Ferramenta para simulação visual de redes de autômatos estocásticos através do cálculo de estados sucessores e predecessores

Abstract

This study aims to develop a tool to visually simulate Stochastic Automata Networks (SAN). The SAN formalism uses the PEPS tool for numerical solution, evaluating unexpected behaviors. These numerical solutions are the core for structured formalisms since they provide numerical results based on mathematical relationships. It is very interesting to complement the numerical solution with visual simulations because it adds more detailed information about models, making them more understandable for broader audiences of researchers and general users. The present dissertation describes a tool for modeling and visual simulation of SAN models, since its internal structure defines a compact storage schema (a Descriptor) for the transition matrix representing the underlying Markov Chain and uses tensor algebra to deal with the vector-descriptor multiplication. The proposed tool will allow both academic users and enthusiasts of the SAN formalism to manipulate models abstracting a deeper knowledge of this structured formalism.