000280107 001__ 280107
000280107 005__ 20210129221218.0
000280107 0247_ $$2doi$$a10.1016/j.cpc.2015.01.009
000280107 0247_ $$2ISSN$$a0010-4655
000280107 0247_ $$2ISSN$$a1386-9485
000280107 0247_ $$2ISSN$$a1879-2944
000280107 0247_ $$2WOS$$aWOS:000351645900005
000280107 037__ $$aFZJ-2015-07853
000280107 082__ $$a004
000280107 1001_ $$0P:(DE-HGF)0$$aBegau, Christoph$$b0$$eCorresponding author
000280107 245__ $$aAdaptive dynamic load-balancing with irregular domain decomposition for particle simulations
000280107 260__ $$aAmsterdam$$bNorth Holland Publ. Co.$$c2015
000280107 3367_ $$0PUB:(DE-HGF)16$$2PUB:(DE-HGF)$$aJournal Article$$bjournal$$mjournal$$s1450775688_21615
000280107 3367_ $$2DataCite$$aOutput Types/Journal article
000280107 3367_ $$00$$2EndNote$$aJournal Article
000280107 3367_ $$2BibTeX$$aARTICLE
000280107 3367_ $$2ORCID$$aJOURNAL_ARTICLE
000280107 3367_ $$2DRIVER$$aarticle
000280107 520__ $$aWe present a flexible and fully adaptive dynamic load-balancing scheme, which is designed for particle simulations of three-dimensional systems with short ranged interactions. The method is based on domain decomposition with non-orthogonal non-convex domains, which are constructed based on a local repartitioning of computational work between neighbouring processors. Domains are dynamically adjusted in a flexible way under the condition that the original topology is not changed, i.e. neighbour relations between domains are retained, which guarantees a fixed communication pattern for each domain during a simulation. Extensions of this scheme are discussed and illustrated with examples, which generalise the communication patterns and do not fully restrict data exchange to direct neighbours. The proposed method relies on a linked cell algorithm, which makes it compatible with existing implementations in particle codes and does not modify the underlying algorithm for calculating the forces between particles. The method has been implemented into the molecular dynamics community code IMD and performance has been measured for various molecular dynamics simulations of systems representing realistic problems from materials science. It is found that the method proves to balance the work between processors in simulations with strongly inhomogeneous and dynamically changing particle distributions, which results in a significant increase of the efficiency of the parallel code compared both to unbalanced simulations and conventional load-balancing strategies.
000280107 536__ $$0G:(DE-HGF)POF3-511$$a511 - Computational Science and Mathematical Methods (POF3-511)$$cPOF3-511$$fPOF III$$x0
000280107 588__ $$aDataset connected to CrossRef
000280107 7001_ $$0P:(DE-Juel1)132274$$aSutmann, Godehard$$b1
000280107 773__ $$0PERI:(DE-600)1466511-6$$a10.1016/j.cpc.2015.01.009$$gVol. 190, p. 51 - 61$$p51 - 61$$tComputer physics communications$$v190$$x0010-4655$$y2015
000280107 8564_ $$uhttps://juser.fz-juelich.de/record/280107/files/1-s2.0-S0010465515000181-main.pdf$$yRestricted
000280107 8564_ $$uhttps://juser.fz-juelich.de/record/280107/files/1-s2.0-S0010465515000181-main.gif?subformat=icon$$xicon$$yRestricted
000280107 8564_ $$uhttps://juser.fz-juelich.de/record/280107/files/1-s2.0-S0010465515000181-main.jpg?subformat=icon-1440$$xicon-1440$$yRestricted
000280107 8564_ $$uhttps://juser.fz-juelich.de/record/280107/files/1-s2.0-S0010465515000181-main.jpg?subformat=icon-180$$xicon-180$$yRestricted
000280107 8564_ $$uhttps://juser.fz-juelich.de/record/280107/files/1-s2.0-S0010465515000181-main.jpg?subformat=icon-640$$xicon-640$$yRestricted
000280107 8564_ $$uhttps://juser.fz-juelich.de/record/280107/files/1-s2.0-S0010465515000181-main.pdf?subformat=pdfa$$xpdfa$$yRestricted
000280107 909CO $$ooai:juser.fz-juelich.de:280107$$pVDB
000280107 9101_ $$0I:(DE-588b)5008462-8$$6P:(DE-Juel1)132274$$aForschungszentrum Jülich GmbH$$b1$$kFZJ
000280107 9131_ $$0G:(DE-HGF)POF3-511$$1G:(DE-HGF)POF3-510$$2G:(DE-HGF)POF3-500$$3G:(DE-HGF)POF3$$4G:(DE-HGF)POF$$aDE-HGF$$bKey Technologies$$lSupercomputing & Big Data$$vComputational Science and Mathematical Methods$$x0
000280107 9141_ $$y2015
000280107 915__ $$0StatID:(DE-HGF)0420$$2StatID$$aNationallizenz
000280107 915__ $$0StatID:(DE-HGF)0100$$2StatID$$aJCR$$bCOMPUT PHYS COMMUN : 2014
000280107 915__ $$0StatID:(DE-HGF)0200$$2StatID$$aDBCoverage$$bSCOPUS
000280107 915__ $$0StatID:(DE-HGF)0300$$2StatID$$aDBCoverage$$bMedline
000280107 915__ $$0StatID:(DE-HGF)0310$$2StatID$$aDBCoverage$$bNCBI Molecular Biology Database
000280107 915__ $$0StatID:(DE-HGF)0199$$2StatID$$aDBCoverage$$bThomson Reuters Master Journal List
000280107 915__ $$0StatID:(DE-HGF)0110$$2StatID$$aWoS$$bScience Citation Index
000280107 915__ $$0StatID:(DE-HGF)0150$$2StatID$$aDBCoverage$$bWeb of Science Core Collection
000280107 915__ $$0StatID:(DE-HGF)0111$$2StatID$$aWoS$$bScience Citation Index Expanded
000280107 915__ $$0StatID:(DE-HGF)1150$$2StatID$$aDBCoverage$$bCurrent Contents - Physical, Chemical and Earth Sciences
000280107 915__ $$0StatID:(DE-HGF)9900$$2StatID$$aIF < 5
000280107 9201_ $$0I:(DE-Juel1)JSC-20090406$$kJSC$$lJülich Supercomputing Center$$x0
000280107 980__ $$ajournal
000280107 980__ $$aVDB
000280107 980__ $$aI:(DE-Juel1)JSC-20090406
000280107 980__ $$aUNRESTRICTED