Replicable parallel branch and bound search
- Submitting institution
-
Heriot-Watt University
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 23510697
- Type
- D - Journal article
- DOI
-
10.1016/j.jpdc.2017.10.010
- Title of journal
- Journal of Parallel and Distributed Computing
- Article number
- -
- First page
- 92
- Volume
- 113
- Issue
- -
- ISSN
- 0743-7315
- Open access status
- Compliant
- Month of publication
- November
- Year of publication
- 2017
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
4
- Research group(s)
-
-
- Citation count
- 6
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Branch and bound algorithms are central to combinatorial search and optimisation and are challenging to parallelise due to high irregularity and dependence on search order. Our approach is (1) generic: using algorithmic skeletons to encode parallel search problems, (2) scalable: to hundreds of cores, (3) replicable: preserving search order achieves predictable parallel performance. We specify multithreaded search using a parallel operational semantics. Published in an A* ranked journal, the work is part of a programme, partially supported by EPSRC CoDiMa (EP/M022641), to scale combinatorial search to HPC platforms and establish symbolic computation as an HPC application domain.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
