Rogério Zanon is a graduate student while also working as a software developer in Brazil. Having graduated in mathematics from the Universidade de Sao Paulo, Rogério Miranda Zanon is pursuing a master’s degree at the Universidade Presbiteriana Mackenzie. There he is working on genetic algorithms and cellular automata. His thesis is about using genetic algorithms to solve computations, and his first problem is the density classification task. Rogério is also project manager at the company Vip Systems Informática, which develops software for access control and shopping.
Project: Non-elementary Neighborhood Behavior in the Density Classification Task
In this work we have a different focus on the density classification task in elementary rule space. We make the cellular automata evolution using a non-uniform neighborhood configuration. The 256 rules of elementary space were tested comparing their performance between elementary and non-elementary neighborhoods. The non-elementary uniform neighborhoods do not have relevant differences, but we found 14 rules that measure relevant differences, in non-elementary non-uniform neighborhoods, compared with elementary neighborhoods, about twice as many. We searched for the best neighborhood configuration in all the possibilities for rule 232 in grid sizes 41, 51, 71 and 101. In each grid the best neighborhood configurations were non-uniform non-elementary with high performance, under 90%. The experiments consisted of 500 initial conditions, periodic boundary, and 150 steps for each evolution.
Favorite Four-Color, Nearest-Neighbor, Totalistic Rule
Rule chosen: 123245
I searched the cellular automata that contain the four states in the last step. My choice is shown using the colors of my country (green, yellow, blue, and white).