Discrete Applied Mathematics Seminar by Kibaek Kim: Computational Aspect for the Dual Decomposition of Stochastic Mixed-Integer Programming

Speaker: Kibaek Kim, computational mathematician, Argonne National Lab

Title: Computational Aspect for the Dual Decomposition of Stochastic Mixed-Integer Programming

Abstract: This talk presents Lagrangian dual decomposition method and its algorithmic approaches that have been developed for solving stochastic mixed-integer programming problems with a finite number of scenarios. The dual decomposition is well-suited for distributed/parallel computing and poses several interesting challenges. In this tutorial, we first discuss the problem structures for which dual decomposition may be preferred to the well-known Benders decomposition. Second, we present the recent development that addresses the computational challenges in the dual decomposition. Last, we present future challenges and opportunities in the dual decomposition and parallel computing.

 

Discrete Applied Math Seminar

Request link