Lagrange multipliers theorem and saddle point optimality criteria in mathematical programming