English

Convex quartic problems: homogenized gradient method and preconditioning

Optimization and Control 2024-04-24 v2

Abstract

We consider a convex minimization problem for which the objective is the sum of a homogeneous polynomial of degree four and a linear term. Such task arises as a subproblem in algorithms for quadratic inverse problems with a difference-of-convex structure. We design a first-order method called Homogenized Gradient, along with an accelerated version, which enjoy fast convergence rates of respectively O(κ2/K2)\mathcal{O}(\kappa^2/K^2) and O(κ2/K4)\mathcal{O}(\kappa^2/K^4) in relative accuracy, where KK is the iteration counter. The constant κ\kappa is the quartic condition number of the problem. Then, we show that for a certain class of problems, it is possible to compute a preconditioner for which this condition number is n\sqrt{n}, where nn is the problem dimension. To establish this, we study the more general problem of finding the best quadratic approximation of an p\ell_p norm composed with a quadratic map. Our construction involves a generalization of the so-called Lewis weights.

Keywords

Cite

@article{arxiv.2306.17683,
  title  = {Convex quartic problems: homogenized gradient method and preconditioning},
  author = {Radu-Alexandru Dragomir and Yurii Nesterov},
  journal= {arXiv preprint arXiv:2306.17683},
  year   = {2024}
}

Comments

27 pages, 2 figures

R2 v1 2026-06-28T11:19:01.112Z