A New Proof of the Maximum Principle in Optimal Control Theory