Another series expansion trouble??

Calculus Level 5

Evaluate

0 x 2 1 + e x d x \int_0^{\infty} \frac{x^2}{1+e^x}\,dx

Note: ζ ( 3 ) = 1.20205 \zeta(3)=1.20205 where ζ ( s ) \zeta(s) is the Riemann zeta function.


The answer is 1.80308.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

3 solutions

Anish Puthuraya
Apr 1, 2014

Im very weak in this concept, so please, if the following is not the cleanest method to solve the problem, then comment below about it.

I ( k ) = 0 x 2 1 + e k x d x I(k) = \int_0^{\infty} \frac{x^2}{1+e^{kx}}dx

I ( k ) = 0 x 3 e k x ( 1 + e k x ) 2 d x I'(k) = \int_0^{\infty} \frac{-x^3e^{kx}}{(1+e^{kx})^2}dx

Let 1 + e k x = t \displaystyle 1+e^{kx} = t

I ( k ) = 1 k 4 2 ( ln ( t 1 ) ) 3 t 2 d t I'(k) = \frac{-1}{k^4} \int_2^{\infty} \frac{(\ln(t-1))^3}{t^2}dt

This integral can be easily evaluated using the expansion of ln ( t 1 ) \displaystyle\ln(t-1) around the point t = 2 \displaystyle t = 2

I ( k ) = 1 k 4 9 2 ζ ( 3 ) I'(k) = \frac{-1}{k^4} \frac{9}{2} \zeta(3)

I ( k ) = 9 2 ζ ( 3 ) 1 k 4 I(k) = \frac{9}{2}\zeta(3) \int\frac{-1}{k^4}

I ( k ) = 9 2 ζ ( 3 ) 1 3 k 3 I(k) = \frac{9}{2}\zeta(3) \frac{1}{3k^3}

Substituting k = 1 \displaystyle k=1 to get the required integral, we have,

I ( 1 ) = 3 2 ζ ( 3 ) = 1.80308 I(1) = \frac{3}{2}\zeta(3) = \boxed{1.80308}

Nicely done! :)

Another method is to rewrite the integral as follows:

0 x 2 e x 1 + e x d x = k = 0 ( 1 ) k 0 x 2 e ( k + 1 ) x d x \displaystyle \int_0^{\infty} \frac{x^2e^{-x}}{1+e^{-x}}\,dx=\displaystyle \sum_{k=0}^{\infty} (-1)^k \int_0^{\infty} x^2 e^{-(k+1)x}\,dx

The above can be easily evaluated using IBP.

Pranav Arora - 7 years, 2 months ago

Log in to reply

Thats a good one. Please correct me if I'm wrong, but is the following also a method?

I = 0 x 2 1 + e x d x I = \int_0^{\infty}\frac{x^2}{1+e^x}dx

I = 0 x 2 ( 1 + e x ) 1 d x I = \int_0^{\infty}x^2(1+e^x)^{-1}dx

I = 0 x 2 ( 1 e x + e 2 x e 3 x + ) d x I = \int_0^{\infty}x^2(1-e^x+e^{2x}-e^{3x}+\ldots)dx

I = 0 x 2 k = 0 ( 1 ) k e k x d x I = \int_0^{\infty} x^2 \sum_{k=0}^{\infty} (-1)^k e^{kx} dx

I = k = 0 ( 1 ) k 0 x 2 e k x d x I = \sum_{k=0}^{\infty} (-1)^k \int_0^{\infty} x^2 e^{kx} dx

I am not sure whether the above step is valid or not. If it is correct, then you can see that the integral 0 x 2 e k x d x \displaystyle\int_0^{\infty}x^2 e^{kx}dx is diverging. So, is it valid?

Anish Puthuraya - 7 years, 2 months ago

Log in to reply

Nope, it is not correct. In my comment, I used the following series expansion:

1 1 + x = k = 0 ( 1 ) k x k \displaystyle \frac{1}{1+x}=\sum_{k=0}^{\infty} (-1)^kx^k

which is valid for x < 1 |x|<1 . Clearly e x e^x lies between 1 to \infty for x ( 0 , ) x \in (0,\infty) so it is not correct to use the above series expansion.

I hope that helps. :)

Pranav Arora - 7 years, 2 months ago

Log in to reply

@Pranav Arora Oh, of course! How could I miss that!! Thanks for that clarification Pranav! I think I just made a fool of myself by committing that silly mistake!

