Extended Abstract:
Advances in computer technology, scientific visualization, and
discretization methods have made possible the numerical simulation of
complex physical phenomena. More often than not, this involves the
numerical solution of systems of partial differential equations
utilizing finite difference, finite element or finite volume methods
over realistic three-dimensional geometries. Inherent to any of these
solution methods is, in addition to the discrete representation of the
differential and/or integral operators in the equations, the discrete
representation of the solution domain. This discrete domain over which
the discrete equations are solved is called a grid [or a mesh]. One
problem with realistic geometries is that they tend to be complex and as
a result, the construction and design of grid structures over such
geometries often becomes a very difficult task. In fact, the generation
of a valid grid is usually the most labor intensive part of any
computational field simulation. Volume grid cells are usually
constructed using a trivariate polynomial map defined on a reference
domain. The simplest and most popular trivariate is the trilinear. The
map and its Jacobian are represented in Bezier form and a pyramid
algorithm is utilized to simultaneously compute points and geometric
characteristics associated with the map. In addition, sufficient
conditions are given for the positivity of the Jacobian determinant and
an iterative algorithm for solving the inversion problem is derived.
The convergence and the accuracy of numerical solutions to partial
differential equations strongly depend on the quality of the grids on
which these solutions are computed. First and second order geometric
characteristics for hexahedral volume grids are formulated and applied
to evaluate the quality of three-dimensional grid structures. Examples
measuring the Jacobian and the orthogonality of geologic grids are
given.
Bio-Sketch:
Vasilis G. Zafiris is a tenured Associate Professor of Computer &
Mathematical Sciences at the University of Houston-Downtown. He
received his B.S. in Aerospace Engineering and M.S. in Applied
Mathematics both from Mississippi State University, and his Ph.D. in
Applied Mathematics from the University of Houston-Central. His primary
research interests focus on Scientific and Geometric Computing. He has
published in journals and proceedings and has presented his work in
national and international conferences. Before joining the university
in 1999, he worked at Landmark Graphics in the Reservoir Simulation
Group, at Los Alamos National Lab in the Earth and Environmental Science
Group, and at Lawrence Livermore National Lab in the Scientific
Computing Group. More recently, he worked as a consultant for Landmark
Graphics and for GX Technologies.