Autômatos celulares com inércia
Abstract
Resumo: Desde que os autômatos celulares (AC) foram criados, vêm sendo muito utilizados em diversas áreas de conhecimento, pois são sistemas simples e de fácil implementação computacional. Em geral, AC são compostos por redes de células, onde cada célula assume um valor numérico que determina seu estado. O tempo é discreto e o valor do estado de cada célula num tempo posterior depende do valor dos estados de seus vizinhos no tempo anterior. A exata conexão entre estas quantidades é estabelecida por uma regra dinâmica específica (a regra de atualização). Existem milhares de regras distintas para AC. Em nosso trabalho utilizaremos uma regra simples, onde o estado da célula no tempo t + 1 depende da soma dos estados de seus vizinhos no tempo t. Consideramos 3 estados, sendo 2 ativos (+1, -1), que competem dinamicamente, um passivo (zero), que não influencia a regra de mudança. Definimos também um "estado interno", a inércia, que é um ingrediente novo no AC. Essa inércia (que pode variar de 0 ao número máximo de vizinhos) confere a cada célula uma resistência à mudança de seu estado. Discutimos então, no caso de um AC bidimensional, como a inércia modifica os padrões de evolução e as propriedades dinâmicas do sistema. Estudamos diferentes aspectos do problema, populações das configurações finais, tempos de convergência, dinâmica de invasão, geração de padrões espaciais, dinâmica da competição entre os estado ativos e assim por diante. De forma geral encontramos que a inércia pode alterar de forma bastante significativa a dinâmica e o comportamento médio típico de um mesmo AC. Abstract: Since cellular automata (CA) were created, they have been widely used in several areas of knowledge because they are systems that are easily and simply implemented in computers. In general, CAs are made up of cell networks in which each cell assumes a numerical value that determines its state. Time is discrete and the state of each cell in a later period depends on the state of its neighbors in the previous period. The relation between these states is established by means of a specific dynamic rule (the updating rule). There are thousands of different rules for CAs. In this dissertation we will use a simple rule by which the state of a cell in time t+1 depends on the sum of the states of its neighbors in time t. We considered three states of which two (+1,-1) are active and compete dynamically, while one is passive (zero). We also defined an "internal state", the inertia, which is an original contribution in this work. This inertia, which can vary from zero to the maximum number of neighbors, gives to each cell a resistance to changes in its state. Then, in the case of a two-dimensional CA, we discuss how this inertia changes the system’s evolutionary patterns and its dynamic properties. We study different aspects of this work, populations of the final configurations, convergence times, invasion dynamics, the generation of spatial patterns, competition dynamics between active states and so forth. Generally, we find that the inertia can change quite significantly the dynamics and typical average behavior of the same CA.
Collections
- Teses & Dissertações [10146]