EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Neural preconditioning via Krylov subspace geometry

Nunzio Dimola, Alessandro Coclite, Paolo Zunino

Bollettino dell'Unione Matematica Italiana · 2026-12 · 19 (4) · 895-917

DOI: 10.1007/s40574-025-00522-2 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Bollettino dell'Unione Matematica Italiana, online erschienen am 20.12.2025. Verfasst von Nunzio Dimola und 2 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Politecnico di Bari.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

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.

Förderung

  • Politecnico di Bari

Zitieren

APA

Dimola, N., Coclite, A., & Zunino, P. (2026). Neural preconditioning via Krylov subspace geometry. Bollettino dell'Unione Matematica Italiana, 19(4), 895-917. https://doi.org/10.1007/s40574-025-00522-2

Vancouver

Dimola N, Coclite A, Zunino P. Neural preconditioning via Krylov subspace geometry. Bollettino dell'Unione Matematica Italiana. 2026;19(4):895-917. doi:10.1007/s40574-025-00522-2

BibTeX
@article{Dimola2026,
  author = {Dimola, Nunzio and Coclite, Alessandro and Zunino, Paolo},
  title = {Neural preconditioning via Krylov subspace geometry},
  journal = {Bollettino dell'Unione Matematica Italiana},
  year = {2026},
  volume = {19},
  number = {4},
  pages = {895--917},
  doi = {10.1007/s40574-025-00522-2}
}
Kostenlos starten →