Matches in SemOpenAlex for { <https://semopenalex.org/work/W2901498907> ?p ?o ?g. }
Showing items 1 to 61 of
61
with 100 items per page.
- W2901498907 abstract "Graph analysis is widely used to study connectivity, centrality, community and path analysis of social networks, biological networks, communication networks and any interacting objects that can be represented as graphs. Graphs are ubiquitous and particularly they are common in social and physical sciences. The graphs are continuously becoming larger and complex; so scalable and parallel algorithms need to be developed to process and analyze such large graphs. Additionally, the high performance computing (HPC) systems are also becoming complex with multiple cores in a processor and multiple levels in the memory subsystems. We need to utilize HPC systems to develop scalable, parallel and high performing algorithmsto analyze large and complex graphs.To analyze connectivity, centrality and robustness of a graph, it is useful to find the densely connected subgraphs (cohesive subgraphs) of a graph. One of the contributions of this thesis is to design parallel algorithms for computing cohesive subgraphs and using them to analyze graphs. The cohesive subgraphs considered are k-core and k-truss of a graph. A parallel algorithm PKC is developed to computek-core decomposition on shared memory systems. PKC uses less memory and has less synchronization overhead as compared to state-of-the-art algorithms. A parallel k-truss decomposition algorithm PKT is also developed that computes trusses of a large social network graph in minutes where as state-of-the-art algorithms take hours. These algorithms are used to sparsify and reorder social networks.In centrality analysis and scientific computing, an important kernel is sparse matrix-vector multiplication (SpMV). Another contribution of this thesis, is to develop a multi-level data structure (CSR-k) to store sparse matrices/graphs to speedup sparse kernels. CSR-k represents the parallelism present in the sparsekernels and also decreases the work load imbalance among the threads. SpMV using CSR-k achieves a speedup of 2x compared to pOSKI on 32 cores. Sparse triangular solution (STS) is also a very useful kernel in scientific computing. We have used CSR-k and graph coloring to represent sparse triangular solution. STSusing CSR-k achieves 2x speedup compared to coloring." @default.
- W2901498907 created "2018-11-29" @default.
- W2901498907 creator A5014673656 @default.
- W2901498907 date "2018-03-01" @default.
- W2901498907 modified "2023-09-23" @default.
- W2901498907 title "HIERARCHICAL SPARSE GRAPH COMPUTATIONS ON MULTICORE PLATFORMS" @default.
- W2901498907 hasPublicationYear "2018" @default.
- W2901498907 type Work @default.
- W2901498907 sameAs 2901498907 @default.
- W2901498907 citedByCount "0" @default.
- W2901498907 crossrefType "journal-article" @default.
- W2901498907 hasAuthorship W2901498907A5014673656 @default.
- W2901498907 hasConcept C106937863 @default.
- W2901498907 hasConcept C114614502 @default.
- W2901498907 hasConcept C132525143 @default.
- W2901498907 hasConcept C133875982 @default.
- W2901498907 hasConcept C173608175 @default.
- W2901498907 hasConcept C33923547 @default.
- W2901498907 hasConcept C41008148 @default.
- W2901498907 hasConcept C48044578 @default.
- W2901498907 hasConcept C53811970 @default.
- W2901498907 hasConcept C77088390 @default.
- W2901498907 hasConcept C80444323 @default.
- W2901498907 hasConceptScore W2901498907C106937863 @default.
- W2901498907 hasConceptScore W2901498907C114614502 @default.
- W2901498907 hasConceptScore W2901498907C132525143 @default.
- W2901498907 hasConceptScore W2901498907C133875982 @default.
- W2901498907 hasConceptScore W2901498907C173608175 @default.
- W2901498907 hasConceptScore W2901498907C33923547 @default.
- W2901498907 hasConceptScore W2901498907C41008148 @default.
- W2901498907 hasConceptScore W2901498907C48044578 @default.
- W2901498907 hasConceptScore W2901498907C53811970 @default.
- W2901498907 hasConceptScore W2901498907C77088390 @default.
- W2901498907 hasConceptScore W2901498907C80444323 @default.
- W2901498907 hasLocation W29014989071 @default.
- W2901498907 hasOpenAccess W2901498907 @default.
- W2901498907 hasPrimaryLocation W29014989071 @default.
- W2901498907 hasRelatedWork W1710447384 @default.
- W2901498907 hasRelatedWork W1971630691 @default.
- W2901498907 hasRelatedWork W2076723445 @default.
- W2901498907 hasRelatedWork W2131183411 @default.
- W2901498907 hasRelatedWork W2142599626 @default.
- W2901498907 hasRelatedWork W2295065913 @default.
- W2901498907 hasRelatedWork W2399269606 @default.
- W2901498907 hasRelatedWork W2417093649 @default.
- W2901498907 hasRelatedWork W2436509158 @default.
- W2901498907 hasRelatedWork W2519958626 @default.
- W2901498907 hasRelatedWork W2725398544 @default.
- W2901498907 hasRelatedWork W2740693040 @default.
- W2901498907 hasRelatedWork W2767319308 @default.
- W2901498907 hasRelatedWork W2912379686 @default.
- W2901498907 hasRelatedWork W2952995885 @default.
- W2901498907 hasRelatedWork W3037631590 @default.
- W2901498907 hasRelatedWork W3046976269 @default.
- W2901498907 hasRelatedWork W3103250330 @default.
- W2901498907 hasRelatedWork W3161374268 @default.
- W2901498907 hasRelatedWork W3176274902 @default.
- W2901498907 isParatext "false" @default.
- W2901498907 isRetracted "false" @default.
- W2901498907 magId "2901498907" @default.
- W2901498907 workType "article" @default.