===firstname:
Scott
===firstname3:
Ilya
===affil6:

===lastname3:
Lashuk
===email:
smaclachlan@mun.ca
===keyword_other2:

===lastname6:

===affil5:

===lastname4:

===lastname7:

===affil7:

===postal:
Department of Mathematics and Statistics
Memorial University of Newfoundland
St. John's, NL
A1C 5S7
Canada
===ABSTRACT:
While discretization and solvers for scalar elliptic PDEs
over regions in the plane or volumes in space are generally
well-understood, many open questions remain when considering PDEs
posed on surfaces.  In this talk, we present an adaptive mesh
refinement scheme for a finite-element discretization of diffusion on
the sphere, motivated by numerical models of the transport of charged
particles near the Fokker-Planck limit.  In such problems, the
right-hand side is typically highly localized and, thus, nonuniform
meshes are needed to efficiently resolve the solutions.  We
approximate the sphere by a piecewise-triangular surface, with
refinement focused around the regions where the source function is
nonzero.  A Fast Adaptive Composite grid (FAC) multigrid method is
used to efficiently solve the resulting linear systems.
===affil3:
Penn State University
===title:
Composite Grid Multigrid for Diffusion on the Sphere
===affil2:
Tufts University
===lastname2:
Adler
===firstname4:

===keyword1:
APP_OTHER
===workshop:
no
===lastname:
MacLachlan
===firstname5:

===keyword2:
NOT_SPECIFIED
===otherauths:

===affil4:

===competition:
no
===firstname7:

===firstname6:

===keyword_other1:
Multigrid methods
===lastname5:

===affilother:

===firstname2:
James