Riemann Rearrangement Theorem
Aryan Das
30 October 2024
Probably YouTube has done more harm to the maths community than you know
— Anant Mudgal
Remark 0.1. Guys I will try to keep this post as intuitive as possible, so if you find it a
less rigorous, orzness on you!
§1 Introduction
Do you remember the famous sum result below?
Proposition 1.1 (Ramanujan Sum)
−1
1 + 2 + 3 + ··· + ∞ =
12
Let’s first prove the above prove:
Proof. Consider
1
A = 1 − 1 + 1 − 1 + · · · = 1 − (1 − 1 + . . . ) =⇒ 2A = 1 =⇒ A =
2
B = 1 − 2 + 3 − 4 + ...
Note that A-B gives:
1
A − B = (1 − 1) + (−1 + 2) + · · · = B =⇒ A = 2B =⇒ B =
4
S = 1 + 2 + 3 + ...
Note that B-S gives us:
−1
B − S = (1 − 1) + (−2 − 2) + (3 − 3) = · · · = −4S =⇒ S =
12
Wow, we have such an amazing and unbelievable result. The sum of infinite positive
integers is a negative fraction... Damn!
1
Aryan Das — 30 October 2024 Riemann Rearrangement Theorem
§2 Another Example
Note this now:
§3 Take backs
Well these are such mind boggling results but sorry to disappoint you, these results are
absolute bullshit. In this post, I will try to make sense of why what I said is true. This
will follow from a theorem I give below which is essentially the title of the post. But
before that I will make some definitions and explanations.
§4 Definitions and Explanations
§4.1 Existence of sum
P
Definition 4.1. If the limit ofPthe nth partial sum Sn of the infinite series an exists
and equals S, then we say an converges and its sum is S. If, as n approaches
infinity,
P the limit of the nth partial sum Sn does not exist, then we say the series
an diverges and has no sum.
Let me explain it to you. Consider you have a geometric series:
∞
X 1
2n
n=1
1
Obviously the answer by the GP formula is = 1. But I would like you to think about
2
1− 12
Pn n
this sum without the formula. Consider Sn = i=1 an , where an = 12 . Note that how
much of a large n you take, you can never quite reach 1. Like take the first trillion values
2
Aryan Das — 30 October 2024 Riemann Rearrangement Theorem
of Sn and each time you will find it < 1. So you might be wondering, how does it go to
1? Well the thing is simple, you are not adding all the terms. Even after you are adding
a billion or even googol values, you will still be remaining with some more. But then you
may ask, how can we add an infinite number of values? Turns out we cant and hence we
invent methods, one of which is the limit method. Note that for a general n the formula
for Sn is:
n i
X 1 2n − 1 1
Sn = = n
=1− n
2 2 2
i=1
But note that it seems like as we get closer to infinity, Sn seems to approach one. So it
will be plausible for the original infinite sum to be 1 by:
∞ n n i
X 1 X 1 1
= lim = lim 1 − n = 1.
2 n→∞ 2 n→∞ 2
i=1 i=1
Now that this thing is clear to you, definition also might be a piece of cake to understand
intuitively.
First of all, I hope till now it is clear to you what a partial sum is, but still let me define
it for you:
Definition 4.2. If a1 , a2 , a3 , · · · ∈ R, then the partial sum Sn = ni=1 ai
P
What the definition says is simple now. That if there exists a real M such that M > Sn
for all n, then the sum of the reals is M and is now a convergent series. If such an M does
not exist, we now call it divergent and it has no sum. Well also M has to be the smallest
such real as there exists many reals say X > M , which also satisfies the condition, but
the sum of the series is not X.
§5 Break and notes
Uff that might have been heavy, but heavier stuff is underway. Before that I would like
to bring into your notice, that there are many tests, that tell us if a sum is convergent
or not. That is beyond the scope of this post, but you can read them independently
whenever you want. Now let’s get back.
§6 2 types of infinite sum
Theorem 6.1
P P
If i=1 |ai | has a sum, then so does i=1 ai , where the vice versa is not necessarily
true. Well when the vice versa is true, then we call it absolute convergence and
when it is not true, we call it conditional convergence
I hope you got the theorem. I am not stating the proof but from here magic starts. The
next theorem by Riemann is magical.
§7 Riemann Rearrangement Theorem
Well this has 2 parts but we will only focus on the second part, because the first part is
not needed with the context of the post.
3
Aryan Das — 30 October 2024 Riemann Rearrangement Theorem
Theorem 7.1 (Riemann Rearrangement Theorem)
P
If an is a conditionally
Pconvergent series and let A be a real number. Then there
exists a permutation of an which converges to A.
The statement is not that hard to understand, but intuitively it makes 0 sense. What
this actually means is that for any conditionally convergent sum, we can do some clever
algebra to get it to converge to any real we want. Like as some clever algebra showed
us that the sum of the real numbers is −112 . So this theorem even goes as far as saying
that we can do some algebra and even prove that the sum of all the natural numbers is 0.
Crazy right?
§8 Conclusion
Well this was nothing. Higher maths is actually much much much more crazier. For
example look at the proof of Miklós Laczkovich, to the irrationality of π. If you want
to know more about it, it would have to wait for another post. Thanks for reading.
Sayonarra!
Remark 8.1. Guys I have the proof of the Riemann Theorem stated above, so if anyone
wants to get a good look at it, please dm me.
