1

I have figured, hopefully, out the volume of the figure of $$(y-x^x)(y-x^{-x})=0\le x\le 1$$ about the line y=1.

This volume came out to be $$V=2π\int_0^1 (1-x)(x^{-x}-x^x)dx= \boxed{-2π\sum_{n=1}^∞\sum_{k=1}^{n-1} \binom{n-1}{k-1}\frac{(-1)^{k+n}+(-1)^n}{k^n}}$$$$=1.774473…$$

All of the context and work is here.

The question is from that the summand does not have the denominator of $n^n$ or another variable which would have made it quite difficult to put into another form except at the upper values for k.

A desired form for this integral is required because this would possibly create a closed form of this integral. In addition, this sum is a little hard to understand which is why an alternate form may make it easier to read.

What is another form of this integral and sum which can be simplified and possibly put in terms of special functions? A closed form would be great, but still optional. Please do not include another integral form as that would be redundant. Anything except integral representations works. Finally, please give me feedback and correct me please!

Тyma Gaidash
  • 12,081
  • 1
    Vague Questions invite vague Answers. While it may be clear to you what (beside "self-advertising") this asks for, you have the responsibility for making it clear to your Readers. – hardmath Apr 22 '21 at 21:58
  • @hardmath How would you suggest this question may be improved? The purpose of this question is to find a variety of answers for what an alternate form of the summation may be. – Тyma Gaidash Apr 22 '21 at 23:04
  • It would help to include context about why alternate forms are expected (or desired?). An example of such an idea might be that you formulated the solid of revolution by a disk/washer method, and that a shell method may give the same value in an alternate expression. I would not know from the present wording of the Question if such an approach would be appropriate. – hardmath Apr 23 '21 at 01:00
  • @hardmath Please click the link! It has all necessary information. – Тyma Gaidash Apr 23 '21 at 12:01
  • I think there may be an error. How does one sum over $k$ twice? I think the first summation should index at $n$, but even with that change, Mathematica gives me a different value. – DecarbonatedOdes Apr 25 '21 at 02:01
  • @DecarbonatedOdes this is called a double sum and is a infinite sum of sums with the inner one having bounds from k=0 to the value of n for which the outer sum is at. I could do another demo if you want.

    Here is proof:formula proof.

    Also, please click the link in my answer and show me your mathematica proof please!

    – Тyma Gaidash Apr 25 '21 at 11:40
  • 1
    I know what a double sum is. Thank you, but the error is there, not in your work, but what you have posted here. You have: $$ -2 \pi \sum _{k=1}^{\infty} \left(\sum _{k=1}^{n} \frac{\left((-1)^{k+n}+(-1)^n\right) \binom{n-1}{k-1}}{k^n}\right) $$ when it should be: $$ -2 \pi \sum _{n=1}^{\infty} \left(\sum _{k=1}^{n-1} \frac{\left((-1)^{k+n}+(-1)^n\right) \binom{n-1}{k-1}}{k^n}\right) $$ I find it astonishing one can come up with that kind of result and miss that or not know what an index is. – DecarbonatedOdes Apr 26 '21 at 06:14
  • @DecarbonatedOdes Thanks for the clarification! I have now updated the answers. You can also see in my link that the correct answer was already posted! – Тyma Gaidash Apr 27 '21 at 12:53
  • @TymaGaidash: If this is to be rotated about the $y$-axis (i.e., the line $x=0$) the radius should be $x$ and not $1-x$. – user26872 May 11 '21 at 19:03
  • 1
    @user26872, yes checking Paul’s notes makes this true. This will then just be about the line y=1. However, I do need to update the source link. Thanks for the catch! – Тyma Gaidash May 11 '21 at 19:24

