Newton and interior-point methods for (constrained) nonconvex-nonconcave minmax optimization with stability and instability guarantees

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Chinchilla, Raphael, Yang, Guosong, Hespanha, Joao P.
Natura: Preprint
Pubblicazione: 2022
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909100828262400
author Chinchilla, Raphael
Yang, Guosong
Hespanha, Joao P.
author_facet Chinchilla, Raphael
Yang, Guosong
Hespanha, Joao P.
contents We address the problem of finding a local solution to a nonconvex-nonconcave minmax optimization using Newton type methods, including interior-point ones. We modify the Hessian matrix of these methods such that, at each step, the modified Newton update direction can be seen as the solution to a quadratic program that locally approximates the minmax problem. Moreover, we show that by selecting the modification in an appropriate way, the only stable equilibrium points of the algorithm's iterations are local minmax points. As a consequence, the algorithm can only converge towards an equilibrium point if such point is a local minmax, and it will escape if the point is not a local minmax. Using numerical examples, we show that the computation time of our algorithm scales roughly linearly with the number of nonzero elements in the Hessian.
format Preprint
id arxiv_https___arxiv_org_abs_2205_08038
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Newton and interior-point methods for (constrained) nonconvex-nonconcave minmax optimization with stability and instability guarantees
Chinchilla, Raphael
Yang, Guosong
Hespanha, Joao P.
Optimization and Control
Systems and Control
We address the problem of finding a local solution to a nonconvex-nonconcave minmax optimization using Newton type methods, including interior-point ones. We modify the Hessian matrix of these methods such that, at each step, the modified Newton update direction can be seen as the solution to a quadratic program that locally approximates the minmax problem. Moreover, we show that by selecting the modification in an appropriate way, the only stable equilibrium points of the algorithm's iterations are local minmax points. As a consequence, the algorithm can only converge towards an equilibrium point if such point is a local minmax, and it will escape if the point is not a local minmax. Using numerical examples, we show that the computation time of our algorithm scales roughly linearly with the number of nonzero elements in the Hessian.
title Newton and interior-point methods for (constrained) nonconvex-nonconcave minmax optimization with stability and instability guarantees
topic Optimization and Control
Systems and Control
url https://arxiv.org/abs/2205.08038