跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2506.20630 · v2

First-order methods for stochastic and finite-sum convex optimization with deterministic constraints

首次提交 2025-06-26 01:26
版本更新 2026-11-18 02:53
雷达观测 2026-11-18 12:34
arXiv ↗PDF ↗

摘要

In this paper, we study a class of stochastic and finite-sum convex optimization problems with deterministic constraints. Existing methods typically aim to find an εε-expected feasible stochastic optimalexpected\ feasible\ stochastic\ optimal solution, in which both the expected constraint violation and the expected optimality gap are within a prescribed tolerance εε. However, in many practical applications, constraints must be nearly satisfied with certainty, rendering such solutions potentially unsuitable due to the risk of substantial violations. To address this issue, we propose stochastic first-order methods for finding an εε-surely feasible stochastic optimalsurely\ feasible\ stochastic\ optimal (εε-SFSO) solution, where the constraint violation is deterministically bounded by εε and the expected optimality gap is at most εε. Our methods apply an accelerated stochastic gradient (ASG) scheme or a modified variance-reduced ASG scheme only onceonly\ once to a sequence of quadratic penalty subproblems with appropriately chosen penalty parameters. We establish first-order oracle complexity bounds for the proposed methods in computing an εε-SFSO solution. Preliminary numerical results are presented to illustrate the performance of the proposed methods.

作者备注:Accepted by Mathematics of Operations Research

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

本站首次收录为 v2,暂无更早版本可对照。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v2 · 2026-11-18 02:53 · 最新观测版本

First-order methods for stochastic and finite-sum convex optimization with deterministic constraints

In this paper, we study a class of stochastic and finite-sum convex optimization problems with deterministic constraints. Existing methods typically aim to find an εε-expected feasible stochastic optimalexpected\ feasible\ stochastic\ optimal solution, in which both the expected constraint violation and the expected optimality gap are within a prescribed tolerance εε. However, in many practical applications, constraints must be nearly satisfied with certainty, rendering such solutions potentially unsuitable due to the risk of substantial violations. To address this issue, we propose stochastic first-order methods for finding an εε-surely feasible stochastic optimalsurely\ feasible\ stochastic\ optimal (εε-SFSO) solution, where the constraint violation is deterministically bounded by εε and the expected optimality gap is at most εε. Our methods apply an accelerated stochastic gradient (ASG) scheme or a modified variance-reduced ASG scheme only onceonly\ once to a sequence of quadratic penalty subproblems with appropriately chosen penalty parameters. We establish first-order oracle complexity bounds for the proposed methods in computing an εε-SFSO solution. Preliminary numerical results are presented to illustrate the performance of the proposed methods.

查看此版本解读 →查看此版本原文 ↗