Introduction

SCOT is an open-source distributed solver for solving Sparse Convex Optimization (SCO) problems over computational networks. SCOT adopts a mixed-integer approach to find exact solutions to SCO problems and combines various techniques to transform the original SCO problem into an equivalent convex Mixed-Integer Nonlinear Programming (MINLP) problem that can benefit from high-performance and parallel computing platforms. SCOT main algorithm, DiHOA, builds upon the LP/NLPbased branch-and-bound and is tailored for this specific problem structure. The DiHOA algorithm combines the so-called single- and multi-tree outer approximation, naturally integrates a decentralized algorithm for distributed convex nonlinear subproblems, and employs enhancement techniques such as quadratic cuts.