Well, the belief is not worth much here. How do you express optimality condition of the form f'(x) = 0 with finite difference? Answer: you don't.

But if this doesn't ring a bell, it's probably better to drop this.