Abstract: In this work, a convergence lemma for function $f$ being finite compositions of analytic mappings and the maximum operator is proved. The lemma shows that the set of $\delta$-stationary points near an isolated local minimum point $x^*$ is shrinking to $x^*$ as $\delta\to 0$. It is a natural extension of the version for strongly convex $C^1$ functions. However, the correctness of the lemma is subtle. Analytic mappings are necessary for the lemma in the sense that replacing it with differentiable or $C^\infty$ mappings makes the lemma false. The proof is based on stratification theorems of semi-analytic sets by {\L}ojasiewicz. An extension of this proof presents a geometric characterization of the set of stationary points of $f$. Finally, a notion of stability on stationary points, called convergence stability, is proposed. It asks, under small numerical errors, whether a reasonable convergent optimization method started near a stationary point should eventually converge to the same stationary point. The concept of convergence stability becomes nontrivial qualitatively only when the objective function is both nonsmooth and nonconvex. Via the convergence lemma, an intuitive equivalent condition for convergence stability of $f$ is proved. These results together provide a new geometric perspective to study the problem of "where-to-converge" in nonsmooth nonconvex optimization.

A bridge connecting convex analysis and complex analysis and $L^2$-estimate of $d$ and $\bar\partial$

$\bar\partial$ Poincaré inequality and an improved $L^2$-estimate of $\bar\partial$ on bounded strictly pseudoconvex domains

Stable and Total Fenchel Duality for Convex Optimization Problems in Locally Convex Spaces

Stable and Total Fenchel Duality for DC Optimization Problems in Locally Convex Spaces.

On the ∂—-Equation with L 2 Estimates on Singular Complex Spaces

Asymptotic Closure Condition and Fenchel Duality for DC Optimization Problems in Locally Convex Spaces

$L^2$ estimates and existence theorems for the $\overline{\partial}$ operators in infinite dimensions, II

Convergence Analysis of a Method for Variational Inclusions

A Converse to the Skoda $L^2$ Division Theorem

Convergence Theorems and Asymptotic Integration for Functional Differential Equations

On supnorm estimates for $\bar\partial$ on infinite type convex domains in $\mathbb C^2$

The Convexity Estimates For The Solutions Of Two Elliptic Equations

Constraint Qualifications for Extended Farkas'S Lemmas and Lagrangian Dualities in Convex Infinite Programming

Majorizing Functions and Convergence of the Gauss-Newton Method for Convex Composite Optimization

Convergence analysis of the Gauss–Newton method for convex inclusion and convex-composite optimization problems

L-2-Estimates on Weakly Q-Convex Domains

A Fenchel-Moreau theorem for $\bar L^0$-valued functions

A Fenchel-Moreau Theorem for (L)over-bar(0)-Valued Functions

Positivity of Riemannian metric and the existence theorem of $L^2$ estimates for $d$ operator

Couplings, gradient estimates and logarithmic Sobolev inequality for Langevin bridges

On Convergence Lemma and Convergence Stability for Piecewise Analytic Functions