On scalable parallel recursive backtracking

Abu-Khzam, Faisal N.; Daudjee, Khuzaima; Mouawad, Amer E.; Nishimura, Nobuya

doi:10.1016/j.jpdc.2015.07.006

Cited by 17 publications

(17 citation statements)

References 23 publications

(73 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since the emergence of the notion of parameterized parallel complexity and even before that, limited has been the scope of research on the topic, apart from experimental work that basically focused on scalability of parallel parameterized algorithms [13,14] To the best of our knowledge, the first FPP algorithm dates back to 1988 and showed that the Feedback Vertex Set problem, on cyclically reducible digraphs, can be solved in O(k * log 2 (n)) time using O(n 4 ) processors [15]. Of course, the notion of fixed-parameter parallelism was not known at that time.…”

Section: Key Results and Methodsmentioning

confidence: 99%

A Brief Survey of Fixed-Parameter Parallelism

Abu-Khzam

Kontar

2020

Algorithms

Self Cite

View full text Add to dashboard Cite

This paper provides an overview of the field of parameterized parallel complexity by surveying previous work in addition to presenting a few new observations and exploring potential new directions. In particular, we present a general view of how known FPT techniques, such as bounded search trees, color coding, kernelization, and iterative compression, can be modified to produce fixed-parameter parallel algorithms.

show abstract

Section: Key Results and Methodsmentioning

confidence: 99%

A Brief Survey of Fixed-Parameter Parallelism

Abu-Khzam

Kontar

2020

Algorithms

Self Cite

View full text Add to dashboard Cite

show abstract

“…3) On-demand work generation bypasses workpools; instead idle workers steal unexplored subtrees directly from other workers. Abu-khzam et al [1] show such techniques to be highly scalable e.g. up to 131,072 workers.…”

Section: Existing Search Approachesmentioning

confidence: 99%

“…Stack-Stealing coordination provides on-demand work generation, triggered by work-stealing, similar to [1]. On receiving a steal request, the victim ships the lowest unexplored subtree from its own stack to the thief.…”

Section: Fig 3: Yewpar System Stackmentioning

confidence: 99%

See 1 more Smart Citation

Implementing YewPar: A Framework for Parallel Tree Search

Archibald

Maier

Stewart

et al. 2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

0000−0003−3699−6658] , Patrick Maier 3[0000−0002−7051−8169] , Robert Stewart 2[0000−0003−0365−693X] , and Phil Trinder 1[0000−0003−0190−7010]Abstract. Combinatorial search is central to many applications yet hard to parallelise. We argue for improving the reuse of parallel searches, and present the design and implementation of a new parallel search framework. YewPar generalises search by abstracting search tree generation, and by providing algorithmic skeletons that support three search types, together with a set of search coordination strategies. The evaluation shows that the cost of YewPar generality is low (6.1%); global knowledge is inexpensively shared between workers; irregular tasks are effectively distributed; and YewPar delivers good runtimes, speedups and efficiency with up to 255 workers on 17 localities.

show abstract

“…This is especially useful for graph problems since the input-partitioning approach is often infeasible. In fact, the search space partitioning approach proved to be highly effective on many had graph theoretic problems (see Abu-Khzam, Daudjee, Mouawad, and Nishimura (2015); Abu-Khzam, Langston, and Shanbhag (2004); Abu-Khzam, Langston, Shanbhag, and Symons (2006); Abu-Khzam and Mouawad (2012) for more information).…”

Section: Introductionmentioning

confidence: 99%

Scalable parallel algorithms for dynamic programming on tree decomposition

Yassine¹

View full text Add to dashboard Cite

Dynamic programming on tree decompositions is often a key for solving a widerange of optimization problems that would otherwise be intractable. Operations on tree decompositions are amenable to parallelization. However, a systematic approach towards such parallelization is still lacking. In this work, we present a generic and flexible framework for parallel dynamic programming on tree decompositions that can be easily adapted to solve any graph theoretic optimization problem on graphs of bounded treewidth. We show the effectiveness of our framework using the Dominating Set problem as a case study. Our experiments show notable speedups compared to the serial approach.

show abstract

On scalable parallel recursive backtracking

Cited by 17 publications

References 23 publications

A Brief Survey of Fixed-Parameter Parallelism

A Brief Survey of Fixed-Parameter Parallelism

Implementing YewPar: A Framework for Parallel Tree Search

Scalable parallel algorithms for dynamic programming on tree decomposition

Contact Info

Product

Resources

About