James P. Crutchfield, Santa Fe Institute
I will review an approach to detecting and analyzing how natural systems compute: How do they store historical information? In what architecture is that information stored? And, how is the stored information processed? This is a complementary approach to engineering new kinds of computation (DNA, quantum, and so on) from a given set of elementary processing units. It will be helpful in finding natural forms of computation and so provide novel information processing substrates on which we can base future computers.
Examples will be drawn from hidden Markov models---often used in speech recognition and bioinformatics---and from cellular automata---often used in high-performance fluid mechanics simulations. The approach blends methods from engineering (information and computation theories), statistical physics, and mathematics (dynamical systems and symbolic dynamics).
Return to Program