Numerical Methods: Euler to Runge–Kutta · Challenge problem

When Euler's method is run in floating-point arithmetic, the total error at a fixed time is modelled by $E(h) = \dfrac{Mh}{2} + \dfrac{\delta}{h}$: truncation error grows with $h$, accumulated round-off grows with the number of steps. With $M = 2$ and $\delta = 10^{-8}$, what is the smallest possible value of $E(h)$ over $h > 0$? Express your answer as a decimal.