Capture Complexity by Partition
Yijia Chen5
and Enshao Shen6 
| (5) |
Laboratoire de Recherche en Informatique, Université de Paris-Sud, Bâtiment 490, F-91405 Orsay Cedex, France |
| (6) |
Department of Computer Science, Shanghai Jiao Tong University, Shanghai, 200030, China |
Abstract
We show in this paper a special extended logic, partition logic based on so called partition quantifiers, is able to capture
some important complexity classes NP, P and NL by its natural fragments. The Fagin’s Theorem and Immerman-Vardi’s Theorem are rephrased and strengthened into a uniform
partition logic setting. Also the dual operators for the partition quantifiers are introduced to expose some of their important
model-theoretic properties. In particular they enable us to show a 0-1 law for the partition logic, even when finite variable
infinitary logic is adjunct to it. As a consequence, partition logic cannot count without built-in ordering on structures.
Considering its better theoretical properties and tools than those of second order logic, partition logic may provide us with
an alternative, yet uniform insight for descriptive complexity.
Supported by Calife, a project of “Réseau National de Recherche en Télécommunications”.
Supported in part by National Foundation of Natural Science of China.
References secured to subscribers.