Multiple Scale Methods For Optimization Of Discretized Continuous Functions

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Richardson, Nicholas J. E., Marusenko, Noah, Friedlander, Michael P.
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912940979912704
author Richardson, Nicholas J. E.
Marusenko, Noah
Friedlander, Michael P.
author_facet Richardson, Nicholas J. E.
Marusenko, Noah
Friedlander, Michael P.
contents A multiscale optimization framework for problems over a space of Lipschitz continuous functions is developed. The method solves a coarse-grid discretization followed by linear interpolation to warm-start project gradient descent on progressively finer grids. Greedy and lazy variants are analyzed and convergence guarantees are derived that show the multiscale approach achieves provably tighter error bounds at lower computational cost than single-scale optimization. The analysis extends to any base algorithm with iterate convergence at a fixed rate. Constraint modification techniques preserve feasibility across scales. Numerical experiments on probability density estimation problems, including geological data, demonstrate speedups of an order of magnitude or better.
format Preprint
id arxiv_https___arxiv_org_abs_2512_13993
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Multiple Scale Methods For Optimization Of Discretized Continuous Functions
Richardson, Nicholas J. E.
Marusenko, Noah
Friedlander, Michael P.
Numerical Analysis
Optimization and Control
65B99, 65D15, 90C59
A multiscale optimization framework for problems over a space of Lipschitz continuous functions is developed. The method solves a coarse-grid discretization followed by linear interpolation to warm-start project gradient descent on progressively finer grids. Greedy and lazy variants are analyzed and convergence guarantees are derived that show the multiscale approach achieves provably tighter error bounds at lower computational cost than single-scale optimization. The analysis extends to any base algorithm with iterate convergence at a fixed rate. Constraint modification techniques preserve feasibility across scales. Numerical experiments on probability density estimation problems, including geological data, demonstrate speedups of an order of magnitude or better.
title Multiple Scale Methods For Optimization Of Discretized Continuous Functions
topic Numerical Analysis
Optimization and Control
65B99, 65D15, 90C59
url https://arxiv.org/abs/2512.13993