initial value problem