Construction of parallel software agents’s behaviour algorithms by methods of genetic programming

The way of parallel algorithm’s construction on the basis of matrix intermediate structure with the use of the description of algorithm in the form of treelike structure is offered. Operations over the treelike structures are entered, allowing to produce parallel algorithms by methods of genetic programming. Logic levels on which evolution of algorithms of software agent’s behaviors proceeds are described and considered.