4

I have been asked to calculate a general formula for $1 + b + 2b^2 + 3b^3 + ... + Nb^N$

I think that a part of this is a geometric sequence, and I have rewritten this as

$f(x) = 1 + \sum_{i=1}^n i\cdot x^i$

(I couldn't figure out a way to make the first term be 1 with the progression so I just removed it from the series instead)

When I look on Wolfram Alpha it says that the partial sum formula for $ \sum_{i=1}^n i\cdot x^i$ is:

$$\sum_{i=1}^n i\cdot x^i = \frac{(nx-n-1)x^{n+1}+x}{(1-x)^2}$$

On this question, an answer said that the general formula for the sum of a finite geometric series is:

$$\sum_{k=0}^{n-1}x^k = \frac{1-x^n}{1-x}$$

But if I substitute my ($i\cdot x^i$) into the formula mentioned above I don't get the same value as the one Wolfram gives me.

Two questions:

  1. Is it correct to pull the first term out of the series so it becomes a geometric series, or is there another way?
  2. How did Wolfram Alpha calculate that expression?
Max
  • 1,039
Joon
  • 157
  • 1
  • 6
  • 5
    The ratio of successive terms is not constant, so this is not a geometric sequence, though it may be related to one – J. W. Tanner Mar 17 '19 at 13:54
  • Thank you @J.W.Tanner, I will edit the question – Joon Mar 17 '19 at 13:59
  • 1
    Ok; by the way, I think we meant series (sum) rather than sequence – J. W. Tanner Mar 17 '19 at 14:02
  • 2
    This page could be helpful: https://math.stackexchange.com/questions/2782812/finite-sum-kxk. Or you can let $S = x + 2x^2 + 3x^3 + \cdots +nx^n$, and consider $S - xS$, and make sure you know the geometric series formula. – Minus One-Twelfth Mar 17 '19 at 14:06

5 Answers5

11

The extra factor of $i$ or $k$ or whatever you call it means we no longer have a geometric series, but we can get one by differentiating with respect to $x$. You'll want to check powers of $x$ carefully too.

J.G.
  • 115,835
6

Hint: Think derivatives w.r.t. $b\,$:

$$1 + b + 2b^2 + 3b^3 + \dotsm + Nb^N=b\sum_{k=1}^{N}k\,b^{k-1}=b\Bigl(\sum_{k=0}^{N}b^k\Bigr)^{\!\!'}$$

Bernard
  • 175,478
6

Your presumptions are wrong. The first series is not a geometric series. and substituting $i*x^i$ into $\sum_{k=0}^N x^k$ is not the same as $\sum_{i=0}^Ni*x^i$.

Try to solve the problem using the principle of induction:

Let us assume the above relation holds for $n$: $\sum_{i=0}^n ix^i=\frac{(nx-n-1)x^{n+1}+x}{(1-x)^2}\equiv=A_n$.

Now let us show it must also holds for $n+1$ and therefore for all $n\in\mathcal{N}$. Thus,

\begin{align} \sum_{i=0}^{n+1}ix^i&=\underbrace{\sum_{i=0}^{n}ix^i}_{=A_n}+(n+1)x^{n+1}\\ &=A_n+(n+1)x^{n+1}\\ &=\frac{(nx-n-1)x^{n+1}+x}{(1-x)^2}+(n+1)x^{n+1}\\ &=\frac{(nx-n-1)x^{n+1}+x}{(1-x)^2}+(n+1)x^{n+1}\frac{(1-x)^2}{(1-x)^2}\\ &=\frac{(nx-n-1)x^{n+1}+x+(n+1)x^{n+1}*(x-1)^2}{(1-x)^2}\\ &=\frac{1}{(1-x)^2}\biggl((nx-n-1)x^{n+1}+x+(n+1)x^{n+1}(1+x^2-2x)\biggr)\\ &=\frac{1}{(1-x)^2}\biggl(x^{n+2}((n+1)x-n-2)+x\biggr)\\ &=\frac{1}{(1-x)^2}\biggl(x^{(n+1)+1}((n+1)x-(n+1)-1)+x\biggr) \end{align}

So if it holds for any $n$ it must hold for $n+1$ too and therefore for all natural numbers. Now, you only have to check that it holds for $n=0$ or $n=1$ and you are done.

3

Start with $$S=1 + b + 2b^2 + 3b^3 + \cdots + Nb^N$$

Multiply by $b$ and then subtract to give $$bS=b + b^2 + 2b^3 + 3b^4 + \cdots +(N-1)b^N + Nb^{N+1}$$ $$(b-1)S = -1 - b^2 - b^3 - b^4 - \cdots -b^N + Nb^{N+1}$$

Do that again $$b(b-1)S = -b - b^3 - b^4 - b^5 - \cdots -b^N-b^{N+1} - Nb^{N+2}$$ $$(b-1)^2S = 1-b +b^2 -(N+1)b^{N+1} + Nb^{N+2}$$

Finally simplify $$S = \dfrac{1-b +b^2 -(N+1)b^{N+1} + Nb^{N+2}}{(b-1)^2}.$$

As an example, with $b=2$ and $N=3$ the first line gives $S=1+2+8+24=35$ while the final line gives $S=\frac{1-2+4-64+96}{1^2}= 35$

Henry
  • 157,058
0

Here's what I did to get the formula for partial sums of this series:

It was too much to type in LaTeX, so just did it on paper. Hope you don't mind.

KKZiomek
  • 4,012
  • Thank you! Could you please explain how you know to add the (1 - x) / (1 - x) to the series? – Joon Mar 17 '19 at 16:31
  • 1
    @Joon it's just a clever trick. There's really no reason why would one multiply and divide by (1-x). I guess it just works :) – KKZiomek Mar 17 '19 at 17:15
  • 1
    If you analyze it closely you see that it cleverly cancels out like terms in a fashion that creates a geometric sequence. So I guess to know to multiply by (1-x) you have to observe and be clever. I didn't come up with this method. I found it in some paper about sequences long time ago – KKZiomek Mar 17 '19 at 17:17
  • 1
    Thank you again, I could understand your method and follow along. Really appreciate it – Joon Mar 17 '19 at 17:23