Last modified 8 years ago Last modified on 06/02/15 11:30:44

Welcome to the DSP page


DSP (Decomposition methods for Structured Programs) is an open-source software package that provides serial and parallel implementations of different decomposition methods for solving large-scale structured programs such as stochastic mixed-integer programming (SMIP) problems. DSP has currently implemented Dual Decomposition method and Benders Decomposition method for SMIP. Shared and distributed-memory implementations are provided. DSP can also solve extensive forms of SMIPs by using off-the-shelf and open-source MIP solvers (e.g., SCIP). DSP enables the user to provide problems through C code, SMPS files, and Julia (a scalable algebraic modeling package for stochastic programming, StochJuMP).