Astral Space: Convex Analysis at Infinity

TL;DR

This paper introduces astral space, a minimal compact extension of Euclidean space, to analyze convex functions at infinity, enabling a better understanding of their minimizers and properties beyond finite points.

Contribution

It develops a new theoretical framework for convex analysis at infinity using astral space, extending concepts like convexity and subdifferentials beyond finite minimizers.

Findings

Astral space is a minimal compact extension of ^n including points at infinity.

Convex functions on astral space have well-characterized minimizers and continuity properties.

The framework supports convergence analysis of descent algorithms at infinity.

Abstract

Not all convex functions on $\mathbb{R}^n$ have finite minimizers; some can only be minimized by a sequence as it heads to infinity. In this work, we aim to develop a theory for understanding such minimizers at infinity. We study astral space, a compact extension of $\mathbb{R}^n$ to which such points at infinity have been added. Astral space is constructed to be as small as possible while still ensuring that all linear functions can be continuously extended to the new space. Although astral space includes all of $\mathbb{R}^n$, it is not a vector space, nor even a metric space. However, it is sufficiently well-structured to allow useful and meaningful extensions of concepts of convexity, conjugacy, and subdifferentials. We develop these concepts and analyze various properties of convex functions on astral space, including the detailed structure of their minimizers, exact characterizations of continuity, and convergence of descent algorithms.