Estudo da variação da entropia de Shannon no jogo de xadrez a partir dos conceitos de teoria da informação

Abstract

Information Theory is a branch of science that quantifies information and its communication processes, and has at its core the reformulation of the fundamental concept of entropy. Shannon's entropy reflects the measure of uncertainty in a communication system during the transmission of information. For the scope of this work, we investigated the applicability of Shannon's Entropy over the course of numerous games of the game of chess using modern computer engines, specifically the Stockfish program, and computer code analysis. Stockfish is the most widespread computer chess engine today because it is an open-source program that allows its resources to be used by a wide range of users, be they professional players, amateurs or computer researchers. The program evaluates positions on the board based on probabilistic calculations that maximize the final result. In this way, entropic variation was computed in scatter plots, linear regression and fitting curves using computer codes. This allowed for a pertinent discussion about the transmission of information and a comparative analysis with the graphs taken from the literature and obtained empirically, thus fostering a timely argument about unpredictability in the context of the game of sixty-four squares