Book chapter
A novel asynchronous parallelism scheme for first-order logic
Automated Deduction — CADE-12, pp.484-498
Lecture Notes in Computer Science, Springer Berlin Heidelberg
05/30/2005
DOI: 10.1007/3-540-58156-1_35
Abstract
We present a new parallelization technique designed to speed the solution of first-order inference problems by detecting and pruning unnecessary search. Our technique, called nagging, is naturally fault tolerant and relatively robust in the presence of high communication latency. As a result, it is well suited to execution on a network of loosely-coupled processors. Our implementation has been tested on as many as 100 workstations and exhibits encouraging scaling properties. In this paper, we introduce nagging in the context of existing work on parallel search for logical inference, present an informal analysis of its operation, and present some empirical results for our PTTP-style prototype.
Details
- Title: Subtitle
- A novel asynchronous parallelism scheme for first-order logic
- Creators
- David B. Sturgill - Cornell UniversityAlberto Maria Segre - Cornell University
- Resource Type
- Book chapter
- Publication Details
- Automated Deduction — CADE-12, pp.484-498
- Publisher
- Springer Berlin Heidelberg; Berlin, Heidelberg
- Series
- Lecture Notes in Computer Science
- DOI
- 10.1007/3-540-58156-1_35
- eISSN
- 1611-3349
- ISSN
- 0302-9743
- Language
- English
- Date published
- 05/30/2005
- Academic Unit
- Fraternal Order of Eagles Diabetes Research Center; Nursing; Computer Science
- Record Identifier
- 9984259417602771
Metrics
5 Record Views