EUVIMEDEUVIMED entdecken
Crossref

Stagnation-Shortening Inexact Newton Method Based on Unsupervised Learning for Highly Nonlinear Algebraic Problems and Applications

Yujie Gong, Li Luo, Xiao-Chuan Cai

SIAM Journal on Scientific Computing · 31.10.2026 · 48 (5) · S513-S534

DOI: 10.1137/24m1672869 ↗

Kurz gefasst

Fachartikel in der Zeitschrift SIAM Journal on Scientific Computing, online erschienen am 12.06.2026. Verfasst von Yujie Gong und 2 weiteren Autor:innen. Die Arbeit nennt 46 Literaturangaben. Gefördert von: NSFC, FDCT, MYRG.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Inexact Newton method is widely used in many scientific and engineering applications, but in some situations its convergence involves a long stagnation period in which the nonlinear residual doesn’t decrease much, and a large percentage of the total compute time is wasted. The reasons for the stagnation are quite difficult to quantify. In this paper, we propose a stagnation-shortening inexact Newton algorithm with a stagnation analysis phase during the Newton iterations in which the norms of the residuals are used as a guiding curve, and the vector values of the residuals are centralized and decomposed into a slow subspace and a regular subspace using an unsupervised learning method based on principal component analysis. A targeted quasi-Newton method is then performed in the slow space whose solution is later projected back to the global space. We show numerically that with such an embedded learning phase, the inexact Newton method converges almost quadratically. As an application, we consider the modeling of the human artery with stenosis using the hyperelasticity equation with multiple material parameters. Due to the significant difference in the material coefficients between the plaques and the healthy parts of the blood vessels, the problem is nonlinearly very difficult. Numerical experiments demonstrate that the proposed method offers significantly reduced number of nonlinear iterations and improved robustness for this rather tough hyperelasticity problem.

Förderung

  • NSFC
  • FDCT
  • MYRG

Zitieren

APA

Gong, Y., Luo, L., & Cai, X. (2026). Stagnation-Shortening Inexact Newton Method Based on Unsupervised Learning for Highly Nonlinear Algebraic Problems and Applications. SIAM Journal on Scientific Computing, 48(5), S513-S534. https://doi.org/10.1137/24m1672869

Vancouver

Gong Y, Luo L, Cai X. Stagnation-Shortening Inexact Newton Method Based on Unsupervised Learning for Highly Nonlinear Algebraic Problems and Applications. SIAM Journal on Scientific Computing. 2026;48(5):S513-S534. doi:10.1137/24m1672869

BibTeX
@article{Gong2026,
  author = {Gong, Yujie and Luo, Li and Cai, Xiao-Chuan},
  title = {Stagnation-Shortening Inexact Newton Method Based on Unsupervised Learning for Highly Nonlinear Algebraic Problems and Applications},
  journal = {SIAM Journal on Scientific Computing},
  year = {2026},
  volume = {48},
  number = {5},
  pages = {S513--S534},
  doi = {10.1137/24m1672869}
}
Kostenlos starten →