LatentDDM is identified as a promising design principle for physical foundation models that improves 20-step field rollouts in fast-pitching airfoil flow, both zero-shot and after few-shot calibration.
Abstract
Neural operators are fast, differentiable surrogates for physical simulation, but their accuracy often degrades when domain geometry, size, or operating conditions differ from training. Supervised adaptation can recover accuracy, but even a small target set requires costly high-fidelity simulations. We therefore ask how pretraining and transfer can be designed together to reduce this deployment cost. LatentDDM first pretrains a neural operator to predict fields on small subdomains. For a new setting, it freezes this operator and trains only a lightweight module that composes the local predictions. We evaluate our method on two complementary problems: steady Darcy flow, where long-range pressure coupling must extend across increasingly large porous domains, and unsteady incompressible flow around a pitching airfoil, where rollout errors compound as target pitching frequencies exceed the training range. Compared with the capacity-matched models that process the full domain at once, LatentDDM's error is 36-56% lower on larger Darcy domains after adaptation with 16 target simulations. It also improves 20-step field rollouts in fast-pitching airfoil flow, both zero-shot and after few-shot calibration. These results identify the co-designed local pretraining and composition-level transfer as a promising design principle for physical foundation models.
This work finds that cross-domain pretraining improves zero- and few-shot performance on held-out datasets relative to both training from scratch and transferring from domain-specific experts, and finds that simply pooling steady-state datasets is both sufficient and effective.
Anthony Zhou, A. Barati Farimani, Shirley Ho et al.· 2 citations
Neural operators can accelerate repeated nonlinear mechanics calculations, but their accuracy can deteriorate as operating conditions move beyond the training range. This work studies whether high-fidelity solutions acquired during use can be reused to adapt a neural operator and improve subsequent mechanics-based comm...
Prashant K. Jha, K. Enakoutsa, Ian Galloway et al.· 0 citations
Neural operators are typically trained in a supervised fashion, which requires a dataset to be generated with a classical solver. Training them physics-informed, i.e., purely from the governing equations, removes this large offline cost and allows fresh samples to be drawn at every optimization step, but has so far bee...
Shi-Zheng Wen, Siddhartha Mishra, Marius Zeinhofer· arXiv.org· 1 citation
This work investigates how established uncertainty quantification approaches behave when adapted to geometry-conditioned neural surrogates and examines whether predicted uncertainties have credible magnitudes, identify locations with larger prediction errors, respond to unfamiliar inputs, and remain informative for der...
Kaustubh Tangsali, M. A. Nabian, Kelvin Lee et al.· arXiv.org· 0 citations
Plasolver, a physics-informed neural operator framework that combines the efficiency of operator learning with the accuracy and robustness of classical numerical solvers, provides an efficient, accurate, and discretization-invariant computational framework for nonlinear, path-dependent elastoplastic problems.
Yi-Zheng Wang, M. Eshaghi, Hua-Dong Zhang et al.· 0 citations
This work shows that the output Jacobian of a neural solution model trained at one condition defines a reusable response space for cross-condition solution variations, and introduces Linearized Subspace Transfer (LST) to exploit this space and recover target solutions by minimizing the target PDE-system residual over r...
Wen-Bo Cao, Wei-Wei Zhang· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduNov 17, 2026
By studying the computational foundations of language learning, Associate Professor Jacob Andreas builds intelligent systems that can communicate effectively with humans.
Microsoft Research Blog· microsoft.comNov 16, 2026
AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduNov 13, 2026