An Effective Increase of the Basin of a Quadratic Functional Global Minimum

AbstractA problem of a quadratic functional minimization in configurational space is considered. To effectively increase the basin of a deep minimum it is suggested to exponentiate the matrix of quadratic functional to some power and to resolve the problem on a new functional constructed on the new matrix. By the example of matrixes of 2-dimensional Ising spin-glass model it is shown that suggested approach leads to a shift of minima spectrum towards a region of deeper minima, sharply reduces the number of found small minima, and allows finding a global minimum with a considerably greater probability.