Penalty formulation for 0-1 linear programming problem: a neural network approach

A lower bound for the penalty parameter μ that ensures the equivalence between 0-1 linear programming problem and concave quadratic penalty, has been proposed by Kalantari et al. (1982, 1987). To determine the lower bound for this parameter some related problems must be solved. In this paper by considering a neural network architecture to solve the equivalent problem, a lower bound is determined easily. The idea here is to find a suitable energy function not necessary concave such that the minima of this energy corresponds exactly to the minima of initial problem. A simulation example is given to show the effectiveness of the authors´ approach