Discrete optimization for the factorization problem

The paper addresses the factorization of natural numbers into prime factors task in the context of discrete optimization and machine learning. The approach with decomposition into summands and the associated function for using with genetic algorithms (as fitness function) and neural networks (as error function) is proposed. The statistical analysis of changes in the discrete transformation function of the optimal divisor is performed in order to approximate the scope of optimal discrete transformations for a trial divisor.