An evaluative algorithm for evolutionary-estimated per-partes-programming the Turing machine

Abstract

This paper deals with description of designed evaluative algorithm which is intended for usage inside cost function of evolutionary algorithms. This algorithm enables evolutionary-estimated per-partes-programming the Turing machine for solving complex tasks and offers flexible way of setting parameters for the programming process. The parameter setting is a key feature of designed algorithm. Parameters of the algorithm have the form of weights which influence the rate of deterministic/stochastic behavior of the algorithm. By changing the weights there can be achieved entirely different results for solving the same problem. The second key feature is an ability to ensure correct programming of the Turing machine which process complex tasks.