DocumentCode
2422742
Title
Multicloud Convergence Acceleration for Complex Applications on Arbitrary Grids
Author
Katz, Aaron ; Jameson, Antony
Author_Institution
Dept. of Aeronaut. & Astronaut., Stanford Univ., Stanford, CA
fYear
2008
fDate
14-17 July 2008
Firstpage
479
Lastpage
485
Abstract
A method is presented which dramatically enhances convergence of discrete systems for a wide variety of discretizations, including finite difference, finite volume, and mixed element approaches. The method, termed ldquomulticloud,rdquo was first introduced in a previous work by Katz and Jameson.[5] This work concerns two major enhancements to multicloud: 1) The integration of multicloud procedures for arbitrary and mixed meshes, and 2) the development of automatic coarsening procedures to obtain coarse levels for the multicloud procedure. The first enhancement involves the use of meshless scheme to damp out unresolvable high frequency modes present on a conventional mesh. Because the meshless scheme is only used on coarse levels, the final solution retains any and all conservative properties of the parent mesh. The second enhancement allows for the automatic creation of coarse point distributions given a conventional mesh. Thus, the multicloud procedure may be used with no extra mesh generation tools needed. These two enhancements create a powerful convergence acceleration technique which is entirely mesh transparent, lending itself well to complex configurations in which mixed grid types are used and for which efficiency is of the utmost importance.
Keywords
computational geometry; convergence of numerical methods; discrete systems; finite difference methods; finite volume methods; mesh generation; discrete systems; finite difference method; finite volume method; mixed element approach; multicloud convergence acceleration; Acceleration; Convergence; Damping; Finite difference methods; Finite volume methods; Frequency; Fuses; Geometry; Mesh generation; Topology;
fLanguage
English
Publisher
ieee
Conference_Titel
DoD HPCMP Users Group Conference, 2008. DOD HPCMP UGC
Conference_Location
Seattle, WA
Print_ISBN
978-1-4244-3323-0
Type
conf
DOI
10.1109/DoD.HPCMP.UGC.2008.64
Filename
4755912
Link To Document