1
$\begingroup$

I'm working with a proof of local and global truncation errors for the euler method in numerical methods, and I can't seem to understand the equality and the less/equal seen below. At first I thought it had to do with geometrical sums, but I still don't understand it, any help would be appreciated:

$$1+e^{Lh}+e^{2Lh}+...+e^{(i-1)Lh} = \frac{e^{iLh}-1}{e^{Lh}-1} \le \frac{e^{L(t_i-a)}-1}{Lh}$$

$\endgroup$
  • 1
    $\begingroup$ What are $L,t_i$ and $a$ ? $\endgroup$ – Yves Daoust Mar 31 '17 at 16:42
3
$\begingroup$

The formula $$\sum^i_{n=0} x^n = \frac{x^{i+1}-1}{x-1}$$ holds for any $x\ne 1$ [this is easily seen for $i=1$ and then proven by induction]. Just plug in $x=e^{Lh}$. The inequality follows since $e^{Lh} -1 \ge Lh$ and presumably $ih = t_i -a$.

$\endgroup$
2
$\begingroup$

The equality is indeed the sum of the geometric series (although $i$ is somewhat confusing: it is the number of intervals, not $\sqrt{-1}$ :-) ).

The inequality arises from two things: $iLh = L(t_i - a)$, so the numerators are equal. For the denominators, $Lh$ is apparently assumed small, so you can approximate $e^{Lh}$ by the first two terms of its Taylor expansion. Since the rest of the expansion is positive, leaving it out reduces the total:

$$ e^{Lh} -1 \ge Lh$$

so inverting inverts the inequality.

EDIT: The inequality is true even if $Lh$ is not small, but when it is small, the inequality is "good" in the sense that it provides a tighter bound.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.