A smoothing method for zero--one constrained extremum problems (Q637567)

The zero-one constrained extremum problem has received a lot of attention due to its difficulty and its numerous practical applications in production planning, telecommunications, cluster analysis, etc. In this paper, the authors consider the zero-one constrained extremum problem that they reformulate as a continuous optimization problem by transforming it into a mathematical program with complementarity constraints and then into an ordinary nonlinear programming problem with the help of the Fischer-Burmeister function. The method proposed within the paper has the main advantage that obtains an equivalent differentiable ordinary nonlinear programming problem due to the strict complementary condition. In addition the aggregate function method is used in order to simplify the computations during which the whole original constraints are reformulated and approximated by only one single smoothing equality constraint with the accuracy of the approximation controlled by one parameter. In this way the approach proposed by the authors is easily programmable and practical.