A LU Decomposition Useful for Antenna Optimization

Abstract

This paper describes a LU
decomposition technique useful when
solving a series of matrix equations in which
only a small fraction of the original matrix
changes from run to run. On the first run,
the entire matrix must be computed and LU
decomposed. However, on the second or
subsequent runs, only those rows and
columns of the matrix which have changed
need be recomputed and re-LU decomposed.
If only a small fraction of the matrix has
changed, this results in a substantial saving
in CPU time both in the computation of the
original matrix and in its LU decomposition.