EUVIMEDEUVIMED entdecken
Crossref

Beyond the Kolmogorov barrier: a learnable weighted hybrid autoencoder for model order reduction

Nithin Somasekharan, Shaowu Pan

Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences · 2025-08 · 481 (2320)

DOI: 10.1098/rspa.2025.0159 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, online erschienen am 27.08.2025. Verfasst von Nithin Somasekharan und 1 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 38 Literaturangaben. Gefördert von: U.S. Department of Energy under Advancements in Artificial Intelligence for Science.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Representation learning for high-dimensional, complex physical systems aims to identify a low-dimensional intrinsic latent space, which is crucial for reduced-order modelling and modal analysis. To overcome the well-known Kolmogorov barrier, deep autoencoders (AEs) have been introduced in recent years, but they often suffer from poor convergence behaviour as the rank of the latent space increases. To address this issue, we propose the learnable weighted hybrid AE , a hybrid approach that combines the strengths of singular value decomposition (SVD) with deep AEs through a learnable weighted framework. We find that the introduction of learnable weighting parameters is essential—without them, the resulting model would either collapse into a standard proper orthogonal decomposition or fail to exhibit the desired convergence behaviour. Interestingly, we empirically find that our trained model has a sharpness thousands of times smaller compared to other models, which in turn enhances its robustness to input noise. Our experiments on classical chaotic PDE systems, including the one-dimensional Kuramoto–Sivashinsky and forced isotropic turbulence datasets, demonstrate that our approach significantly improves generalization performance compared to several competing frameworks. In addition, when combined with time series modelling techniques (e.g. Koopman operator, long-short-term memory), the proposed technique offers significant improvements for surrogate modelling of high-dimensional multiscale PDE systems.

Förderung

  • U.S. Department of Energy under Advancements in Artificial Intelligence for Science

Zitieren

APA

Somasekharan, N., & Pan, S. (2025). Beyond the Kolmogorov barrier: a learnable weighted hybrid autoencoder for model order reduction. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 481(2320). https://doi.org/10.1098/rspa.2025.0159

Vancouver

Somasekharan N, Pan S. Beyond the Kolmogorov barrier: a learnable weighted hybrid autoencoder for model order reduction. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences. 2025;481(2320). doi:10.1098/rspa.2025.0159

BibTeX
@article{Somasekharan2025,
  author = {Somasekharan, Nithin and Pan, Shaowu},
  title = {Beyond the Kolmogorov barrier: a learnable weighted hybrid autoencoder for model order reduction},
  journal = {Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences},
  year = {2025},
  volume = {481},
  number = {2320},
  doi = {10.1098/rspa.2025.0159}
}
Kostenlos starten →