Penalty Method for Nonlinear Programming