Title: Analysis problems for sequential dynamical systems and communicating state machines

A simple sequential dynamical system (SDS) is a triple (G, F, {pi}), where (i) G(V, E) is an undirected graph with n nodes with each node having a 1-bit state, (ii) F = {l_brace} f{sub 1},f{sub 2},...,f{sub n}{r_brace} is a set of local transition functions with f{sub i} denoting a Boolean function associated with node Vv{sub i} and (iii) {pi} is a fixed permutation of (i.e., a total order on) the nodes in V. A single SDS transition is obtained by updating the states of the nodes in V by evaluating the function associated with each of them in the order given by {pi}. Such a (finite) SDS is a mathematical abstraction of simulation systems [BMR99, BR99]. In this paper, we characterize the computational complexity of determining several phase space properties of SDSs. The properties considered are t-REACHABILITY ('Can a given SDS starting from configuration I reach configuration B in t or fewer transitions?'), REACHABILITY('Can a given SDS starting from configuration I ever reach configuration B?') and FIXED POINT REACHABILITY ('Can a given SDS starting from configuration I ever reach configuration in which it stays for ever?'). Our main result is a sharp dichotomy between classes of SDSs whose behavior is 'easy' to predict and those whose behavior is 'hard' to predict. Specifically, we show the following. (1) The t-REACHABILITY, REACHABILITY and the FIXED POINT REACHABILITY problems for SDSs are PSPACE-complete, even when restricted to graphs of bounded bandwidth (and hence of bounded pathwidth and treewidth) and when the function associated with each node is symmetric. The result holds even for regular graphs of constant degree where all the nodes compute the same symmetric Boolean function. (2) In contrast, the t-REACHABILITYm REACHABILITY and FIXED POINT REACHABILITY problems are solvable in polynomial time for SDSs when the Boolean function associated with each node is symmetric and monotone. Two important consequences of our results are the following: (i) The close correspondence between SDSs and cellular automata (CA), in conjunctio with our lower bounds for SDSs, yields stronger lower bounds on the complexity of reachability problems for CA than known previously. (ii) REACHABILITY problems for hierarchically-specified linearly inter-connected copies of a single finite automaton are EXPSPACE-hard. The results can be combined with our related results to show hardness of a number of equivalence relations for such automata. The results can also be used to demonstrate that determining the sensitivity to initial conditions of such automata (as proposed in [Mo90, BPT91]) is computationally intractable.