Publication View

Randomized Parallel Algorithms For The Homing Sequence Problem (1996)

Abstract
Homing sequences play an important role in the testing of finite state systems and have been recently used in learning algorithms due to Rivest and Schapire [17] and Freund et al. [4]. It is well-known that every minimal DFA has a homing sequence of length O(n 2 ) which can be constructed sequentially in time O(n 3 ). But no efficient parallel algorithm was known for this problem. In this work, we present two RNC algorithms of time complexity O(log 2 n) for this problem. We show that one of our RNC algorithms produces a homing sequence of length O(n log 2 n) for almost all DFA's with n states using a random model of Traktenbrot and Barzdin. We also discuss connections between the homing sequence problem and other problems in the field of hardware fault-testing and protocol verification. 1. INTRODUCTION Locating the current state in a finite-state system is a fundamental problem in map-learning and robotics. This problem has many variations. The version we consider in this pape...

Publication details
Download http://citeseer.ist.psu.edu/34679.html
Source http://homepage.cs.uri.edu/faculty/ravikumar/icpp.ps
Publisher unknown
Contributors The Pennsylvania State University CiteSeer Archives
Repository CiteSeer (United States)
Keywords B. Ravikumar,X. Xiong Randomized Parallel Algorithms For The Homing Sequence Problem
Language Englisch
Relation oai:CiteSeerPSU:23165, oai:CiteSeerPSU:443298, oai:CiteSeerPSU:93406