2

In solving this problem, I come up with the following integral:

$$\int_{-1/(4\pi)}^0\frac{(s\log{(-4\pi s)})^{(2+n)/2}}{s^2}ds$$

where $n=1,2,3...$

By using Mathematica, I could get that the integral is equal to $$\frac{8\pi^{-n/2}\Gamma(2+n/2)}{2^{1+n/2}n^{(4+n)/2}}$$

My question is: How do we calculate it by hand? Is there any clever way to do it?

Tomás
  • 22,559
  • 1
    It seems like you can factor out a lot of constants for starters - that may make things easier. – preferred_anon Oct 15 '13 at 23:37
  • After throwing away the constants, you're left with $$\int_{-1/(4\pi)}^0!!!\frac {(s\log (-4\pi s))^{1+n/2}}{s^2},ds.$$ Still not too pretty – dfeuer Oct 15 '13 at 23:46

2 Answers2

5

Rewrite your integral as $$\frac{(2n)^{1+n/2}}{2+n}\int_{-1/(4\pi)}^{0}s^{-1+n/2}[\log(-4\pi s)]^{1+n/2}ds$$ Then make the substitution $u=\log(-4\pi s),$ giving $du=1/s$ and $s^{n/2}=(-\frac{1}{4\pi} e^{u})^{n/2}$. Your limits become $0$ (upper) and $-\infty$ (lower) - switch them, using up a minus sign. The gamma function should become very apparent.

Note: You should be much more careful than I have been with minus signs in logarithms and square roots. This is purely a sketch to give you some ideas.

4

$\newcommand{\angles}[1]{\left\langle #1 \right\rangle}% \newcommand{\braces}[1]{\left\lbrace #1 \right\rbrace}% \newcommand{\bracks}[1]{\left\lbrack #1 \right\rbrack}% \newcommand{\dd}{{\rm d}}% \newcommand{\ds}[1]{\displaystyle{#1}}% \newcommand{\expo}[1]{{\rm e}^{#1}}% \newcommand{\ic}{{\rm i}}% \newcommand{\imp}{\Longrightarrow}% \newcommand{\pars}[1]{\left( #1 \right)}% \newcommand{\pp}{{\cal P}}% \newcommand{\ul}[1]{\underline{#1}}% \newcommand{\verts}[1]{\left\vert #1 \right\vert}$

\begin{align} &\overbrace{\int_{-1/\pars{4\pi}}^{{0}}{% \bracks{s\log{\pars{-4\pi s}}}^{\pars{2 + n}/2} \over s^{2}}\,\dd s}^{\ds{s\ =\ -\,{x \over 4\pi}\,,\quad x = -4\pi s}}\ =\ \pars{-1}^{\pars{n/2} -1}\,\,{\pi^{-n/2} \over 2^{n}}\quad \overbrace{\int_{0}^{1}x^{\pars{n/2} - 1}\ln^{\pars{n/2} + 1}\pars{x}\,\dd x} ^{\ds{x\ =\ \expo{-z}\,,\quad z = -\ln\pars{x}}} \\[3mm]&= \pars{-1}^{\pars{n/2} - 1}\,\, {\pi^{-n/2} \over 2^{n}}\,\pars{-1}^{\pars{n/2} + 1} \int_{0}^{\infty}\expo{-nz/2}\,z^{\pars{n/2} + 1}\,\dd z \\[3mm]&= \pars{-1}^{n}\,\,{\pi^{-n/2} \over 2^{n}}\,{1 \over \pars{n/2}^{\pars{n/2} + 2}} \quad \overbrace{\int_{0}^{\infty}\expo{-z}\,z^{\bracks{\pars{n/2} + 2} - 1}\,\dd z} ^{\ds{\Gamma\pars{{n \over 2} + 2}}} \\[3mm]&= \pars{-1}^{n}\,\,{\pi^{-n/2} \over 2^{-2 + \pars{n/2}}}\, {1 \over n^{\pars{n/2} + 2}}\Gamma\pars{2 + {n \over 2}} = \pars{-1}^{n}\,\,{8\pi^{-n/2} \over 2^{1 + \pars{n/2}}}\, {1 \over n^{\pars{4 + n}/2}}\Gamma\pars{2 + {n \over 2}} \\[1cm]& \end{align}

$$ {\large\int_{-1/\pars{4\pi}}^{{0}}{% \bracks{s\log{\pars{-4\pi s}}}^{\pars{2 + n}/2} \over s^{2}}\,\dd s = \color{#ff0000}{\large% \pars{-1}^{n}\, {8\pi^{-n/2}\,\Gamma\pars{2 + n/2} \over 2^{1 + n/2\,}n^{\pars{4 + n}/2}}}} $$ Notice that we got a pre factor $\pars{-1}^{n}$ besides the result the OP reports from the Mathematica package. We check explicitly ( we started from the initial integral with $n = 1$ ) our result for $n = 1$ and we found it's negative !!!. We would like the OP recheck it with Mathematica ( Sorry, I don't have Mathematica ).

Felix Marin
  • 89,464
  • Are you sure that your calculation is correct? – Pedro Oct 18 '13 at 00:36
  • @Pedro I'll check everything. 0 k. – Felix Marin Oct 18 '13 at 01:56
  • @Pedro I recheck everything from the very beginning. I got the prefactor $\left(-1\right)^{n}$. Read my comment at the end of the calculation. – Felix Marin Oct 18 '13 at 03:50
  • Thanks a lot. I made the changes that you suggested and it seems that worked. However, I made all changes before factor out the constants. So, we get

    $$\int_{-1/(4\pi)}^0\frac{(s\log{(-4\pi s)})^{(2+n)/2}}{s^2},ds$$

    $$=\int_0^1\frac{\left((-x/4\pi)\ln x\right)^{n/2+1}}{(-x/4\pi)^2}(-1/(4\pi),dx$$

    $$=\int_0^\infty \frac{((-\exp({-z})/4\pi)(-z))^{n/2+1}}{(-\exp({-z})/4\pi)^2}(-1/(4\pi)(-\exp({-z})),dz$$

    $$=\frac{8\pi^{-n/2}\Gamma(2+n/2)}{2^{1+n/2}n^{(4+n)/2}}$$

    – Pedro Oct 18 '13 at 08:46
  • I think that in your solution, the error comes up in the second equality, when you uses $\sqrt{ab}=\sqrt{a}\sqrt{b}$ with $a,b<0$ (this can be dangerous). Notice that after you make the two changes this problem disappears. – Pedro Oct 18 '13 at 08:50
  • @Pedro I'll check tomorrow because it's too late ( 4:30 in the morning ). Thanks. – Felix Marin Oct 18 '13 at 09:05
  • @Pedro In my second checking ( with $n = 1$ ), I have the feeling the problem could be related to $n$ odd character. – Felix Marin Oct 18 '13 at 09:07
  • Maybe you are right, because it seems that my calculation above is wrong too. I'll check it (again). – Pedro Oct 18 '13 at 09:28
  • @FelixMarin, thank you for your help. Note that the integrand is positive, hence, the answer must be positive. – Tomás Oct 23 '13 at 13:19
  • @Tomás I'll remove the factor later. I'm so busy now. Thanks. – Felix Marin Oct 24 '13 at 00:57
  • I have posted a compilation of the calculation here. – Pedro Apr 09 '17 at 20:11