English

No quantum speedup over gradient descent for non-smooth convex optimization

Data Structures and Algorithms 2020-10-06 v1 Optimization and Control Quantum Physics

Abstract

We study the first-order convex optimization problem, where we have black-box access to a (not necessarily smooth) function f:RnRf:\mathbb{R}^n \to \mathbb{R} and its (sub)gradient. Our goal is to find an ϵ\epsilon-approximate minimum of ff starting from a point that is distance at most RR from the true minimum. If ff is GG-Lipschitz, then the classic gradient descent algorithm solves this problem with O((GR/ϵ)2)O((GR/\epsilon)^{2}) queries. Importantly, the number of queries is independent of the dimension nn and gradient descent is optimal in this regard: No deterministic or randomized algorithm can achieve better complexity that is still independent of the dimension nn. In this paper we reprove the randomized lower bound of Ω((GR/ϵ)2)\Omega((GR/\epsilon)^{2}) using a simpler argument than previous lower bounds. We then show that although the function family used in the lower bound is hard for randomized algorithms, it can be solved using O(GR/ϵ)O(GR/\epsilon) quantum queries. We then show an improved lower bound against quantum algorithms using a different set of instances and establish our main result that in general even quantum algorithms need Ω((GR/ϵ)2)\Omega((GR/\epsilon)^2) queries to solve the problem. Hence there is no quantum speedup over gradient descent for black-box first-order convex optimization without further assumptions on the function family.

Keywords

Cite

@article{arxiv.2010.01801,
  title  = {No quantum speedup over gradient descent for non-smooth convex optimization},
  author = {Ankit Garg and Robin Kothari and Praneeth Netrapalli and Suhail Sherif},
  journal= {arXiv preprint arXiv:2010.01801},
  year   = {2020}
}

Comments

25 pages

R2 v1 2026-06-23T19:01:51.456Z