Matches in SemOpenAlex for { <https://semopenalex.org/work/W2307160900> ?p ?o ?g. }
- W2307160900 endingPage "A558" @default.
- W2307160900 startingPage "A532" @default.
- W2307160900 abstract "Many fields require computing the trace of the inverse of a large, sparse matrix. Since dense matrix methods are not practical, the typical method used for such computations is the Hutchinson method, which is a Monte Carlo (MC) averaging over matrix quadratures. To improve its slow convergence, several variance reduction techniques have been proposed. In this paper, we study the effects of deflating the near null singular value space. We make two main contributions: one theoretical, and one by engineering a solution to a real-world application. First, we analyze the variance of the Hutchinson method as a function of the deflated singular values and vectors. By assuming additionally that the singular vectors are random unitary matrices, we arrive at concise formulas for the deflated variance that include only the variance and the mean of the singular values. In this case, we make the remarkable observation that deflation may increase variance for Hermitian matrices but not for non-Hermitian ones. The theory can be used as a model for predicting the benefits of deflation. Experimentation shows that the model is robust even when the singular vectors are not random. Second, we use deflation in the context of a large scale application of “disconnected diagrams” in lattice quantum chromodynamics. On lattices, hierarchical probing (HP) has previously provided significant variance reduction over MC by removing “error” from neighboring nodes of increasing distance in the lattice. Although deflation used directly on MC yields a limited improvement of 30% in our problem, when combined with HP they reduce variance by a factor of about 60 over MC. We explain this synergy theoretically and provide a thorough experimental analysis. One of the important steps of our solution is the precomputation of 1000 smallest singular values of an ill-conditioned matrix of size 25 million. Using the state-of-the-art packages PRIMME and a domain-specific algebraic multigrid preconditioner, we solve this large eigenvalue computation on 32 nodes of a Cray Edison in about 1.5 hours and at a fraction of the cost of our trace computation." @default.
- W2307160900 created "2016-06-24" @default.
- W2307160900 creator A5048499869 @default.
- W2307160900 creator A5079391957 @default.
- W2307160900 creator A5090417045 @default.
- W2307160900 date "2017-01-01" @default.
- W2307160900 modified "2023-10-16" @default.
- W2307160900 title "Deflation as a Method of Variance Reduction for Estimating the Trace of a Matrix Inverse" @default.
- W2307160900 cites W1834106819 @default.
- W2307160900 cites W1851008583 @default.
- W2307160900 cites W1990457366 @default.
- W2307160900 cites W2024364296 @default.
- W2307160900 cites W2027257397 @default.
- W2307160900 cites W2037931139 @default.
- W2307160900 cites W2057309399 @default.
- W2307160900 cites W2057450058 @default.
- W2307160900 cites W2073737032 @default.
- W2307160900 cites W2087338340 @default.
- W2307160900 cites W2089663769 @default.
- W2307160900 cites W2108214478 @default.
- W2307160900 cites W2280006024 @default.
- W2307160900 cites W2964321151 @default.
- W2307160900 cites W985544890 @default.
- W2307160900 doi "https://doi.org/10.1137/16m1066361" @default.
- W2307160900 hasPublicationYear "2017" @default.
- W2307160900 type Work @default.
- W2307160900 sameAs 2307160900 @default.
- W2307160900 citedByCount "22" @default.
- W2307160900 countsByYear W23071609002016 @default.
- W2307160900 countsByYear W23071609002017 @default.
- W2307160900 countsByYear W23071609002018 @default.
- W2307160900 countsByYear W23071609002019 @default.
- W2307160900 countsByYear W23071609002020 @default.
- W2307160900 countsByYear W23071609002021 @default.
- W2307160900 countsByYear W23071609002022 @default.
- W2307160900 countsByYear W23071609002023 @default.
- W2307160900 crossrefType "journal-article" @default.
- W2307160900 hasAuthorship W2307160900A5048499869 @default.
- W2307160900 hasAuthorship W2307160900A5079391957 @default.
- W2307160900 hasAuthorship W2307160900A5090417045 @default.
- W2307160900 hasBestOaLocation W23071609003 @default.
- W2307160900 hasConcept C105795698 @default.
- W2307160900 hasConcept C106487976 @default.
- W2307160900 hasConcept C109282560 @default.
- W2307160900 hasConcept C121332964 @default.
- W2307160900 hasConcept C126285488 @default.
- W2307160900 hasConcept C151730666 @default.
- W2307160900 hasConcept C158693339 @default.
- W2307160900 hasConcept C159985019 @default.
- W2307160900 hasConcept C162324750 @default.
- W2307160900 hasConcept C17744445 @default.
- W2307160900 hasConcept C192562407 @default.
- W2307160900 hasConcept C19499675 @default.
- W2307160900 hasConcept C199539241 @default.
- W2307160900 hasConcept C202444582 @default.
- W2307160900 hasConcept C2779343474 @default.
- W2307160900 hasConcept C28826006 @default.
- W2307160900 hasConcept C33923547 @default.
- W2307160900 hasConcept C556758197 @default.
- W2307160900 hasConcept C62520636 @default.
- W2307160900 hasConcept C62644790 @default.
- W2307160900 hasConcept C64812099 @default.
- W2307160900 hasConcept C67820243 @default.
- W2307160900 hasConcept C8112001 @default.
- W2307160900 hasConcept C86803240 @default.
- W2307160900 hasConcept C94940 @default.
- W2307160900 hasConcept C96214315 @default.
- W2307160900 hasConceptScore W2307160900C105795698 @default.
- W2307160900 hasConceptScore W2307160900C106487976 @default.
- W2307160900 hasConceptScore W2307160900C109282560 @default.
- W2307160900 hasConceptScore W2307160900C121332964 @default.
- W2307160900 hasConceptScore W2307160900C126285488 @default.
- W2307160900 hasConceptScore W2307160900C151730666 @default.
- W2307160900 hasConceptScore W2307160900C158693339 @default.
- W2307160900 hasConceptScore W2307160900C159985019 @default.
- W2307160900 hasConceptScore W2307160900C162324750 @default.
- W2307160900 hasConceptScore W2307160900C17744445 @default.
- W2307160900 hasConceptScore W2307160900C192562407 @default.
- W2307160900 hasConceptScore W2307160900C19499675 @default.
- W2307160900 hasConceptScore W2307160900C199539241 @default.
- W2307160900 hasConceptScore W2307160900C202444582 @default.
- W2307160900 hasConceptScore W2307160900C2779343474 @default.
- W2307160900 hasConceptScore W2307160900C28826006 @default.
- W2307160900 hasConceptScore W2307160900C33923547 @default.
- W2307160900 hasConceptScore W2307160900C556758197 @default.
- W2307160900 hasConceptScore W2307160900C62520636 @default.
- W2307160900 hasConceptScore W2307160900C62644790 @default.
- W2307160900 hasConceptScore W2307160900C64812099 @default.
- W2307160900 hasConceptScore W2307160900C67820243 @default.
- W2307160900 hasConceptScore W2307160900C8112001 @default.
- W2307160900 hasConceptScore W2307160900C86803240 @default.
- W2307160900 hasConceptScore W2307160900C94940 @default.
- W2307160900 hasConceptScore W2307160900C96214315 @default.
- W2307160900 hasFunder F4320306076 @default.
- W2307160900 hasFunder F4320306084 @default.
- W2307160900 hasFunder F4320332359 @default.
- W2307160900 hasFunder F4320337513 @default.
- W2307160900 hasIssue "2" @default.