Symmetric and Positive Definite Broyden Update for Unconstrained Optimization

Broyden update is one of the one-rank updates which solves the unconstrained optimization problem but this update does not guarantee the positive definite and the symmetric property of Hessian matrix. In this paper the guarantee of positive definite and symmetric property for the Hessian matrix will be established by updating the vector which represents the difference between the next gradient and the current gradient of the objective function assumed to be twice continuous and differentiable .Numerical results are reported to compare the proposed method with the Broyden method under standard problems.