aiqbench analysis
This research introduces PCGBandit, a method that leverages online learning to adaptively optimize solver configurations for transient PDEs without requiring extensive pre-training data. By utilizing feedback from the linear solver itself via bandit algorithms, it achieves immediate speedups in fluid and magnetohydrodynamics simulations within standard open-source frameworks like OpenFOAM. This approach addresses key bottlenecks in ML-for-science by demonstrating practical efficiency gains over classical baselines while minimizing computational overhead during the learning phase.
One-shot acceleration of transient PDE solvers via online-learned preconditioners
arXiv:2509.08765v4 Announce Type: replace-cross Abstract: Data-driven acceleration of scientific computing workflows has been a high-profile aim of machine learning (ML) for science, with numerical simulation of transient partial differential equations (PDEs) being one of the main applications. The focus thus far has been on methods that require classical simulations to train, which when combined with the data-hungriness and optimization challenges of neural networks has caused difficulties in demonstrating a convincing advantage against strong classical baselines. We consider an alternative paradigm in which the learner uses a classical solver's own data to accelerate it, enabling a one-shot speedup of the simulation. Concretely, since transient PDEs often require solving a sequence of related linear systems, the feedback from repeated calls to a linear solver such as preconditioned conjugate gradient (PCG) can be used by a bandit algorithm to online-learn an adaptive sequence of solver configurations (e.g. preconditioners). The method we develop, PCGBandit, is implemented directly on top of the popular open-source software OpenFOAM, which we use to show its effectiveness on a set of fluid and magnetohydrodynamics (MHD) problems.
Abstract:Data-driven acceleration of scientific computing workflows has been a high-profile aim of machine learning (ML) for science, with numerical simulation of transient partial differential equations (PDEs) being one of the main applications. The focus thus far has been on methods that require classical simulations to train, which when combined with the data-hungriness and optimization challenges of neural networks has caused difficulties in demonstrating a convincing advantage against strong classical baselines. We consider an alternative paradigm in which the learner uses a classical solver's own data to accelerate it, enabling a one-shot speedup of the simulation. Concretely, since transient PDEs often require solving a sequence of related linear systems, the feedback from repeated calls to a linear solver such as preconditioned conjugate gradient (PCG) can be used by a bandit algorithm to online-learn an adaptive sequence of solver configurations (e.g. preconditioners). The method we develop, PCGBandit, is implemented directly on top of the popular open-source software OpenFOAM, which we use to show its effectiveness on a set of fluid and magnetohydrodynamics (MHD) problems.
Computational Physics (physics.comp-ph); Machine Learning (cs.LG); Numerical Analysis (math.NA); Machine Learning (stat.ML)
arXiv:2509.08765 [physics.comp-ph]
(or arXiv:2509.08765v4 [physics.comp-ph] for this version)
https://doi.org/10.1016/j.cpc.2026.110304 DOI(s) linking to related resources
Full text under the publisher's open licence · 198 words
Related coverage
Accelerating Reinforcement Learning via MPC Solver-Gradient Guidance for Weights-varying MPC
arXiv · Sep 2, 2026
At SIGGRAPH, NVIDIA Advances Graphics and Simulation With Agentic and Physical AI
NVIDIA · Jul 20, 2026
Inference for Newton Methods with Accelerated Sketch-and-Project via Random Scaling
arXiv · Sep 14, 2026
Applying Two-Grid Preconditioner for Subsurface Flow Simulation using Attention-enhanced Hybrid Network to Accelerate Multiscale Discretization in High-contrast Media
arXiv · Jun 3, 2026