A New Transformed VM-algorithm for Unconstrained Optimization

Abstract

The Broyden class of Quasi-Newton updates is an approximation for the inverse Hessian matrix, are transformed to the standard BFGS update, which makes it possible to generalize the well-known t-BFGS formula. One of the variants, the simpler of them, is given in this study and does not require any additional matrix storage by vector multiplications. Experimental results indicate that the new proposed algorithm was more efficient than the standard BFGS algorithm.