Maximum independent set (stable set) problem: A mathematical programming model with valid inequalities; Computational testing with binary search and alternate optimal basic solutions (extreme points)

This paper deals with the maximum independent set (M.I.S.) problem, also known as the stable set problem. The basic mathematical programming model that captures this problem is an Integer Program (I.P.) with zero-one variables and only the edge inequalities. We present an enhanced model by adding a polynomial number of linear constraints, known as valid inequalities; this new model is still polynomial in the number of vertices in the graph. We carried out computational testing of the Linear Relaxation of the new Integer Program. We tested about 7000 instances of randomly generated (and connected) graphs with up to 64 vertices (as well as all 64, 128, and 256-vertex instances at the "challenge" website OEIS.org). In each of these instances, the Linear Relaxation returned an optimal solution with (i) every variable having an integer value, and (ii) the optimal solution value of the Linear Relaxation was the same as that of the original (basic) Integer Program. Our computational experience has been that a binary search on the objective function value is a powerful tool which yields a (weakly) polynomial algorithm.