Metraipolis › archive › dc_94f3e8a2b2bf
Latent-MoE: Domain-Aware Mixture-of-Experts for PDEs with Multi-Regime Physics — arXiv:2609.07814
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.07814
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T13:57:56.969Z
- SHA-256 of the archived text
fc8eeec7a81c8f6b0914f74eb996e912ebae0dba6dbf0fb51be95acb09b6d6bb- Stable URL of this page
https://metraipolis.com/documento/dc_94f3e8a2b2bf
How to cite
Latent-MoE: Domain-Aware Mixture-of-Experts for PDEs with Multi-Regime Physics — arXiv:2609.07814 — arXiv (en), https://arxiv.org/abs/2609.07814 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_94f3e8a2b2bf (Metraipolis archive dc_94f3e8a2b2bf, sha256 fc8eeec7a81c8f6b0914f74eb996e912ebae0dba6dbf0fb51be95acb09b6d6bb)
BibTeX
@misc{dc94f3e8a2b2bf,
title = {Latent-MoE: Domain-Aware Mixture-of-Experts for PDEs with Multi-Regime Physics — arXiv:2609.07814},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.07814},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_94f3e8a2b2bf; sha256 fc8eeec7a81c8f6b0914f74eb996e912ebae0dba6dbf0fb51be95acb09b6d6bb},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.07814. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Physics-informed neural networks (PINNs) struggle on PDEs whose governing physics varies across the domain. We trace this to a structural property of standard coordinate networks: their neural tangent kernel (NTK) is translation-variant and lets training points of large coordinate magnitude disproportionately influence predictions elsewhere, producing long-range coupling and gradient conflict during training. We show analytically and empirically that mixture-of-experts (MoE) architectures with centered, compact-support routers yield a uniformly banded NTK whose kernel-regression weights decay exponentially with distance, localizing the learning. Building on this, we propose \emph{Latent-MoE}, which interleaves domain-aware MoE blocks within a shared backbone. Unlike FB-PINNs or X-PINNs, which rigidly partition both the domain and the parameters so that the parameters on different subdomains are updated independently, Latent-MoE is designed to preserve the localization benefit of domain-aware routing while allowing capacity to flow across regions through the shared backbone. On standard homogeneous-physics benchmarks Latent-MoE is competitive with established baselines; on benchmarks with multi-stage time-variable physics, where global models and rigid domain decompositions both fall into spurious solutions, it improves over them by more than an order of magnitude, with markedly reduced gradient conflict during training.
Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-09 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG) ; Computational Physics (physics.comp-ph).
Pagina del lavoro: https://arxiv.org/abs/2609.07814 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T13:39:42.992Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).