Talk:Ramanujan summation


 * $$2+3+5+7+11+\cdots = 24133 (\Re)$$

holds as well, but I haven't any article to take citation from. It is P(-1), where P(x) is Prime zeta function (http://mathworld.wolfram.com/PrimeZetaFunction.html).

This is hard to believe. Sums of integers are usually 0 (e.g. 1+4+9+16+...), fractions (e.g. 1+2+3+4+...) or irrational (e.g. 1+2+6+24+...). How did you arrive at that specific integer? I computed a value of 2.92... for the sum of prime numbers. Doubledork (talk) 22:03, 30 July 2009 (UTC)

Explanation?
Does anybody know a rationale to how this sum is figured? Thanks. Root4(one) 05:14, 5 June 2007 (UTC)

I consider a Ramanujan sum as the finite term of a polynomial of the surreal infinity ω. I could get into OR about this, but for Wikipedia purposes, has anyone else considered such a definition? Collin238 16:55, 27 September 2007 (UTC)

Not invented by Ramanujan
Ramanujan did not invent the technique as this article claims. Euler used it long before him. Look at E55 ("Methodus universalis series summandi ulterius promota") and E352 ("Remarques sur un beau rapport entre les series des puissances tant directes que reciproques ") here http://www.math.dartmouth.edu/~euler/ —Preceding unsigned comment added by 209.67.107.10 (talk) 00:01, 13 June 2009 (UTC)

Another way to calculate Ramanujan summation
I found another way to calculate Ramanujan summation:

Suppose we want to calculate $$f\left( 1 \right)+f\left( 2 \right)+f\left( 3 \right)+\ldots $$.

Let $$g\left( x \right)$$ be an analytic continuation of $$\sum\limits_{m=1}^{x}{\int\limits_{0}^{m}{f\left( t \right)dt}}$$

Then $$f\left( 1 \right)+f\left( 2 \right)+f\left( 3 \right)+\ldots =-{g}'\left( 0 \right) (\Re)$$.

This method agrees with the regular Ramanujan summation when $$f\left( x \right)$$ is a polynomial, and this method is also linear, so I suppose it agrees with the regular Ramanujan summation in every series.

77.126.36.120 (talk) 14:32, 30 September 2009 (UTC)


 * Nice, great, but can you deliver proof of this result? If you can so, I'll suggest writing paper rather than forcing solution in talk page of wikipedia article. 82.160.31.58 (talk) 20:56, 8 July 2023 (UTC)

Ramanujan summation not unique?
Let R*(x) = R(x) + p(x), where p(x) is a 1-periodic function whose integral from 1 to 2 is 0 (such a function could be, e.g. p(x) = sin(2 pi x)) Then the integral of R*(x) from 1 to 2 is 0 by linearity of the integration operator, and R*(x) also solves the functional equation. Thus the integral does not guarantee a unique solution and the statement it does is wrong. If there are additional criteria that guarantee the uniqueness, they should be mentioned here. mike4ty4 (talk) 23:20, 17 October 2009 (UTC)


 * Integral of R*(x) from 1 to 2 is equal ot integral of R(x) over this interval, not 0, unless you precise what do you mean by R(x).
 * R*(x) can solve some functional equations, unless you specify what do you mean. In the article there are no functional equations for whicch this statement makes sense imo.
 * 82.160.31.58 (talk) 20:54, 8 July 2023 (UTC)

+ + + +… or + - + -…  ?
"For even powers we have:
 * $$1 + 2^{2k} + 3^{2k} + \cdots = 0\ (\Re)$$"

but 1_−_2_%2B_3_−_4_%2B_·_·_· says: For even n, this reduces to
 * $$1-2^{2k}+3^{2k}-\cdots = 0.$$

What's right? 46.115.58.168 (talk) 01:43, 30 August 2012 (UTC)


 * they are BOTH right remember that $$ \eta(s)= (1-2^{1-s}) \zeta (s) $$ so they share some zeros. — Preceding unsigned comment added by 85.85.12.24 (talk • contribs) 19:15, 11 September 2013‎

They are both wrong, Ramanujan made a mistake, simple as that. Even in the context of Ramanujan Summation, only the 1-1+1- ... = 1/2 and 1-2+3-4 ... = 1/4 are 'correct'. The others do not follow the context and ignore zeroes, and are hence incorrect completely. Ramanujan Summation refers to the impact of the series, and hence zeroes cannot be ignored : http://rahulraj-says.blogspot.in/2014/01/the-power-of-zero.html — Preceding unsigned comment added by 120.61.22.134 (talk) 17:35, 19 January 2014 (UTC)

Not very comprehensible.
It doesn't bother anybody else that this article is more or less incomprehensible? I have no idea what a Ramanujan sum is, how it's calculated, or why it's interesting, and I'm not likely to learn it from the article, since the very first equation has two variables (Bk+1 and Rp) that are not defined anywhere in the article Geoffrey.landis (talk) 21:31, 18 January 2014 (UTC)

