Activity: Talk or presentation › Public or invited talk
Description
This talk focuses on Sparse Convex Optimization (SCO) problems in distributed computing environments where several computing nodes collaborate to solve optimization problems with a sparsity constraint. Inspired by algorithmic improvements in mixed-integer optimization and the availability of high-performance computing environments, a distributed Mixed- nteger Programming (MIP) framework is introduced which includes several distributed algorithms and heuristics implemented in the Sparse Convex Optimization Toolkit (SCOT). The proposed algorithms, such as the Relaxed Hybrid Alternating Direction Method of Multipliers (RH-ADMM), extend existing methods, leading to the design of Distributed Primal Outer Approximation (DiPOA) and Distributed Hybrid Outer Approximation (DiHOA) algorithms.