Key points are not available for this paper at this time.
A novel method of function minimization that combines the power of the diagonal approximation to the normal matrix with conjugate directions is described. This method approaches closer to the local minimum than the methods that are commonly used in macromolecular refinement. The weakness of the current methods are analyzed to explain the advantage of the conjugate-direction method.
Dale E. Tronrud (1992) studied this question.