The epsilon variation method in two-point boundary-value problems