Skip to content
Preprint

Establishing Boundary KKT Convergence of Mirror Descent through Reparameterization

Aug 2026 · 0 citations · 39 references
Mathematics Computer Science

Abstract

Sequence convergence to a boundary Karush--Kuhn--Tucker (KKT) point has long remained unclear for nonconvex mirror descent with Legendre kernels. The difficulty arises from the blow-up of the gradient of the Legendre kernel at the boundary. Recent work~\cite{dingtoh2026nonkkt} shows that mirror descent can accumulate at non-KKT boundary points despite decreasing objective values, precluding a convergence guarantee to KKT points in general. Despite this negative result, mirror descent remains effective in many real applications. Motivated by this contrast, we address the boundary difficulty directly and establish KKT convergence of mirror descent for a broad class of structured nonconvex problems. We analyze mirror descent in reparameterized variables, where the Hessian metric is flattened and remains nondegenerate as the boundary is approached. Under extension and definability conditions jointly coupling the objective, the Legendre kernel, and the feasible region, the reparameterized sequence has finite length and converges, thereby recovering convergence to a KKT point of the original sequence. Our general framework applies to some concrete instances: Shannon entropy, Fermi--Dirac entropy, and power kernels on polyhedron.

View source

Similar papers

#machine learning Preprint Aug 2026

Non-KKT Accumulation in Entropic Mirror Descent

For mirror descent generated by a Legendre kernel, perhaps one of the most basic question in optimization is this: must every accumulation point of a bounded mirror descent sequence be Karush--Kuhn--Tucker (KKT) stationary under proper stepsizes? We show that the answer is no. A longstanding obstacle to resolving this question is the boundary blow-up of the Legendre gradient: it keeps every mirror step in the interior, while at a boundary limit, the inverse entropy metric vanishes on active coordinates and can erase the dual-feasibility in the KKT system. We construct $C^\infty$ objectives and bounded sequences generated by the Shannon-entropic mirror descent on the nonnegative orthant $\R_+^n$, for every $n\geq 3$, and on the probability simplex $\Delta_n$, for every $n\geq 4$, such that, in each case, the set of accumulation points is a smooth boundary circle containing a nonempty relatively open arc of non-KKT points. The steps satisfy $\alpha_k\asymp k^{-\beta}$ with $\beta\in(1/2,1)$, the objective values are nonincreasing, and the objectives are entropy-relatively smooth. Hence the pathology stems from the degeneracy of the Bregman geometry at the boundary, rather than from failure of descent, or improper stepsizes. To the best of our knowledge, these provide the first counterexamples to KKT accumulation for bounded mirror descent sequences with nonincreasing objective values.

Kuangyu Ding, Kim-Chuan Toh · 1 citation
Jun 2026

Second-Order KKT Guarantees for Bregman ADMM in Nonconvex and Non-Lipschitz Optimization

It is shown that on an invariant open state-space domain, one iteration of Bregman ADMM defines a smooth primal--dual fixed-point map whose strict-saddle KKT points are unstable fixed points; consequently, from random initialization the iterates converge to a strict saddle with probability zero, which yields almost-sure second-order stationarity of limiting KKT points.

S. Li, Zhihui Zhu, Qiuwei Li · 0 citations
Preprint Aug 2026

Mirror Polyak and a Primal-Dual Lifting

This work revisits a variant of the Polyak step-size based on Bregman projections due to Kiwiel (1997), and shows that mirror Polyak enjoys guarantees similar to its Euclidean counterpart, automatically adapting to relative notions of smoothness, Lipschitz continuity, or strong convexity.

Frederik Kunstner, Ryan D'Orazio, V. S. Portella et al. · 0 citations
Preprint Jul 2026

Fully Convergent Projection-based Methods with Momentum under Nonconvex Geometric

This work can specifically ensure, without any smoothness assumptions, convergence to Mordukhovich stationarity as long as the base directions asymptotically revert to the negative gradient for small stepsizes.

Matteo Lapucci, Diego Scuppa · 0 citations
Preprint Aug 2026

On the Iterate Convergence of Bregman Projected Gradient Method

A novel convergence analysis framework for the BPGM with the Shannon entropy kernel is developed, yielding strong convergence results for a broad class of objective functions under linear constraints.

He Chen, Anthony Man-Cho So · 1 citation
Preprint Jul 2026

Semismooth Newton methods for degenerate polyhedral projection

In this paper, we study dual semismooth Newton (SSN) methods for degenerate polyhedral projection problems, where generalized Jacobians of the dual residual may remain singular even arbitrarily close to the solution set. Rather than regularizing these singular systems, we exploit the nonuniqueness of the dual representation. We introduce a primal--dual lifted projection-equivalent set that always possesses extreme points without additional structural assumptions on the polyhedron, and show that its extreme-point geometry identifies dual representatives at which nonsingular generalized Jacobians of the dual residual can be constructed. This geometry is further linked to a full-column-rank condition and a generalized weak strict Robinson constraint qualification, showing that the regularity required by the Newton step can be recovered rather than imposed \emph{a priori}. We also establish displacement bounds that connect representative selection throughout the algorithm with the local Newton mechanism. Building on this variational framework, we develop an inexact dual SSN method with local superlinear convergence and a globalized version combining monotone representative selection with a Wolfe line search. The resulting method is globally convergent and eventually recovers the fast local rate. Numerical experiments on regularized optimal transport, battery-scheduling feasibility restoration, and occupation-measure projection demonstrate its robustness in highly degenerate settings.

Chao Ding, Fuxiaoyue Feng, Xudong Li · 0 citations