Approximate Matrix Inversion
RB Flavell · Journal of the Operational Research Society · 1977
IN MANY branches of applied mathematics, iterative algorithms that require a different matrix inverse on each iteration are used. Frequently, in an attempt to reduce the overall computational time, the use of an approximate matrix inversion formula for several successive iterations is proposed, with computational accuracy being regained later in the algorithm. A particular example is the non-linear programming algorithm Generalised Reduced Gradient for which Abadie and Guigou' suggest the use of