Fast convergence of inertial dynamics with Hessian-driven damping under geometry assumptions
hal.structure.identifier | Institut de Mathématiques de Bordeaux [IMB] | |
dc.contributor.author | AUJOL, Jean-François | |
hal.structure.identifier | Institut National des Sciences Appliquées - Toulouse [INSA Toulouse] | |
dc.contributor.author | DOSSAL, Charles | |
hal.structure.identifier | Institut National des Sciences Appliquées - Toulouse [INSA Toulouse] | |
dc.contributor.author | HOÀNG, Văn Hào | |
hal.structure.identifier | Institut National des Sciences Appliquées - Toulouse [INSA Toulouse] | |
dc.contributor.author | LABARRIÈRE, Hippolyte | |
hal.structure.identifier | Institut National des Sciences Appliquées - Toulouse [INSA Toulouse] | |
hal.structure.identifier | Équipe Recherche Opérationnelle, Optimisation Combinatoire et Contraintes [LAAS-ROC] | |
dc.contributor.author | RONDEPIERRE, Aude | |
dc.date.accessioned | 2024-04-04T02:41:11Z | |
dc.date.available | 2024-04-04T02:41:11Z | |
dc.identifier.uri | https://oskar-bordeaux.fr/handle/20.500.12278/191128 | |
dc.description.abstractEn | First-order optimization algorithms can be considered as a discretization of ordinary differential equations (ODEs). In this perspective, studying the properties of the corresponding trajectories may lead to convergence results which can be transfered to the numerical scheme. In this paper we analyse the following ODE introduced by Attouch et al.: ∀t ⩾ t0, ẍ(t) + α t ẋ(t) + βHF (x(t)) ẋ(t) + ∇F (x(t)) = 0, where α > 0, β > 0 and HF denotes the Hessian of F. This ODE can be derived to build numerical schemes which do not require F to be twice differentiable as shown by Attouch et al. We provide strong convergence results on the error F (x(t)) − F * and integrability properties on ∥∇F (x(t))∥ under some geometry assumptions on F such as quadratic growth around the set of minimizers. In particular, we show that the decay rate of the error for a strongly convex function is O(t −α−ε) for any ε > 0. These results are briefly illustrated at the end of the paper. | |
dc.description.sponsorship | Mathématiques de l'optimisation déterministe et stochastique liées à l'apprentissage profond - ANR-19-CE23-0017 | |
dc.language.iso | en | |
dc.subject.en | Convex optimization | |
dc.subject.en | Hessian-driven damping | |
dc.subject.en | Lyapunov analysis | |
dc.subject.en | Lojasiewicz property | |
dc.subject.en | ODEs | |
dc.title.en | Fast convergence of inertial dynamics with Hessian-driven damping under geometry assumptions | |
dc.type | Document de travail - Pré-publication | |
dc.subject.hal | Mathématiques [math]/Optimisation et contrôle [math.OC] | |
dc.identifier.arxiv | 2206.06853 | |
bordeaux.hal.laboratories | Institut de Mathématiques de Bordeaux (IMB) - UMR 5251 | * |
bordeaux.institution | Université de Bordeaux | |
bordeaux.institution | Bordeaux INP | |
bordeaux.institution | CNRS | |
hal.identifier | hal-03693218 | |
hal.version | 1 | |
hal.origin.link | https://hal.archives-ouvertes.fr//hal-03693218v1 | |
bordeaux.COinS | ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.au=AUJOL,%20Jean-Fran%C3%A7ois&DOSSAL,%20Charles&HO%C3%80NG,%20V%C4%83n%20H%C3%A0o&LABARRI%C3%88RE,%20Hippolyte&RONDEPIERRE,%20Aude&rft.genre=preprint |
Files in this item
Files | Size | Format | View |
---|---|---|---|
There are no files associated with this item. |