The present software package includes scripts for solving SDP problems in general form:
min <C,X>
s.t. <Ai,X> <= bi, i=1,...,l;
<Aj,X> = bj, j=l+1,...,m;
X >= L
X psd
In particular it contains the ADMM for large-scale SDPs in general form and the post-processing procedure able to compute a valid bound on the primal optimal value detailed in the manuscript:
Federico Battista, Marianna De Santis "Dealing with inequality constraints in large scale semidefinite relaxations for graph coloring and maximum clique problems"
The MAIN code for the Alternating Direction Augmented Lagrangian along with safe bounding procedures is
- ADALplus_bounds.py
N.B. the dual bound procedure calls the LP solver of Gurobi. You need to install Gurobi in order to run it
Instances can be formulated by the python code at: https://github.com/batt95/SDP_MSSP_GCP