Anish Puthuraya - 7 years, 2 months ago

Log in to reply

@Anish Puthuraya and @Pranav Arora , I think you are not completely wrong Anish. If you know the analytic continuation of Riemann zeta function or might be familiar with Grandi's series , your method might be justified. Note that, the series expansion also valid for x = 1 x=1 in String Theory.

Tunk-Fey Ariawan - 7 years, 2 months ago

I too did it using Integration by Parts;

Differentiate x 2 x^{2} and integrate e x 1 + e x \displaystyle\frac{e^{-x}}{1+e^{-x}} to obtain:

0 2 x ln ( 1 + e x ) d x \displaystyle\int_{0}^{\infty} 2x\ln(1+e^{-x})dx

Now use series expansion of l n ( 1 + x ) ln(1+x) and resultant integrals can also be easily evaluated using Integration by Parts.

By the way, Anish how did you type that solution so fast? It takes me ages to type anything in LaTeX.

Karthik Kannan - 7 years, 2 months ago

Log in to reply

I did in same way as you!

jatin yadav - 7 years, 2 months ago

Nice solution.

I see that you recently wrote your first solution. So, if you keep posting solutions, you can get pretty fast at writing them.

Anish Puthuraya - 7 years, 2 months ago

Log in to reply

Thanks ! : )

Karthik Kannan - 7 years, 2 months ago

@Anish Puthuraya Can you please post a solution to A tank full of helium in space . I get v = f R T M \displaystyle v= \sqrt{\frac{fRT}{M}} , where f f are degrees of freedom.

jatin yadav - 7 years, 2 months ago

Log in to reply

@jatin yadav Done!

Anish Puthuraya - 7 years, 2 months ago
Tunk-Fey Ariawan
Apr 2, 2014

This is another approach in addition to Anish's solution. From the formulation of Dirichlet eta function we have: η ( s ) = ( 1 2 1 s ) ζ ( s ) \eta(s)=\left(1-2^{1-s}\right)\zeta(s) and η ( s ) = 1 Γ ( s ) 0 x s 1 1 + e x d x . \eta(s)=\frac{1}{\Gamma(s)}\int_0^\infty\frac{x^{s-1}}{1+e^x}\,dx. where Γ ( s ) \Gamma(s) is gamma function and Γ ( s ) = ( s 1 ) ! \Gamma(s)=(s-1)! for s Z + s\in\mathbb{Z}_+ .

Therefore, the solution of the integral becomes straightforward by using Dirichlet eta function. 0 x 2 1 + e x d x = ( 1 2 2 ) Γ ( 3 ) ζ ( 3 ) = 3 4 2 ! 1.20205 = 1.80308 \begin{aligned} \int_0^\infty\frac{x^2}{1+e^x}\,dx &=(1-2^{-2})\Gamma(3)\zeta(3)\\ &=\frac{3}{4}\cdot 2!\cdot1.20205\\ &=\boxed{\color{#3D99F6}{1.80308}} \end{aligned}


# Q . E . D . # \Large\color{#3D99F6}{\text{\# }\mathbb{Q.E.D.}\text{ \#}}

@Anish Puthuraya , @Pranav Arora , and @Karthik Kannan . Sorry guys, I win! Mine is shorter. Hehehe... (^_^)V

Just_Kidding

Tunk-Fey Ariawan - 7 years, 2 months ago

Log in to reply

Hi-Five! you won!

Anish Puthuraya - 7 years, 2 months ago

Log in to reply

You, @Pranav Arora , and @Karthik Kannan are awesome. It never crossed my mind using IBP. Nice work guys! (y) (y)

Tunk-Fey Ariawan - 7 years, 2 months ago

Log in to reply

@Tunk-Fey Ariawan But how do you prove that integral representation of η ( s ) \eta(s) ?

