Techniques of the simplex basis LU factorization update

The objective of this work is to compare the developed LU factorization update with results from MINOS. This technique, use a matrix columns static reordering and they are rearranged in accordance with the increasing number of nonzero entries and triangularized, leading to sparse basis factorization without computational effort to reorder the columns. Only the columns factorizations actually modified by the change of basis are carried through due to matrix sparse structure. Computational results in Matlab for problems from the Netlib show that this is a very promising idea, since there is no need to refactorize the matrix in the tested problems