TY - GEN
T1 - Lazy abstraction for size-change termination
AU - Codish, Michael
AU - Fuhs, Carsten
AU - Giesl, Jürgen
AU - Schneider-Kamp, Peter
PY - 2010/1/1
Y1 - 2010/1/1
N2 - Size-change termination is a widely used means of proving termination where source programs are first abstracted to size-change graphs which are then analyzed to determine if they satisfy the size-change termination property. Here, the choice of the abstraction is crucial to the success of the method, and it is an open problem how to choose an abstraction such that no critical loss of precision occurs. This paper shows how to couple the search for a suitable abstraction and the test for size-change termination via an encoding to a single SAT instance. In this way, the problem of choosing the right abstraction is solved en passant by a SAT solver. We show that for the setting of term rewriting, the integration of this approach into the dependency pair framework works smoothly and gives rise to a new class of size-change reduction pairs. We implemented size-change reduction pairs in the termination prover AProVE and evaluated their usefulness in extensive experiments.
AB - Size-change termination is a widely used means of proving termination where source programs are first abstracted to size-change graphs which are then analyzed to determine if they satisfy the size-change termination property. Here, the choice of the abstraction is crucial to the success of the method, and it is an open problem how to choose an abstraction such that no critical loss of precision occurs. This paper shows how to couple the search for a suitable abstraction and the test for size-change termination via an encoding to a single SAT instance. In this way, the problem of choosing the right abstraction is solved en passant by a SAT solver. We show that for the setting of term rewriting, the integration of this approach into the dependency pair framework works smoothly and gives rise to a new class of size-change reduction pairs. We implemented size-change reduction pairs in the termination prover AProVE and evaluated their usefulness in extensive experiments.
UR - https://www.scopus.com/pages/publications/85037544890
U2 - 10.1007/978-3-642-16242-8_16
DO - 10.1007/978-3-642-16242-8_16
M3 - Conference contribution
AN - SCOPUS:85037544890
SN - 364216241X
SN - 9783642162411
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 217
EP - 232
BT - Logic for Programming, Artificial Intelligence, and Reasoning - 17th International Conference, LPAR-17, Proceedings
PB - Springer Verlag
ER -