Abstract
Parallel computing is applied to the solution of an inverse problem arising from a system of hyperbolic partial differential equations. A sequential algorithm is converted to a parallel algorithm, and it is executed on a 64-processor computer with a hypercube architecture. Each processor is assigned to perform the computations for every point along a particular characteristic of the equations. Two different algorithms are described and implemented, and their relative merits are compared. Various methods for solving the direct problem arising from the same system of equations are discussed; a solution to the direct problem produces the data for input to the routine that solves the inverse problem. The problem is solved for different numbers of discretization points, and the results are examined for accuracy and time of execution. Measures of speedup and efficiency for the parallelization are given.
Phillips, Mike Randall (1993). A parallel solution of the inverse problem associated with a hyperbolic partial differential equation. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1993 -THESIS -P558.