Abstract
A bounded-transient finite automaton is an automaton for which a single change of an input will not affect the output "far away." This paper investigates the class of events which bounded-transient automata can compute and shows that this class of events is a generalization of the class of definite events. A method of determining whether or not a finite automaton is bounded-transient is described, and the connection between the results of this paper and Kilmer's results is indicated.