FMint-SDE: A Multimodal Foundation Model for Accelerating Numerical Simulation of SDEs via Error Correction
SIAM Journal on Scientific Computing · 31.12.2026 · 48 (6) · C1491-C1513
DOI: 10.1137/25m1817170 ↗Kurz gefasst
Fachartikel in der Zeitschrift SIAM Journal on Scientific Computing, online erschienen am 16.11.2026. Verfasst von Jiaxin Yuan und 2 weiteren Autor:innen. Die Arbeit nennt 65 Literaturangaben. Gefördert von: U.S. Department of Energy, Air Force Office of Scientific Research, Defense Advanced Research Projects Agency.
Abstract (Originalsprache)
Fast and accurate simulation of dynamical systems is a fundamental challenge across scientific and engineering domains. Traditional numerical integrators often face a trade-off between accuracy and computational efficiency, while existing neural network–based approaches typically require training a separate model for each case. To overcome these limitations, we introduce a novel multimodal foundation model for large-scale simulations of differential equations: FMint-SDE (Foundation Model based on Initialization for stochastic differential equations). Based on a decoder-only transformer with in-context learning, FMint-SDE leverages numerical and textual modalities to learn a universal error-correction scheme. It is trained using prompted sequences of coarse solutions generated by conventional solvers, enabling broad generalization across diverse systems. We evaluate our models on a suite of challenging SDE benchmarks spanning applications in molecular dynamics, mechanical systems, finance, and biology. Experimental results show that our approach achieves a superior accuracy-efficiency tradeoff compared to classical solvers, underscoring the potential of FMint-SDE as a general-purpose simulation tool for dynamical systems. Reproducibility of computational results. This paper has been awarded the “SIAM Reproducibility Badge: Code and Data Available” as a recognition that the authors have followed reproducibility principles valued by SISC and the scientific computing community. Code and data that allow readers to reproduce the results in this paper are available at https://github.com/margotyjx/FMint-SDE and in the supplementary materials ( M181717_Supplementary_Materials_1.pdf [7.87MB], FMint-SDE-main.zip [206KB]), linked from the main article webpage. [Formula: see text]
Förderung
- U.S. Department of Energy
- Air Force Office of Scientific Research
- Defense Advanced Research Projects Agency
- Office of Naval Research
- National Science Foundation
Zitieren
APA
Yuan, J., Yang, H., & Cameron, M. (2026). FMint-SDE: A Multimodal Foundation Model for Accelerating Numerical Simulation of SDEs via Error Correction. SIAM Journal on Scientific Computing, 48(6), C1491-C1513. https://doi.org/10.1137/25m1817170
Vancouver
Yuan J, Yang H, Cameron M. FMint-SDE: A Multimodal Foundation Model for Accelerating Numerical Simulation of SDEs via Error Correction. SIAM Journal on Scientific Computing. 2026;48(6):C1491-C1513. doi:10.1137/25m1817170
BibTeX
@article{Yuan2026,
author = {Yuan, Jiaxin and Yang, Haizhao and Cameron, Maria},
title = {FMint-SDE: A Multimodal Foundation Model for Accelerating Numerical Simulation of SDEs via Error Correction},
journal = {SIAM Journal on Scientific Computing},
year = {2026},
volume = {48},
number = {6},
pages = {C1491--C1513},
doi = {10.1137/25m1817170}
}
EUVIMED