Gulayeva N. Local Algorithms and Their Effective Implementation in the Multiprocessor Computing Systems

This thesis is devoted to the model of local computations. It is shown that class of local algorithms proposed by Iu.I.Zhuravlev, class of local algorithms proposed by S.V.Iablonskii and A.Iu.Chernov, classes of local predicative and computational algorithms proposed by V.A.Ievstigne'ev are effectively translated to class of local algorithms proposed by A.V.Anisimov, hereinafter A-local algorithms. It is shown that class of finite-state automaton programs, class of classical and extended Petri Nets, class of genetic algorithms implemented via diffusion model, class of artificial neural networks and class of Turing machines are effectively translated to class of A-local algorithms. New A-local algorithms to solve practically important discrete problems are proposed. Effective ways of local algorithms implementation in the multiprocessor computing systems are given. In particular, A-local algorithms effective implementation methods in the networks of processor elements where the number of input/output ports per processor is equal to 4, in the homogeneous computing environment and with the help of artificial neural networks are given.