Yes, it is too technical to understand unless you are already well versed in calculus and the summations of infinite series. I have added a flag to the top of the article. YouTube videos from Numberphile and Mathologer on the "1 + 2 + 3 . . ." series can provide insight to those less thoroughly versed in the specific symbols and notations used here.Kemery720 (talk) 07:22, 5 March 2018 (UTC)

Recent edits citing "Rahul Raj" are not from academic sources
These edits should be reverted; one look at Rahul Raj's cited blog post reveals that he's an amateur (not a bad thing in itself) and doesn't know or understand things beyond Cesaro summation. — Preceding unsigned comment added by Innerproduct (talk • contribs) 22:07, 19 January 2014 (UTC)

For someone who claims to be a mathematics graduate, Innerproduct does not understand the difference between Dirichlet series and the zeta function. Reverting reference to "Rahul Raj" as it provides a more coherent explanation of the Cesaro sum, which is all that is needed for this Summation. — Preceding unsigned comment added by 123.63.97.113 (talk) 18:01, 20 January 2014 (UTC)

Why are the edits "Bogus"? The linked text is amateurish but correct. This particular page on Ramanujan Summation is being quoted as proof that the sum of the infinite series 1+2+3+4+...= - 1/12. Ramanujan in the first reference quoted does not provide any proof of the same. The page should be edited to reflect that the claim is incorrect and Inconsistent and students should not follow the method cited. — Preceding unsigned comment added by 123.63.97.142 (talk) 02:16, 20 January 2014 (UTC)

For the record, I didn't add any changes or revert any of the changes made by others. I merely pointed out (on this talk page) that references to the substandard blog post by Rahul Raj are not reliable and should be reverted. Here are three far better sources (all of them by Terry Tao): (1.) http://terrytao.wordpress.com/2010/04/10/the-euler-maclaurin-formula-bernoulli-numbers-the-zeta-function-and-real-variable-analytic-continuation/ (2.) https://plus.google.com/114134834346472219368/posts/ZuJDv3daT9n, and (3.) http://terrytao.wordpress.com/career-advice/there%E2%80%99s-more-to-mathematics-than-rigour-and-proofs/. -- Innerproduct (talk) 18:43, 20 January 2014 (UTC)

One should also note that there are multiple ways of "summing" infinite series the usual epsilon-delta definition or the slightly less well known Cesaro summation are just two such ways. The answer by Sridhar Ramesh at https://www.quora.com/Mathematics/Theoretically-speaking-how-can-the-sum-of-all-positive-integers-be-1-12 clarifies this point in a relatively easy to understand manner. -- Innerproduct (talk) 19:46, 20 January 2014 (UTC)

I also wish to publicly thank User:NorthAce. -- Innerproduct (talk) 20:11, 20 January 2014 (UTC)

The Ramanujan summation of the three series mentioned in this article is in fact correct (see above http://terrytao.wordpress.com/2010/04/10/the-euler-maclaurin-formula-bernoulli-numbers-the-zeta-function-and-real-variable-analytic-continuation/). That's why the symbol $$(\Re)$$ is used here. I calculated the sums myself via the Euler-Maclaurin Method mentioned above. The result coincides with the one from Terry Tao's method of smoothed asymptotics. Therefore again, I reverted the statement that the Ramanujan summation is wrong. Conversely: If properly defined Ramanujan summation is correct and the value obtained is a property of the partial sums of a series. However, it does not equal the sum of a series in the sense of absolute convergence – which this article does not state anyway, thus the symbol $$(\Re)$$ was used (I personally, for reasons of formality would prefer to put the Ramanujan R to the equality sign, but i am not sure if there exists a general convention and what it is). The blog post by Raj Rahul does not cover the real problem here (the confusion of equality with the process of Ramanujan summation) but instead gives arguments, why it is not Césaro summable. This is not even disputed, the series 1+2+3... is not Césaro summable but it is Ramanujan summable. I conclude that this paragraph would need serious rewriting but for the moment it is sufficient to point out that the Ramanujan summation is not the sum of the series and cite Terry Tao's article. NorthAce (talk) 13:55, 21 January 2014 (UTC)

Thanks User:NorthAce for the explanation. Can you share the derivation either here or through a web link? The formal derivation should add value to the article. — Preceding unsigned comment added by 115.112.84.23 (talk) 14:41, 21 January 2014 (UTC)

The formal derivation of the Ramanujan sums can be taken from Terry Tao's article http://terrytao.wordpress.com/2010/04/10/the-euler-maclaurin-formula-bernoulli-numbers-the-zeta-function-and-real-variable-analytic-continuation/. NorthAce (talk) 19:07, 21 January 2014 (UTC)

Important note
This article needs serious revision Infinite Geometry (talk) 03:58, 17 February 2024 (UTC)


 * I agree. Is there a way to flag this up to the editors? Blitzer99 (talk) 15:33, 22 April 2024 (UTC)