mlsanityData to 7 Oct 2026

UniFluids: Unified Neural Operator Learning with Conditional Flow-matching

arXiv 2603.22309 ↗First posted 19 Mar 2026cs.LG cs.AIcited by 0

On the map: Diffusion models → Flow matching, time series and science

Abstract (arXiv)

Partial differential equation (PDE) simulation holds extensive significance in scientific research. Currently, the integration of deep neural networks to learn solution operators of PDEs has introduced great potential. In this paper, we present UniFluids, a conditional flow-matching framework that harnesses the scalability of diffusion Transformer to unify learning of solution operators across diverse PDEs with varying dimensionality and physical variables. Unlike the autoregressive PDE foundation models, UniFluids adopts flow-matching to achieve parallel sequence generation, making it the first such approach for unified operator learning. Specifically, the introduction of a unified four-dimensional spatiotemporal representation for the heterogeneous PDE datasets enables joint training and conditional encoding. Furthermore, we find the effective dimension of the PDE dataset is much lower than its patch dimension. We thus employ x-prediction in the flow-matching operator learning, which is verified to significantly improve prediction accuracy. We conduct a large-scale evaluation of UniFluids on several PDE datasets covering spatial dimensions 1D, 2D and 3D. Experimental results show that UniFluids achieves strong prediction accuracy and demonstrates good scalability and cross-scenario generalization capability. The code will be released later.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Flow matching, time series and science, one of 14 directions in Diffusion models. Builds on 52 of its 67 references that are in the map. First posted on arXiv 19 Mar 2026.

Builds on

52 of 67 references.

Closest ancestors · 8

Same direction, newest first

  1. 2025Physics-Informed Distillation of Diffusion Models for PDE-Constrained Generation28 May
  2. 2025Mitigating Spectral Bias in Neural Operators via High-Frequency Scaling for Physical Systems17 Mar
  3. 2024Wavelet Diffusion Neural Operator6 Dec
  4. 2023DYffusion: A Dynamics-informed Diffusion Model for Spatiotemporal Forecasting3 Jun
  5. 2023A Denoising Diffusion Model for Fluid Field Prediction27 Jan
  6. 2022Efficient Video Prediction via Sparsely Conditioned Flow Matching26 Nov
  7. 2022Flow Matching for Generative Modeling6 Oct
  8. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow7 Sep

Other references · 44

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  2. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  3. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,082
  4. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  5. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,339
  6. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  7. 2022Scalable Diffusion Models with Transformerscited by 5,931
  8. 2022Classifier-Free Diffusion Guidancecited by 5,717
  9. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,581
  10. 2018Neural Ordinary Differential Equationscited by 3,733
  11. 2022Elucidating the Design Space of Diffusion-Based Generative Modelscited by 2,994
  12. 2020Fourier Neural Operator for Parametric Partial Differential Equationscited by 2,806
32 more
  1. 2022Video Diffusion Modelscited by 2,280
  2. 2019DeepONet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operatorscited by 1,965
  3. 2018On the Spectral Bias of Neural Networkscited by 1,321
  4. 2022FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operatorscited by 748
  5. 2022Scientific Machine Learning through Physics-Informed Neural Networks: Where we are and What's nextcited by 603
  6. 2017Solving high-dimensional partial differential equations using deep learningcited by 549
  7. 2021Learning the solution operator of parametric partial differential equations with physics-informed DeepOnetscited by 537
  8. 2021Physics-Informed Neural Operator for Learning Partial Differential Equationscited by 516
  9. 2022PDEBENCH: An Extensive Benchmark for Scientific Machine Learningcited by 396
  10. 2022Latent Video Diffusion Models for High-Fidelity Long Video Generationcited by 385
  11. 2022Message Passing Neural PDE Solverscited by 368
  12. 2021Choose a Transformer: Fourier or Galerkincited by 333
  13. 2023ClimaX: A foundation model for weather and climatecited by 325
  14. 2023GNOT: A General Neural Operator Transformer for Operator Learningcited by 321
  15. 2023Geometry-Informed Neural Operator for Large-Scale 3D PDEscited by 293
  16. 2022Transformer for Partial Differential Equations' Operator Learningcited by 270
  17. 2021Adaptive Fourier Neural Operators: Efficient Token Mixers for Transformerscited by 237
  18. 2022Towards Multi-spatiotemporal-scale Generalized PDE Modelingcited by 190
  19. 2024Poseidon: Efficient Foundation Models for PDEscited by 186
  20. 2023GenCast: Diffusion-based ensemble forecasting for medium-range weathercited by 169
  21. 2024DPOT: Auto-Regressive Denoising Operator Transformer for Large-Scale PDE Pre-Trainingcited by 137
  22. 2023Towards Foundation Models for Scientific Machine Learning: Characterizing Scaling and Transfer Behaviorcited by 128
  23. 2023In-Context Operator Learning with Data Prompts for Differential Equation Problemscited by 98
  24. 2023Multiple Physics Pretraining for Physical Surrogate Modelscited by 90
  25. 2023CFDBench: A Large-Scale Benchmark for Machine Learning Methods in Fluid Dynamicscited by 40
  26. 2023Self-Supervised Learning with Lie Symmetries for Partial Differential Equationscited by 29
  27. 2024BENO: Boundary-embedded Neural Operators for Elliptic PDEscited by 22
  28. 2023NUNO: A General Framework for Learning Parametric PDEs with Non-Uniform Datacited by 15
  29. 2025Mixture-of-Experts Operator Transformer for Large-Scale PDE Pre-Trainingcited by 10
  30. 2025CoD: A Diffusion Foundation Model for Image Compressioncited by 9
  31. 2024OmniArch: Building Foundation Model For Scientific Computingcited by 8
  32. 2025Sig-DEG for Distillation: Making Diffusion Models Faster and Lightercited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025CFO: Learning Continuous-Time PDE Dynamics via Flow-Matched Neural Operators
  2. 2025Residual-augmented flow matching operators for probabilistic partial differential equations
  3. 2024Diffusion models as probabilistic neural operators for recovering unobserved states of dynamical systems
  4. 2025Merging Memory and Space: A State Space Neural Operator
  5. 2026Learning Physical Operators using Neural Operators
  6. 2026OpInf-LLM: Parametric PDE Solving with LLMs via Operator Inference
  7. 2025Generative Latent Neural PDE Solver using Flow Matching
  8. 2025Flow marching for a generative PDE foundation model