A formula that I learnt in school to sum an infinite geometric series with a common ratio between \(-1\) and \(1\) is,

\(\displaystyle \sum _{n=0}^{\infty } a r^n=\frac{a}{1-r}\text{ if }|r| < 1\)

and the proof is as given below:

Take \(S=a+a r+a r^2+a r^3+\text{...}\)

then \(r S=a r+a r^2+a r^3+a r^4+\text{...}\)

subtracting these two we get

\((1-r) S=a\)

Therefore, \(S=\displaystyle \frac{a}{1-r}\)

My question is: why isn't this formula valid for all \(r\) excluding \(1\)?

## Comments

Sort by:

TopNewestWell, an infinite summation doesn't make sense in itself. It must have a definition, and usually we take \(\sum_{i=0}^\infty f (i)=\lim_{n\to\infty}\sum_{i=0}^n f (i)\) which leads us to the \(a^{big}\to0\) argumentation.

Regarding your demonstration, again, we have to take the infinite summation as a finite one: then if we set \(S (n)=a+ar+\dots+ar^n\) we get \(rS (n)=ar+ar^2+\dots+ar^{n+1}\) and so \(S(n)(1-r)=a-r^{n+1}\); if we take this to the limit, if \(|r|<1\) then again \(r^n\) converges to 0, otherwise it diverges to infinity.

For example: let \(S=1+2+4+8+\dots\); then \(2S=2+4+8+\dots\) and subtracting we'd get \(S=-1\), which obviously is absurd.

Another example: how do you evaluate \(S=1-1+1-1+\dots\)? you could say that \(S=(1-1)+(1-1)+\dots=0\), or \(S=1+(-1+1)+(-1+1)+\dots=1\), or even \(S=1-S\), which is \(S=1/2\).

Just to say that you must be VERY careful when you sum something to infinity! – Riccardo Zanotto · 3 years ago

Log in to reply

this limit doesn't equal \[\frac{1}{4}\] , Its obvious when you check the pure definition of the limit of a sequence. I think that limit here doesn't exist. – Ahmed Taha · 3 years ago

Log in to reply

– Riccardo Zanotto · 3 years ago

Yes, of course it doesn't exist... in fact the three "values" are there to show that summing to infinity without care is really dangerous.Log in to reply

See here.Grandi's sum actually has no sum, but it

It doesn't have a value in the usual sense.But by Cesaro summing it you will arrive at \(\frac{1}{2}\).should be \(\frac{1}{2}\)– Bogdan Simeonov · 3 years agoLog in to reply

shouldbe \(1\). – Riccardo Zanotto · 3 years agoLog in to reply

– Bogdan Simeonov · 3 years ago

ExactlyLog in to reply

Also, I tried \(\zeta (-1)\) (The riemann-zeta function) in mathematica; it returned \(\displaystyle \frac{-1}{12}\)

\(\displaystyle \zeta (s)=\frac{1}{1^s} + \frac{1}{2^s} + \frac{1}{3^s} + ...\)

so \(\displaystyle \zeta (-1)=1+2+3+... = -\frac{1}{12}\) This seems so bizarre that my head hurts. :P – Danish Mohammed · 3 years ago

Log in to reply

check here – Bogdan Simeonov · 3 years ago

I have written a post about itLog in to reply

– Sharky Kesa · 3 years ago

If you want more about infinit geometric sequences, read my summation notes (summation, summation part two and summation part two method two).Log in to reply

– Eddie The Head · 3 years ago

I was referring to this 1 and -1 fallacy ....... Nice post.....Log in to reply

The proof may also be given as the following in which your answer becomes clear:

Sum of the infinite geometric series can be taken to be the sum as number of a geometric series whose tend to \(\infty\).

\(\displaystyle \large S_n=\displaystyle\lim_{n\to\infty} \frac{a(1-r^{n})}{1-r}\)

With \(n\to\infty\) the above limit exists only if \(r<1\) as then \(r^{something \hspace{1mm} big}=0\) and the numerator becomes 1. So the result follows. – Krishna Jha · 3 years ago

Log in to reply

– Eddie The Head · 3 years ago

Exactly,..that is the interval of convergence...Log in to reply

If r is greater than 1, then clearly the sum becomes diverging,....yes in that case also it seems that we can apply this formula but clearly that finite answer has no meaning because as r tens to infinity the terms of the sequence becomes incresingly large and the whole expression that is the sum of these terms become infinity.....you can notice another interesting fallacy when r =1...in that case writing the sum in 2 different gives gives two different answers..... – Eddie The Head · 3 years ago

Log in to reply

– Eddie The Head · 3 years ago

Also the expression infinity -infinity has no meaning....this crops up when r is greater than 1....but on the contrary in the convergence interval as the sequence is decresing we can say that after infinite steps all the terms will become zero and hence the sum becomes finite and hence we can subtract these two numbers...I would love to hear what others say about this,.,,Log in to reply