A note on the use of the conjugate gradient method in the solution of a large system of sparse equations

Abstract

The method of conjugate gradients, as an iterative, method for solving simultaneous equations, was found to be useful in a problem which involved fitting a large amount of photographic data to co-ordinates of visual landmarks on the plant Mars. The method proved to be useful because the normal matrix obtained was very sparse. A somewhat more complicated metric than traditionally used was needed to control the error for problems in which the dimension of the normal matrix was large. Successful runs were obtained with 4500×4500 normal matrices.