Regular Temporal Cost Functions

Abstract

Regular cost functions have been introduced recently as an extension to the notion of regular languages with counting capabilities. The specificity of cost functions is that exact values are not considered, but only estimated. In this paper, we study the strict subclass of regular temporal cost functions. In such cost functions, it is only allowed to count the number of occurrences of consecutive events. For this reason, this model intends to measure the length of intervals, i.e., a discrete notion of time. We provide various equivalent representations for functions in this class, using automata, and 'clock based' reduction to regular languages. We show that the conversions are much simpler to obtain, and much more efficient than in the general case of regular cost functions. Our second aim in this paper is to use temporal cost function as a test-case for exploring the algebraic nature of regular cost functions. Following the seminal ideas of Schutzenberger, this results in a decidable algebraic characterization of regular temporal cost functions inside the class of regular cost functions.