I have seen this sum a couple of times now in the textbook and in the problems book but I don't know how to solve it. I know this might be more suited for math.stackexchange but since this is a fairly common sum and is related I'm pretty sure someone can help. Thank you!
-
I'm not sure I undestand what you need help with. The equation in the red box is equal to $\alpha^n(1+1/\alpha+1/\alpha^2 + \ldots + 1/\alpha^n)$. Is that helpful? – MBaz Oct 16 '17 at 18:58
2 Answers
\begin{align} (1-x)\sum_{k=0}^{n}x^k &= \sum_{k=0}^{n}x^k - x\sum_{k=0}^{n}x^k \\ \\ &= \sum_{k=0}^{n}x^k - \sum_{k=1}^{n+1}x^k \\ \\ &=1 + x+x^2+ \dots+x^n \ \\ & \qquad -x-x^2-\dots-x^n-x^{n+1}\\ \\ &= 1 - x^{n+1} \end{align}
Thus if $x \ne 1$, dividing both sides by $(1-x)$ results in $$\sum_{k=0}^{n}x^k = \frac{1-x^{n+1}}{1-x}$$
Replace $x = \alpha^{-1}$, it becomes $$\sum_{k=0}^{n}\alpha^{-k} = \frac{1-(1/\alpha)^{n+1}}{1-1/\alpha}$$
- 20,661
- 4
- 38
- 76
- 6,585
- 2
- 21
- 38
-
it seems to me that this answer suffices to answer the question. (but this summation is in the calculus textbook. any college sophomore should know it.) it should get the check mark. – robert bristow-johnson Oct 17 '17 at 00:10
@AlexTP provides the proof for the summation from $k=0$ to $k=n$. Let me put the more general case: Assume that the integers $a,b$ are finite, then for any $\beta$ we have $$ \sum_{k=a}^{b} \beta^k = \frac{ ~~\beta^a - \beta^{b+1} }{ 1 - \beta} $$
When any of the summation limits $a$ or $b$ are infinite, then a limiting process should be considered and we have: $$ \sum_{k=a}^{\infty} \beta^k = \frac{ ~~\beta^a }{ 1 - \beta} $$ for $|\beta| < 1$ , or else it won't converge. And for the infinity at the lower limit we have:
$$ \sum_{k=-\infty}^{b} \beta^k = \frac{ - \beta^{b+1} }{ 1 - \beta} $$ for $|\beta| > 1$ , or else it won't converge.
Applying to your case with $a=0$, $b=n$ and $\beta = \alpha^{-1}$ :
$$ \alpha^{n} \sum_{k=0}^{n} (\alpha^{-1})^k = \alpha^{n} \frac{ 1 - \alpha^{-n-1} }{1 - \alpha^{-1} } = \frac{ \alpha^{n+1} - 1 }{\alpha - 1 }= \frac{ 1 - \alpha^{n+1} }{1 - \alpha } $$
- 28,152
- 3
- 24
- 50
-
1Does the first summation apply for $\beta=1$? It looks like it should be equal to $b-a+1$, assuming $b \geq a$, but that's not obvious from the RHS. – MBaz Oct 16 '17 at 21:31
-
neither of them shall be used for $\beta=1$ . However for the first sum you can get the correct result of course using the limit approach as follows: $$\sum_{k=a}^{b} 1^k = \lim_{x \to 1} \frac{x^a - x^{b+1}}{1-x}$$ Evaluated using L'hopital approach as: $$\sum_{k=a}^{b} 1 = \lim_{x \to 1} \frac{a x^{a-1} - (b+1)x^b}{-1} = b+1-a$$ – Fat32 Oct 16 '17 at 22:12
-
Thanks. The reason for my comment is that you state that the summation formula is valid for "any $\beta$". – MBaz Oct 16 '17 at 22:21
-
but yes the 1st finite length sum is (therefore) valid for any $\beta$ including $\beta=1$ ? – Fat32 Oct 16 '17 at 22:24
-
As I said in my first comment, it wasn't obvious to me that it's valid for $\beta=1$, so thanks for showing that it indeed it is. – MBaz Oct 16 '17 at 22:37
-
haaa I didn't get that from your comment, so now it's ok... btw from a pure algebraic point of view, the RHS fraction cannot be used to compute the value of the sum as the fraction becomes indeterminate form of $0/0$ at $x=1$. However it's calculus that lets us use fraction formlula with the limit approach without paying attention to the intermediate stages of division by zero etc... btw you know all these things why do you ask anyway ? ;-) – Fat32 Oct 16 '17 at 22:52
-
1:-D Sorry if I came across the wrong way... I wanted to point out a way to improve your answer, and maybe get you to work out the details without me having to ;) – MBaz Oct 16 '17 at 23:06
-
4It is silly to work out the value of $$\sum_{k=a}^b \beta^k$$ when $\beta = 1$ by starting from the fancy formula and using L'Hopital's rule etc. More directly, $$\sum_{k=a}^b \beta^k\big|{\beta = 1}=\sum{k=a}^b 1^k = \sum_{k=a}^b 1 = b-a+1,$$ no muss, no fuss. – Dilip Sarwate Oct 17 '17 at 02:45
