3 Stars
Updated Last
5 Months Ago
Started In
June 2015


The DSPsolver.jl package provides an interface for using an open-source software package DSP for solving stochastic mixed-integer programming problems from the Julia language. You need to download and install DSP package.

This package requires the StructJuMP.jl and MPI.jl packages for Julia. The StructJuMP.jl package provides a scalable algebraic modeling tool for stochastic programming. The MPI.jl package enables the Julia script to run on a distributed computing system via MPI communication.


You can install DSPsolver.jl using the Julia package system.



The following example shows a Julia script that uses DSPsolver.jl for solving a two-stage stochastic integer programming problem.

using DSPsolver, StructJuMP, MPI; # Load packages

# Initialize MPI

# random parameter
xi = [[7,7] [11,11] [13,13]];

# StructJuMP.jl model scripts
m = StructuredModel(num_scenarios=3);
@variable(m, 0 <= x[i=1:2] <= 5, Int);
@objective(m, Min, -1.5*x[1]-4*x[2]);
for s = 1:3
	q = StructuredModel(parent=m, prob=1/3);
	@variable(q, y[j=1:4], Bin);
	@objective(q, Min, -16*y[1]+19*y[2]+23*y[3]+28*y[4]);
	@constraint(q, 2*y[1]+3*y[2]+4*y[3]+5*y[4]<=xi[1,s]-x[1]);
	@constraint(q, 6*y[1]+1*y[2]+3*y[3]+2*y[4]<=xi[2,s]-x[2]);

DSPsolver.loadProblem(m);       # Load model m to DSP
DSPsolver.solve(DSP_SOLVER_DD); # Solve problem using dual decomposition

# print out upper/lower bounds
println("Upper Bound: ", DSPsolver.getPrimalBound());
println("Lower Bound: ", DSPsolver.getDualBound());

# Finalize MPI