Title
Searching for non-regular neighborhood cellular automata rules applied to scheduling task and guided by a forecast dynamical behavior parameter
Searching for non-regular neighborhood cellular automata rules applied to scheduling task and guided by a forecast dynamical behavior parameter
Download via this paper's page on the MIT Press ECAL 2015 Proceedings website.
Cellular automata (CA) have been recently considered for the scheduling task. Since the problem of forecasting dynamic behavior of CA is undecidable, several parameter-based approximations have been developed to the problem. Sensitivity parameter is one of the most-studied CA forecast parameters and it has shown efficient to identify the dynamical behaviour of CA rules characterized by regular neighborhood. Here we perform an investigation about the usage of sensitivity to identify dynamical classes in cellular automata rules with a non-linear neighborhood model in a scheduling task. The results show that sensitivity can help the identification of longcycle rules, avoiding this undesirable behavior in the search for CA rules appropriate to schedule the tasks of parallel programs in a multiprocessor architecture.