Descriptive Complexity for Counting Complexity Classes
Marcelo Arena (Pontificia Universidad Católica de Chile)
Abstract: Descriptive Complexity has been very successful in characterizing complexity classes of decision problems in terms of the properties definable in some logics. However, descriptive complexity for counting complexity classes, such as FP and #P, has not been systematically studied, and it is not as developed as its decision counterpart. In this talk, we will present a framework based on Weighted Logics to address this issue. Specifically, by focusing on the natural numbers we obtain a logic called Quantitative Second Order Logics (QSO), and show how some of its fragments can be used to capture fundamental counting complexity classes such as FP, #P and FPSPACE, among others. Moreover, we use QSO to define a hierarchy inside #P, identifying counting complexity classes with good closure and approximation properties, and which admit natural complete problems.
computational complexitylogic in computer sciencelogic
Audience: researchers in the topic
Series comments: Description: Seminar on all areas of logic
Organizer: | Wesley Calvert* |
*contact for this listing |