On the Set of Possible Minimizers of a Sum of Convex Functions

Moslem Zamani,François Glineur,Julien M. Hendrickx
2024-03-09
Abstract:Consider a sum of convex functions, where the only information known about each individual summand is the location of a minimizer. In this work, we give an exact characterization of the set of possible minimizers of the sum. Our results cover several types of assumptions on the summands, such as smoothness or strong convexity. Our main tool is the use of necessary and sufficient conditions for interpolating the considered function classes, which leads to shorter and more direct proofs in comparison with previous work. We also address the setting where each summand minimizer is assumed to lie in a unit ball, and prove a tight bound on the norm of any minimizer of the sum.
Optimization and Control
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to determine the set of all possible minimum points of the sum of convex functions when only the positions of the minimum values of each individual convex function term are known. Specifically, for an optimization problem of the form \( \min_{x \in \mathbb{R}^n} f(x) := \sum_{i \in [m]} f_i(x) \), where \( f_i: \mathbb{R}^n \to (-\infty, \infty] \) are convex functions, and the minimum point \( x_i^\star \) of each \( f_i \) is known, that is, \( x_i^\star \in \arg\min_{x \in \mathbb{R}^n} f_i(x) \), the goal of the paper is to use these known minimum points and some additional properties of the functions (such as smoothness or strong convexity) to accurately describe the set of all possible minimum points of \( f(x) \). The main contributions of the paper include: 1. **Exact Characterization**: Give the exact set of all possible minimum points when each convex function has smoothness and strong convexity. 2. **Geometric Method**: Provide a more direct and algebraic method to deal with these problems using interpolation conditions. 3. **Extension to Multiple Functions**: Extend the results to cases involving an arbitrary number of functions and consider the situation where one of the functions is not smooth. 4. **Minimum Points within the Sphere**: Under the assumption that each minimum point is located within the unit sphere, provide a tight bound on the norm of the minimum points. Through this work, the paper provides a theoretical basis for understanding and analyzing related problems in distributed optimization, elastic distributed optimization, and federated learning.