelingcited by 190
  9. 2024Poseidon: Efficient Foundation Models for PDEscited by 186
  10. 2024DPOT: Auto-Regressive Denoising Operator Transformer for Large-Scale PDE Pre-Trainingcited by 137
  11. 2024The Well: a Large-Scale Collection of Diverse Physics Simulations for Machine Learningcited by 134
  12. 2023Upscale-A-Video: Temporal-Consistent Diffusion Model for Real-World Video Super-Resolutioncited by 131
  13. 2022Why Exposure Bias Matters: An Imitation Learning Perspective of Error Accumulation in Language Generationcited by 106
  14. 2024DiffusionPDE: Generative PDE-Solving Under Partial Observationcited by 100
  15. 2023In-Context Operator Learning with Data Prompts for Differential Equation Problemscited by 98
  16. 2022Continuous PDE Dynamics Forecasting with Implicit Neural Representationscited by 71
  17. 2024Ca2-VDM: Efficient Autoregressive Video Diffusion Model with Causal Generation and Cache Sharingcited by 70
  18. 2024Progressive Autoregressive Video Diffusion Modelscited by 65
  19. 2024Towards a Foundation Model for Partial Differential Equations: Multi-Operator Learning and Extrapolationcited by 45
  20. 2024Integrating Neural Operators with Diffusion Models Improves Spectral Representation in Turbulence Modelingcited by 36
  21. 2024PROSE-FD: A Multimodal PDE Foundation Model for Learning Multiple Operators for Forecasting Fluid Dynamicscited by 33
  22. 2025Guided Diffusion Sampling on Function Spaces with Applications to PDEscited by 33
  23. 2024Unisolver: PDE-Conditional Transformers Towards Universal Neural PDE Solverscited by 31
  24. 2025PDE-Transformer: Efficient and Versatile Transformers for Physics Simulationscited by 29
  25. 2024Wavelet Diffusion Neural Operatorcited by 28
  26. 2025Generative Latent Neural PDE Solver using Flow Matchingcited by 28
  27. 2023Physics Informed Token Transformer for Solving Partial Differential Equationscited by 21
  28. 2025Recurrent Neural Operators: Stable Long-Term PDE Predictioncited by 15
  29. 2024Accelerating Phase Field Simulations Through a Hybrid Adaptive Fourier Neural Operator with U-Net Backbonecited by 10
  30. 2022GAN-DUF: Hierarchical Deep Generative Models for Design Under Free-Form Geometric Uncertaintycited by 9
  31. 2024Auto-Regressive Moving Diffusion Models for Time Series Forecastingcited by 9
  32. 2025AERIS: Argonne Earth Systems Model for Reliable and Skillful Predictionscited by 8
  33. 2025Physics-aware generative models for turbulent fluid flows through energy-consistent stochastic interpolantscited by 7
  34. 2025PI-MFM: Physics-informed multimodal foundation model for solving partial differential equationscited by 6
  35. 2025Generative Reliability-Based Design Optimization Using In-Context Learning Capabilities of Large Language Modelscited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Flow Learners for PDEs: Toward a Physics-to-Physics Paradigm for Scientific Computingcited by 4
  2. 2026Learnable composition for neural operatorscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Flow marching for a generative PDE foundation model
  2. 2024Text2PDE: Latent Diffusion Models for Accessible Physics Simulation
  3. 2026Geometry-aware Latent Autoregressive Generative Model for PDEs in Complex Domains
  4. 2023LatentPINNs: Generative physics-informed neural networks via a latent representation learning
  5. 2026Agentic Exploration of PDE Spaces using Latent Foundation Models for Parameterized Simulations
  6. 2026Variational Grey-Box Dynamics Matching
  7. 2026Stabilizing Physics-Informed Consistency Models via Structure-Preserving Training
  8. 2025Physics-Constrained Flow Matching: Sampling Generative Models with Hard Constraints