# Finding Euclidean Distance to a Convex Cone Generated by a Large Number of Discrete Points

## TL;DR

This paper introduces a Frank-Wolfe based algorithm to efficiently compute the Euclidean distance to a convex cone generated by discrete points, with applications in various fields like machine learning and manufacturing.

## Contribution

It presents a novel algorithm for high-dimensional distance computation to convex cones generated by discrete points, including convergence analysis and practical effectiveness.

## Key findings

- Effective algorithm for high-dimensional problems
- Convergence properties established
- Numerical experiments demonstrate success

## Abstract

In this paper, we study the problem of finding the Euclidean distance to a convex cone generated by a set of discrete points in $\mathbb{R}^n_+$. In particular, we are interested in problems where the discrete points are the set of feasible solutions of some binary linear programming constraints. This problem has applications in manufacturing, machine learning, clustering, pattern recognition, and statistics. Our problem is a high-dimensional constrained optimization problem. We propose a Frank-Wolfe based algorithm to solve this non-convex optimization problem with a convex-noncompact feasible set. Our approach consists of two major steps: presenting an equivalent convex optimization problem with a non-compact domain, and finding a compact-convex set that includes the iterates of the algorithm. We discuss the convergence property of the proposed approach. Our numerical work shows the effectiveness of this approach.