Bogdan Simeonov - 7 years, 2 months ago

Its always great to see different approaches to the same problem, thank you for sharing this! :)

(Though this one went above my head, I am not introduced to gamma functions yet. :P )

Pranav Arora - 7 years, 2 months ago
Lucas Tell Marchi
Jan 20, 2015

First of all, note that

1 1 + e x = 1 e x ( 1 1 + 1 e x ) = 1 e x k = 0 e k x ( 1 ) k = k = 0 ( 1 ) k e x ( k + 1 ) e x < 1 \frac{1}{1+e^x} = \frac{1}{e^x} \left ( \frac{1}{1+\frac{1}{e^x}} \right ) = \frac{1}{e^x} \sum_{k=0}^{\infty} e^{-kx} (-1)^k = \sum_{k=0}^{\infty} (-1)^k e^{-x(k+1)} \;\;\; |e^x|<1

Then

0 x 2 1 + e x d x = k = 0 ( 1 ) k 0 x 2 e x ( k + 1 ) d x \int_{0}^{\infty} \frac{x^2}{1+e^x} \; dx = \sum_{k=0}^{\infty} (-1)^k \int_{0}^{\infty} x^2e^{-x(k+1)} \; dx

Now, integration by parts will give us

0 x 2 e x ( k + 1 ) d x = ( x 2 e x ( k + 1 ) k + 1 ) 0 + 2 0 x e x ( k + 1 ) k + 1 d x \int_{0}^{\infty} x^2e^{-x(k+1)} \; dx = - \left (\frac{x^2e^{-x(k+1)}}{k+1} \right )_{0}^{\infty} + 2 \int_{0}^{\infty}x \; \frac{e^{-x(k+1)}}{k+1} \; dx

And

0 x e x ( k + 1 ) k + 1 d x = ( x e x ( k + 1 ) k + 1 ) 0 + 0 e x ( k + 1 ) ( k + 1 ) 2 d x \int_{0}^{\infty}x \; \frac{e^{-x(k+1)}}{k+1} \; dx = - \left ( \frac{xe^{-x(k+1)}}{k+1} \right )_{0}^{\infty} + \int_{0}^{\infty} \frac{e^{-x(k+1)}}{(k+1)^2} \; dx

Therefore

0 x 2 e x ( k + 1 ) d x = ( x 2 e x ( k + 1 ) k + 1 ) 0 2 ( x e x ( k + 1 ) k + 1 ) 0 + 2 0 e x ( k + 1 ) ( k + 1 ) 2 d x \int_{0}^{\infty} x^2e^{-x(k+1)} \; dx = - \left (\frac{x^2e^{-x(k+1)}}{k+1} \right )_{0}^{\infty} - 2 \left ( \frac{xe^{-x(k+1)}}{k+1} \right )_{0}^{\infty} + 2 \int_{0}^{\infty} \frac{e^{-x(k+1)}}{(k+1)^2} \; dx

Now using L'Hospital

lim x x 2 e x ( k + 1 ) = lim x 2 x ( k + 1 ) e x ( k + 1 ) = 2 ( k + 1 ) 2 lim x 1 e x ( k + 1 ) = 0 \lim_{x\rightarrow \infty} \frac{x^2}{e^{x(k+1)}} = \lim_{x\rightarrow \infty} \frac{2x}{(k+1)e^{x(k+1)}} = \frac{2}{(k+1)^2} \lim_{x\rightarrow \infty} \frac{1}{e^{x(k+1)}} = 0

And therefore

0 x 2 e x ( k + 1 ) d x = 2 0 e x ( k + 1 ) ( k + 1 ) 2 d x = 2 ( e x ( k + 1 ) ( k + 1 ) 3 ) 0 = 2 ( 1 k + 1 ) 3 \int_{0}^{\infty} x^2e^{-x(k+1)} \; dx = 2 \int_{0}^{\infty} \frac{e^{-x(k+1)}}{(k+1)^2} \; dx = -2\left ( \frac{e^{-x(k+1)}}{(k+1)^3} \right )_{0}^{\infty} = 2\left ( \frac{1}{k+1} \right )^3

Thus the original integral is

0 x 2 1 + e x d x = 2 k = 0 ( 1 ) k ( 1 k + 1 ) 3 = 2 n = 1 ( 1 ) n 1 1 n 3 = ( 7 8 1 8 ) ζ ( 3 ) 1.803 \int_{0}^{\infty} \frac{x^2}{1+e^x} \; dx = 2\sum_{k=0}^{\infty} (-1)^k \left ( \frac{1}{k+1} \right )^3 = 2 \sum_{n=1}^{\infty} (-1)^{n-1}\frac{1}{n^3} = \left ( \frac{7}{8} - \frac{1}{8} \right ) \zeta(3) \approx 1.803

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...