1 Answers1

1

We'll show $$ V=-2\pi\sum_{n=1}^\infty\sum_{k=1}^{n-1}{n-1\choose k-1}\frac{(-1)^{k+n}+(-1)^n}{k^n} =4\pi\sum_{r=1}^\infty\left(\frac{1}{(2r)^{2r}}-\frac{1}{(2r+1)^{2r}}\right). $$ First observe that for odd $k$ the terms vanish because $(-1)^{k+n}+(-1)^n=0$. Hence we can restrict the sum to even $k=2r$: $$ V=-2\pi\sum_{n=1}^\infty\sum_{r=1}^{\left[\frac{n-1}{2}\right]}{n-1\choose 2r-1}\frac{2\,(-1)^n}{(2r)^n}. $$ Now, rearrange the summation $$1\le n,\quad 1\le r\le\frac{n-1}{2}\quad\Longleftrightarrow\quad1\le r,\quad2r+1\le n$$ to get $$ V=-4\pi\sum_{r=1}^\infty\sum_{n=2r+1}^\infty{n-1\choose 2r-1}\left(-\frac{1}{2r}\right)^n=-4\pi\sum_{r=1}^\infty\sum_{n=2r}^\infty{n\choose 2r-1}\left(-\frac{1}{2r}\right)^{n+1}.\qquad(*) $$ The next step will evaluate the inner sum to a closed expression. For doing that, we start with $$ \sum_{k=0}^\infty x^k=\frac{1}{1-x},\quad(|x|<1). $$ Differentiating $a$ times by $x$ gives $$ \sum_{k=0}^\infty k(k-1)\ldots(k-a+1)x^{k-a}=\frac{a!}{(1-x)^{a+1}},\quad(|x|<1). $$ Multiplying this identity with $\frac{x^a}{a!}$ and observing that the terms in the sum for $0\le k<a$ vanish, we get $$ \sum_{k=a}^\infty {k\choose a}x^{k}=\frac{x^a}{(1-x)^{a+1}},\quad(|x|<1). $$ That is almost what we need for $\sum_{n=2r}^\infty{n\choose 2r-1}\left(-\frac{1}{2r}\right)^{n+1}$. Multiplying with $x$ and subtracting the first term completes it: $$ \sum_{k=a+1}^\infty {k\choose a}x^{k+1} = \sum_{k=a}^\infty {k\choose a}x^{k+1}-x^{a+1}=\frac{x^{a+1}}{(1-x)^{a+1}}-x^{a+1},\quad(|x|<1). $$ Setting $a=2r-1$, $k=n$, $x=-\frac{1}{2r}$, we find that indeed $|x|\le\frac{1}{2}<1$, and $$ \sum_{n=2r}^\infty{n\choose 2r-1}\left(-\frac{1}{2r}\right)^{n+1}= \frac{\left(-\frac{1}{2r}\right)^{2r}}{\left(1+\frac{1}{2r}\right)^{2r}}- \left(-\frac{1}{2r}\right)^{2r}=\frac{1}{(2r+1)^{2r}}-\frac{1}{(2r)^{2r}}. $$ Inserting this in (*), we get the simplified expression $$ V=4\pi\sum_{r=1}^\infty\left(\frac{1}{(2r)^{2r}}-\frac{1}{(2r+1)^{2r}}\right), $$ which converges very rapidly: $$ \begin{array}{cll} r&\frac{1}{(2r)^{2r}}-\frac{1}{(2r+1)^{2r}}&\mbox{partial sum up to $r$}\\ 1&0.138888888888889&0.138888888888889\\ 2&0.00230625&0.141195138888889\\ 3&1.29336107552305\cdot10^{-5}&0.141208072499644\\ 4&3.63740716499719\cdot10^{-8}&0.141208108873716\\ 5&6.14456710570468\cdot10^{-11}&0.141208108935161\\ 6&6.92346723666235\cdot10^{-14}&0.141208108935231\\ 7&5.57378713899953\cdot10^{-17}&0.141208108935231 \end{array} $$ giving $V=4\pi\cdot0.1412081089352\ldots=1.77447343063\ldots$.

Uwe
  • 721
  • Amazing! It is weird how my 2 bounties each got answered 3 days or less before expiration when they are the top of the front page in the questions tab. I have always wanted to see this strategy in action. I may have to update my page with these results and maybe even use this to find the general formula of this surface about the y=a from 0 to b or from 0 to 1 at least. This is very useful for this simplification to simply a difference of Sophomore’s Dream related summations. Please tell me if there is a way to simplify this even further! – Тyma Gaidash May 13 '21 at 23:37
  • @TymaGaidash: The shift in the index bound is achieved by separating the first term from the sum. It is just needed to obtain the exact form matching the inner sum. I tried to find also further simplifications but since there are terms involved with increasing powers it seems unlikely to me that there is one. – Uwe May 14 '21 at 01:10