Full metadata record
DC FieldValueLanguage
dc.contributor.authorHwang, Feng-Nanen_US
dc.contributor.authorCai, Shang-Rongen_US
dc.contributor.authorShao, Yun-Longen_US
dc.contributor.authorWu, Jong-Shinnen_US
dc.date.accessioned2014-12-08T15:48:26Z-
dc.date.available2014-12-08T15:48:26Z-
dc.date.issued2010-09-01en_US
dc.identifier.issn0010-4655en_US
dc.identifier.urihttp://dx.doi.org/10.1016/j.cpc.2010.05.003en_US
dc.identifier.urihttp://hdl.handle.net/11536/32263-
dc.description.abstractWe investigate fully parallel Newton-Krylov-Schwarz (NKS) algorithms for solving the large sparse nonlinear systems of equations arising from the finite element discretization of the three-dimensional Poisson-Boltzmann equation (PBE), which is often used to describe the colloidal phenomena of an electric double layer around charged objects in colloidal and interfacial science. The NKS algorithm employs an inexact Newton method with backtracking (INB) as the nonlinear solver in conjunction with a Krylov subspace method as the linear solver for the corresponding Jacobian system. An overlapping Schwarz method as a preconditioner to accelerate the convergence of the linear solver. Two test cases including two isolated charged particles and two colloidal particles in a cylindrical pore are used as benchmark problems to validate the correctness of our parallel NKS-based PBE solver. In addition, a truly three-dimensional case, which models the interaction between two charged spherical particles within a rough charged micro-capillary, is simulated to demonstrate the applicability of our PBE solver to handle a problem with complex geometry. Finally, based on the result obtained from a PC cluster of parallel machines, we show numerically that NKS is quite suitable for the numerical simulation of interaction between colloidal particles, since NKS is robust in the sense that INB is able to converge within a small number of iterations regardless of the geometry, the mesh size, the number of processors. With help of an additive preconditioned Krylov subspace method NKS achieves parallel efficiency of 71% or better on up to a hundred processors for a 3D problem with 5 million unknowns. (C) 2010 Elsevier B.V. All rights reserved.en_US
dc.language.isoen_USen_US
dc.subjectPoisson-Boltzmann equationen_US
dc.subjectOverlapping Schwarz preconditioningen_US
dc.subjectInexact Newtonen_US
dc.subjectFinite elementen_US
dc.subjectParallel processingen_US
dc.subjectColloidal particlesen_US
dc.titleParallel Newton-Krylov-Schwarz algorithms for the three-dimensional Poisson-Boltzmann equation in numerical simulation of colloidal particle interactionsen_US
dc.typeArticleen_US
dc.identifier.doi10.1016/j.cpc.2010.05.003en_US
dc.identifier.journalCOMPUTER PHYSICS COMMUNICATIONSen_US
dc.citation.volume181en_US
dc.citation.issue9en_US
dc.citation.spage1529en_US
dc.citation.epage1537en_US
dc.contributor.department機械工程學系zh_TW
dc.contributor.departmentDepartment of Mechanical Engineeringen_US
dc.identifier.wosnumberWOS:000280873500006-
dc.citation.woscount0-
Appears in Collections:Articles


Files in This Item:

  1. 000280873500006.pdf

If it is a zip file, please download the file and unzip it, then open index.html in a browser to view the full text content.