Answers are grounded in this paper and cite their sources.
We propose a geometry-aware strategy for training neural preconditioners tailored to parametrized linear systems arising from the discretization of mixed-dimensional partial differential equations (PDEs). Such systems are typically ill-conditioned due to embedded lower-dimensional structures and are solved using Krylov subspace methods. Our approach yields an approximation of the inverse operator employing a learning algorithm consisting of a two-stage training framework: an initial static pretraining phase, based on residual minimization, followed by a dynamic fine-tuning phase that incorporates solver convergence dynamics into the training process via a novel loss functional. This dynamic loss is defined by the principal angles between the residuals and the Krylov subspaces. It is evaluated using a differentiable implementation of the Flexible GMRES algorithm, which enables backpropagation through both the Arnoldi process and Givens rotations. The resulting neural preconditioner is explicitly optimized to enhance early-stage convergence and reduce iteration counts across a family of 3D–1D mixed-dimensional problems exhibiting geometric variability in the 1D domain. Numerical experiments show that our solver-aligned approach significantly improves convergence rate, robustness, and generalization.
No takes yet. Share an insight, caveat, or question.
Dimola et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: