How to solve this improper integral $int _{-infty}^{infty} e^{ax}/(e^x+1) dx$? [closed]












0












$begingroup$


I need to solve the following integral.
$$int _{-infty }^{infty }::dfrac{e^{ax}}{e^x+1}dx$$



where $0<a<1$.










share|cite|improve this question











$endgroup$



closed as off-topic by Brahadeesh, Shaun, user10354138, mrtaurho, Paul Frost Dec 20 '18 at 18:18


This question appears to be off-topic. The users who voted to close gave this specific reason:


  • "This question is missing context or other details: Please provide additional context, which ideally explains why the question is relevant to you and our community. Some forms of context include: background and motivation, relevant definitions, source, possible strategies, your current progress, why the question is interesting or important, etc." – Brahadeesh, Shaun, user10354138, mrtaurho, Paul Frost

If this question can be reworded to fit the rules in the help center, please edit the question.












  • 2




    $begingroup$
    Your integral seems divergent.
    $endgroup$
    – Awe Kumar Jha
    Dec 20 '18 at 10:39






  • 1




    $begingroup$
    Come on, people! It is convergent and is equal to $mathrm{B}(a,1-a)=dfrac{pi}{sin api}$.
    $endgroup$
    – metamorphy
    Dec 20 '18 at 12:01








  • 1




    $begingroup$
    Did you find $mathrm{B}(x,y)=displaystyleint_0^inftyfrac{t^{x-1},dt}{(1+t)^{x+y}}$ there? Please read the article before asking further questions - I wouldn't want to reproduce the whole of it here ;)
    $endgroup$
    – metamorphy
    Dec 20 '18 at 12:18








  • 1




    $begingroup$
    @John You could solve this integral using a rectangular contour which goes from -R to R on the real axis and which is of height $2pi i$. Thus you have a pole at $x=pi i$. If you edit your question so that it fits the rules of the site, I can maybe answer your question, because I had to solve this exact same integral today also using contour integration.
    $endgroup$
    – Poujh
    Dec 20 '18 at 20:45








  • 1




    $begingroup$
    @Poujh That's honestly much easier than the approach I took. So thanks for helping me learn today.
    $endgroup$
    – Dylan
    Dec 21 '18 at 3:56


















0












$begingroup$


I need to solve the following integral.
$$int _{-infty }^{infty }::dfrac{e^{ax}}{e^x+1}dx$$



where $0<a<1$.










share|cite|improve this question











$endgroup$



closed as off-topic by Brahadeesh, Shaun, user10354138, mrtaurho, Paul Frost Dec 20 '18 at 18:18


This question appears to be off-topic. The users who voted to close gave this specific reason:


  • "This question is missing context or other details: Please provide additional context, which ideally explains why the question is relevant to you and our community. Some forms of context include: background and motivation, relevant definitions, source, possible strategies, your current progress, why the question is interesting or important, etc." – Brahadeesh, Shaun, user10354138, mrtaurho, Paul Frost

If this question can be reworded to fit the rules in the help center, please edit the question.












  • 2




    $begingroup$
    Your integral seems divergent.
    $endgroup$
    – Awe Kumar Jha
    Dec 20 '18 at 10:39






  • 1




    $begingroup$
    Come on, people! It is convergent and is equal to $mathrm{B}(a,1-a)=dfrac{pi}{sin api}$.
    $endgroup$
    – metamorphy
    Dec 20 '18 at 12:01








  • 1




    $begingroup$
    Did you find $mathrm{B}(x,y)=displaystyleint_0^inftyfrac{t^{x-1},dt}{(1+t)^{x+y}}$ there? Please read the article before asking further questions - I wouldn't want to reproduce the whole of it here ;)
    $endgroup$
    – metamorphy
    Dec 20 '18 at 12:18








  • 1




    $begingroup$
    @John You could solve this integral using a rectangular contour which goes from -R to R on the real axis and which is of height $2pi i$. Thus you have a pole at $x=pi i$. If you edit your question so that it fits the rules of the site, I can maybe answer your question, because I had to solve this exact same integral today also using contour integration.
    $endgroup$
    – Poujh
    Dec 20 '18 at 20:45








  • 1




    $begingroup$
    @Poujh That's honestly much easier than the approach I took. So thanks for helping me learn today.
    $endgroup$
    – Dylan
    Dec 21 '18 at 3:56
















0












0








0


2



$begingroup$


I need to solve the following integral.
$$int _{-infty }^{infty }::dfrac{e^{ax}}{e^x+1}dx$$



where $0<a<1$.










share|cite|improve this question











$endgroup$




I need to solve the following integral.
$$int _{-infty }^{infty }::dfrac{e^{ax}}{e^x+1}dx$$



where $0<a<1$.







integration improper-integrals contour-integration






share|cite|improve this question















share|cite|improve this question













share|cite|improve this question




share|cite|improve this question








edited Dec 20 '18 at 11:02









Brahadeesh

6,36442363




6,36442363










asked Dec 20 '18 at 10:31









JohnJohn

8110




8110




closed as off-topic by Brahadeesh, Shaun, user10354138, mrtaurho, Paul Frost Dec 20 '18 at 18:18


This question appears to be off-topic. The users who voted to close gave this specific reason:


  • "This question is missing context or other details: Please provide additional context, which ideally explains why the question is relevant to you and our community. Some forms of context include: background and motivation, relevant definitions, source, possible strategies, your current progress, why the question is interesting or important, etc." – Brahadeesh, Shaun, user10354138, mrtaurho, Paul Frost

If this question can be reworded to fit the rules in the help center, please edit the question.







closed as off-topic by Brahadeesh, Shaun, user10354138, mrtaurho, Paul Frost Dec 20 '18 at 18:18


This question appears to be off-topic. The users who voted to close gave this specific reason:


  • "This question is missing context or other details: Please provide additional context, which ideally explains why the question is relevant to you and our community. Some forms of context include: background and motivation, relevant definitions, source, possible strategies, your current progress, why the question is interesting or important, etc." – Brahadeesh, Shaun, user10354138, mrtaurho, Paul Frost

If this question can be reworded to fit the rules in the help center, please edit the question.








  • 2




    $begingroup$
    Your integral seems divergent.
    $endgroup$
    – Awe Kumar Jha
    Dec 20 '18 at 10:39






  • 1




    $begingroup$
    Come on, people! It is convergent and is equal to $mathrm{B}(a,1-a)=dfrac{pi}{sin api}$.
    $endgroup$
    – metamorphy
    Dec 20 '18 at 12:01








  • 1




    $begingroup$
    Did you find $mathrm{B}(x,y)=displaystyleint_0^inftyfrac{t^{x-1},dt}{(1+t)^{x+y}}$ there? Please read the article before asking further questions - I wouldn't want to reproduce the whole of it here ;)
    $endgroup$
    – metamorphy
    Dec 20 '18 at 12:18








  • 1




    $begingroup$
    @John You could solve this integral using a rectangular contour which goes from -R to R on the real axis and which is of height $2pi i$. Thus you have a pole at $x=pi i$. If you edit your question so that it fits the rules of the site, I can maybe answer your question, because I had to solve this exact same integral today also using contour integration.
    $endgroup$
    – Poujh
    Dec 20 '18 at 20:45








  • 1




    $begingroup$
    @Poujh That's honestly much easier than the approach I took. So thanks for helping me learn today.
    $endgroup$
    – Dylan
    Dec 21 '18 at 3:56
















  • 2




    $begingroup$
    Your integral seems divergent.
    $endgroup$
    – Awe Kumar Jha
    Dec 20 '18 at 10:39






  • 1




    $begingroup$
    Come on, people! It is convergent and is equal to $mathrm{B}(a,1-a)=dfrac{pi}{sin api}$.
    $endgroup$
    – metamorphy
    Dec 20 '18 at 12:01








  • 1




    $begingroup$
    Did you find $mathrm{B}(x,y)=displaystyleint_0^inftyfrac{t^{x-1},dt}{(1+t)^{x+y}}$ there? Please read the article before asking further questions - I wouldn't want to reproduce the whole of it here ;)
    $endgroup$
    – metamorphy
    Dec 20 '18 at 12:18








  • 1




    $begingroup$
    @John You could solve this integral using a rectangular contour which goes from -R to R on the real axis and which is of height $2pi i$. Thus you have a pole at $x=pi i$. If you edit your question so that it fits the rules of the site, I can maybe answer your question, because I had to solve this exact same integral today also using contour integration.
    $endgroup$
    – Poujh
    Dec 20 '18 at 20:45








  • 1




    $begingroup$
    @Poujh That's honestly much easier than the approach I took. So thanks for helping me learn today.
    $endgroup$
    – Dylan
    Dec 21 '18 at 3:56










2




2




$begingroup$
Your integral seems divergent.
$endgroup$
– Awe Kumar Jha
Dec 20 '18 at 10:39




$begingroup$
Your integral seems divergent.
$endgroup$
– Awe Kumar Jha
Dec 20 '18 at 10:39




1




1




$begingroup$
Come on, people! It is convergent and is equal to $mathrm{B}(a,1-a)=dfrac{pi}{sin api}$.
$endgroup$
– metamorphy
Dec 20 '18 at 12:01






$begingroup$
Come on, people! It is convergent and is equal to $mathrm{B}(a,1-a)=dfrac{pi}{sin api}$.
$endgroup$
– metamorphy
Dec 20 '18 at 12:01






1




1




$begingroup$
Did you find $mathrm{B}(x,y)=displaystyleint_0^inftyfrac{t^{x-1},dt}{(1+t)^{x+y}}$ there? Please read the article before asking further questions - I wouldn't want to reproduce the whole of it here ;)
$endgroup$
– metamorphy
Dec 20 '18 at 12:18






$begingroup$
Did you find $mathrm{B}(x,y)=displaystyleint_0^inftyfrac{t^{x-1},dt}{(1+t)^{x+y}}$ there? Please read the article before asking further questions - I wouldn't want to reproduce the whole of it here ;)
$endgroup$
– metamorphy
Dec 20 '18 at 12:18






1




1




$begingroup$
@John You could solve this integral using a rectangular contour which goes from -R to R on the real axis and which is of height $2pi i$. Thus you have a pole at $x=pi i$. If you edit your question so that it fits the rules of the site, I can maybe answer your question, because I had to solve this exact same integral today also using contour integration.
$endgroup$
– Poujh
Dec 20 '18 at 20:45






$begingroup$
@John You could solve this integral using a rectangular contour which goes from -R to R on the real axis and which is of height $2pi i$. Thus you have a pole at $x=pi i$. If you edit your question so that it fits the rules of the site, I can maybe answer your question, because I had to solve this exact same integral today also using contour integration.
$endgroup$
– Poujh
Dec 20 '18 at 20:45






1




1




$begingroup$
@Poujh That's honestly much easier than the approach I took. So thanks for helping me learn today.
$endgroup$
– Dylan
Dec 21 '18 at 3:56






$begingroup$
@Poujh That's honestly much easier than the approach I took. So thanks for helping me learn today.
$endgroup$
– Dylan
Dec 21 '18 at 3:56












1 Answer
1






active

oldest

votes


















1












$begingroup$

The substitution $e^x mapsto x$ gives



$$ I = int_{-infty}^infty frac{e^{ax}}{e^x+1}dx = int_0^infty frac{1}{x^{1-a}(1+x)}dx $$



where $0 < 1-a < 1$



We can use complex integration here. Due to the pole at $z=-1$, we pick a branch cut on the positive real axis such that $0 le arg(z) < 2pi$. The contour is a standard "keyhole contour" consisting of





  • $C_1$: left semicircle centered at $0$ with radius $epsilon$, going clockwise from $-epsilon i$ to $epsilon i$


  • $C_2$: straight line from $epsilon i$ to $R + epsilon i$


  • $C_3$: circle centered at $0$ with radius $R$, going counterclockwise from $R+iepsilon$ to $R - epsilon i$


  • $C_4$: straight line from $R - epsilon i$ to $-epsilon i$


In the limits $epsilon to 0$ and $R to infty$, the two straight line integrals converge to



$$ int_{C_2} f(z) dz to int_0^infty frac{1}{x^{1-a}e^{i0(1-a)}(1+x)}dx = I $$



$$ int_{C_4} f(z) dz to -int_0^infty frac{1}{x^{1-a}e^{i2pi(1-a)}(1+x)}dx = -e^{i2pi a}I $$



The remaining integrals should go to $0$



$$ leftvertint_{C_1} f(z)dzrightvert le frac{pi epsilon}{|z|^{1-a}|1+z|} le frac{pi epsilon^a}{1-epsilon} to 0 $$



$$ leftvertint_{C_3} f(z) dz rightvert le frac{2pi R}{|z|^{1-a}|1+z|} le frac{2pi R^a}{R-1} to 0 $$



Therefore



$$ (1-e^{i2pi a})I = 2pi ioperatorname*{Res}_{z=-1} f(z) = -2pi i e^{ipi a} $$



So finally



$$ I = frac{2pi ie^{ipi a}}{e^{2pi a}-1} = frac{2pi i}{e^{ipi a}-e^{-ipi a}} = frac{pi}{sin(api)} $$





Edit: Here's a simpler method, credit to @Poujh



$$ I = int_{-infty}^infty frac{e^{ax}}{e^x+1}dx = int_{-infty}^infty f(x) dx $$



Take a rectangular contour:





  • $C_1$: Straight line from $-R$ to $R$


  • $C_2$: Straight line from $R$ to $R + i2pi$


  • $C_3$: Straight line from $R + i2pi$ to $-R + i2pi$


  • $C_4$: Straight line from $-R + i2pi$ to $-R$




In the limit of $Rtoinfty$, we have



$$ leftvertint_{C_2} frac{e^{az}}{e^z+1} dzrightvert = leftvertint_0^{2pi} frac{e^{a(R+iy)}}{e^{R+iy}+1} dyrightvert le 2pi leftvertfrac{e^{aR}}{e^R-1}rightvert to 0 $$



$$ leftvertint_{C_4} frac{e^{az}}{e^z+1} dzrightvert = leftvertint_0^{2pi} frac{e^{a(-R+iy)}}{e^{-R+iy}+1} rightvert le 2pi leftvert frac{e^{-aR}}{1-e^{-R}} rightvert to 0 $$



For the horizontal lines



$$ int_{C_1} frac{e^{az}}{e^z+1} dz to int_{-infty}^infty frac{e^{ax}}{e^x+1} dx = I $$



$$ int_{C_3} frac{e^{az}}{e^z+1} dz to -int_{-infty}^infty frac{e^{ax}e^{i2pi a}}{e^x+1} = -e^{i2pi a} I $$



We end up with the same result



$$ (1-e^{i2pi a})I = 2pi ioperatorname*{Res}_{z=ipi} frac{e^{az}}{e^z+1} $$



where the residue is computed using the limit



$$ lim_{zto ipi} e^{az}frac{z-ipi}{e^z+1} = -e^{ipi a} $$






share|cite|improve this answer











$endgroup$













  • $begingroup$
    What about the other poles in this second method? Of course $exp(z)+1$ has zeros at $(2n+1)pi i$.
    $endgroup$
    – John
    Dec 23 '18 at 12:22






  • 1




    $begingroup$
    @John The other poles are outside the contour
    $endgroup$
    – Dylan
    Dec 23 '18 at 13:16








  • 1




    $begingroup$
    yes I got it and the residue is $- exp(i pi a)$, you can edit it.
    $endgroup$
    – John
    Dec 23 '18 at 13:23


















1 Answer
1






active

oldest

votes








1 Answer
1






active

oldest

votes









active

oldest

votes






active

oldest

votes









1












$begingroup$

The substitution $e^x mapsto x$ gives



$$ I = int_{-infty}^infty frac{e^{ax}}{e^x+1}dx = int_0^infty frac{1}{x^{1-a}(1+x)}dx $$



where $0 < 1-a < 1$



We can use complex integration here. Due to the pole at $z=-1$, we pick a branch cut on the positive real axis such that $0 le arg(z) < 2pi$. The contour is a standard "keyhole contour" consisting of





  • $C_1$: left semicircle centered at $0$ with radius $epsilon$, going clockwise from $-epsilon i$ to $epsilon i$


  • $C_2$: straight line from $epsilon i$ to $R + epsilon i$


  • $C_3$: circle centered at $0$ with radius $R$, going counterclockwise from $R+iepsilon$ to $R - epsilon i$


  • $C_4$: straight line from $R - epsilon i$ to $-epsilon i$


In the limits $epsilon to 0$ and $R to infty$, the two straight line integrals converge to



$$ int_{C_2} f(z) dz to int_0^infty frac{1}{x^{1-a}e^{i0(1-a)}(1+x)}dx = I $$



$$ int_{C_4} f(z) dz to -int_0^infty frac{1}{x^{1-a}e^{i2pi(1-a)}(1+x)}dx = -e^{i2pi a}I $$



The remaining integrals should go to $0$



$$ leftvertint_{C_1} f(z)dzrightvert le frac{pi epsilon}{|z|^{1-a}|1+z|} le frac{pi epsilon^a}{1-epsilon} to 0 $$



$$ leftvertint_{C_3} f(z) dz rightvert le frac{2pi R}{|z|^{1-a}|1+z|} le frac{2pi R^a}{R-1} to 0 $$



Therefore



$$ (1-e^{i2pi a})I = 2pi ioperatorname*{Res}_{z=-1} f(z) = -2pi i e^{ipi a} $$



So finally



$$ I = frac{2pi ie^{ipi a}}{e^{2pi a}-1} = frac{2pi i}{e^{ipi a}-e^{-ipi a}} = frac{pi}{sin(api)} $$





Edit: Here's a simpler method, credit to @Poujh



$$ I = int_{-infty}^infty frac{e^{ax}}{e^x+1}dx = int_{-infty}^infty f(x) dx $$



Take a rectangular contour:





  • $C_1$: Straight line from $-R$ to $R$


  • $C_2$: Straight line from $R$ to $R + i2pi$


  • $C_3$: Straight line from $R + i2pi$ to $-R + i2pi$


  • $C_4$: Straight line from $-R + i2pi$ to $-R$




In the limit of $Rtoinfty$, we have



$$ leftvertint_{C_2} frac{e^{az}}{e^z+1} dzrightvert = leftvertint_0^{2pi} frac{e^{a(R+iy)}}{e^{R+iy}+1} dyrightvert le 2pi leftvertfrac{e^{aR}}{e^R-1}rightvert to 0 $$



$$ leftvertint_{C_4} frac{e^{az}}{e^z+1} dzrightvert = leftvertint_0^{2pi} frac{e^{a(-R+iy)}}{e^{-R+iy}+1} rightvert le 2pi leftvert frac{e^{-aR}}{1-e^{-R}} rightvert to 0 $$



For the horizontal lines



$$ int_{C_1} frac{e^{az}}{e^z+1} dz to int_{-infty}^infty frac{e^{ax}}{e^x+1} dx = I $$



$$ int_{C_3} frac{e^{az}}{e^z+1} dz to -int_{-infty}^infty frac{e^{ax}e^{i2pi a}}{e^x+1} = -e^{i2pi a} I $$



We end up with the same result



$$ (1-e^{i2pi a})I = 2pi ioperatorname*{Res}_{z=ipi} frac{e^{az}}{e^z+1} $$



where the residue is computed using the limit



$$ lim_{zto ipi} e^{az}frac{z-ipi}{e^z+1} = -e^{ipi a} $$






share|cite|improve this answer











$endgroup$













  • $begingroup$
    What about the other poles in this second method? Of course $exp(z)+1$ has zeros at $(2n+1)pi i$.
    $endgroup$
    – John
    Dec 23 '18 at 12:22






  • 1




    $begingroup$
    @John The other poles are outside the contour
    $endgroup$
    – Dylan
    Dec 23 '18 at 13:16








  • 1




    $begingroup$
    yes I got it and the residue is $- exp(i pi a)$, you can edit it.
    $endgroup$
    – John
    Dec 23 '18 at 13:23
















1












$begingroup$

The substitution $e^x mapsto x$ gives



$$ I = int_{-infty}^infty frac{e^{ax}}{e^x+1}dx = int_0^infty frac{1}{x^{1-a}(1+x)}dx $$



where $0 < 1-a < 1$



We can use complex integration here. Due to the pole at $z=-1$, we pick a branch cut on the positive real axis such that $0 le arg(z) < 2pi$. The contour is a standard "keyhole contour" consisting of





  • $C_1$: left semicircle centered at $0$ with radius $epsilon$, going clockwise from $-epsilon i$ to $epsilon i$


  • $C_2$: straight line from $epsilon i$ to $R + epsilon i$


  • $C_3$: circle centered at $0$ with radius $R$, going counterclockwise from $R+iepsilon$ to $R - epsilon i$


  • $C_4$: straight line from $R - epsilon i$ to $-epsilon i$


In the limits $epsilon to 0$ and $R to infty$, the two straight line integrals converge to



$$ int_{C_2} f(z) dz to int_0^infty frac{1}{x^{1-a}e^{i0(1-a)}(1+x)}dx = I $$



$$ int_{C_4} f(z) dz to -int_0^infty frac{1}{x^{1-a}e^{i2pi(1-a)}(1+x)}dx = -e^{i2pi a}I $$



The remaining integrals should go to $0$



$$ leftvertint_{C_1} f(z)dzrightvert le frac{pi epsilon}{|z|^{1-a}|1+z|} le frac{pi epsilon^a}{1-epsilon} to 0 $$



$$ leftvertint_{C_3} f(z) dz rightvert le frac{2pi R}{|z|^{1-a}|1+z|} le frac{2pi R^a}{R-1} to 0 $$



Therefore



$$ (1-e^{i2pi a})I = 2pi ioperatorname*{Res}_{z=-1} f(z) = -2pi i e^{ipi a} $$



So finally



$$ I = frac{2pi ie^{ipi a}}{e^{2pi a}-1} = frac{2pi i}{e^{ipi a}-e^{-ipi a}} = frac{pi}{sin(api)} $$





Edit: Here's a simpler method, credit to @Poujh



$$ I = int_{-infty}^infty frac{e^{ax}}{e^x+1}dx = int_{-infty}^infty f(x) dx $$



Take a rectangular contour:





  • $C_1$: Straight line from $-R$ to $R$


  • $C_2$: Straight line from $R$ to $R + i2pi$


  • $C_3$: Straight line from $R + i2pi$ to $-R + i2pi$


  • $C_4$: Straight line from $-R + i2pi$ to $-R$




In the limit of $Rtoinfty$, we have



$$ leftvertint_{C_2} frac{e^{az}}{e^z+1} dzrightvert = leftvertint_0^{2pi} frac{e^{a(R+iy)}}{e^{R+iy}+1} dyrightvert le 2pi leftvertfrac{e^{aR}}{e^R-1}rightvert to 0 $$



$$ leftvertint_{C_4} frac{e^{az}}{e^z+1} dzrightvert = leftvertint_0^{2pi} frac{e^{a(-R+iy)}}{e^{-R+iy}+1} rightvert le 2pi leftvert frac{e^{-aR}}{1-e^{-R}} rightvert to 0 $$



For the horizontal lines



$$ int_{C_1} frac{e^{az}}{e^z+1} dz to int_{-infty}^infty frac{e^{ax}}{e^x+1} dx = I $$



$$ int_{C_3} frac{e^{az}}{e^z+1} dz to -int_{-infty}^infty frac{e^{ax}e^{i2pi a}}{e^x+1} = -e^{i2pi a} I $$



We end up with the same result



$$ (1-e^{i2pi a})I = 2pi ioperatorname*{Res}_{z=ipi} frac{e^{az}}{e^z+1} $$



where the residue is computed using the limit



$$ lim_{zto ipi} e^{az}frac{z-ipi}{e^z+1} = -e^{ipi a} $$






share|cite|improve this answer











$endgroup$













  • $begingroup$
    What about the other poles in this second method? Of course $exp(z)+1$ has zeros at $(2n+1)pi i$.
    $endgroup$
    – John
    Dec 23 '18 at 12:22






  • 1




    $begingroup$
    @John The other poles are outside the contour
    $endgroup$
    – Dylan
    Dec 23 '18 at 13:16








  • 1




    $begingroup$
    yes I got it and the residue is $- exp(i pi a)$, you can edit it.
    $endgroup$
    – John
    Dec 23 '18 at 13:23














1












1








1





$begingroup$

The substitution $e^x mapsto x$ gives



$$ I = int_{-infty}^infty frac{e^{ax}}{e^x+1}dx = int_0^infty frac{1}{x^{1-a}(1+x)}dx $$



where $0 < 1-a < 1$



We can use complex integration here. Due to the pole at $z=-1$, we pick a branch cut on the positive real axis such that $0 le arg(z) < 2pi$. The contour is a standard "keyhole contour" consisting of





  • $C_1$: left semicircle centered at $0$ with radius $epsilon$, going clockwise from $-epsilon i$ to $epsilon i$


  • $C_2$: straight line from $epsilon i$ to $R + epsilon i$


  • $C_3$: circle centered at $0$ with radius $R$, going counterclockwise from $R+iepsilon$ to $R - epsilon i$


  • $C_4$: straight line from $R - epsilon i$ to $-epsilon i$


In the limits $epsilon to 0$ and $R to infty$, the two straight line integrals converge to



$$ int_{C_2} f(z) dz to int_0^infty frac{1}{x^{1-a}e^{i0(1-a)}(1+x)}dx = I $$



$$ int_{C_4} f(z) dz to -int_0^infty frac{1}{x^{1-a}e^{i2pi(1-a)}(1+x)}dx = -e^{i2pi a}I $$



The remaining integrals should go to $0$



$$ leftvertint_{C_1} f(z)dzrightvert le frac{pi epsilon}{|z|^{1-a}|1+z|} le frac{pi epsilon^a}{1-epsilon} to 0 $$



$$ leftvertint_{C_3} f(z) dz rightvert le frac{2pi R}{|z|^{1-a}|1+z|} le frac{2pi R^a}{R-1} to 0 $$



Therefore



$$ (1-e^{i2pi a})I = 2pi ioperatorname*{Res}_{z=-1} f(z) = -2pi i e^{ipi a} $$



So finally



$$ I = frac{2pi ie^{ipi a}}{e^{2pi a}-1} = frac{2pi i}{e^{ipi a}-e^{-ipi a}} = frac{pi}{sin(api)} $$





Edit: Here's a simpler method, credit to @Poujh



$$ I = int_{-infty}^infty frac{e^{ax}}{e^x+1}dx = int_{-infty}^infty f(x) dx $$



Take a rectangular contour:





  • $C_1$: Straight line from $-R$ to $R$


  • $C_2$: Straight line from $R$ to $R + i2pi$


  • $C_3$: Straight line from $R + i2pi$ to $-R + i2pi$


  • $C_4$: Straight line from $-R + i2pi$ to $-R$




In the limit of $Rtoinfty$, we have



$$ leftvertint_{C_2} frac{e^{az}}{e^z+1} dzrightvert = leftvertint_0^{2pi} frac{e^{a(R+iy)}}{e^{R+iy}+1} dyrightvert le 2pi leftvertfrac{e^{aR}}{e^R-1}rightvert to 0 $$



$$ leftvertint_{C_4} frac{e^{az}}{e^z+1} dzrightvert = leftvertint_0^{2pi} frac{e^{a(-R+iy)}}{e^{-R+iy}+1} rightvert le 2pi leftvert frac{e^{-aR}}{1-e^{-R}} rightvert to 0 $$



For the horizontal lines



$$ int_{C_1} frac{e^{az}}{e^z+1} dz to int_{-infty}^infty frac{e^{ax}}{e^x+1} dx = I $$



$$ int_{C_3} frac{e^{az}}{e^z+1} dz to -int_{-infty}^infty frac{e^{ax}e^{i2pi a}}{e^x+1} = -e^{i2pi a} I $$



We end up with the same result



$$ (1-e^{i2pi a})I = 2pi ioperatorname*{Res}_{z=ipi} frac{e^{az}}{e^z+1} $$



where the residue is computed using the limit



$$ lim_{zto ipi} e^{az}frac{z-ipi}{e^z+1} = -e^{ipi a} $$






share|cite|improve this answer











$endgroup$



The substitution $e^x mapsto x$ gives



$$ I = int_{-infty}^infty frac{e^{ax}}{e^x+1}dx = int_0^infty frac{1}{x^{1-a}(1+x)}dx $$



where $0 < 1-a < 1$



We can use complex integration here. Due to the pole at $z=-1$, we pick a branch cut on the positive real axis such that $0 le arg(z) < 2pi$. The contour is a standard "keyhole contour" consisting of





  • $C_1$: left semicircle centered at $0$ with radius $epsilon$, going clockwise from $-epsilon i$ to $epsilon i$


  • $C_2$: straight line from $epsilon i$ to $R + epsilon i$


  • $C_3$: circle centered at $0$ with radius $R$, going counterclockwise from $R+iepsilon$ to $R - epsilon i$


  • $C_4$: straight line from $R - epsilon i$ to $-epsilon i$


In the limits $epsilon to 0$ and $R to infty$, the two straight line integrals converge to



$$ int_{C_2} f(z) dz to int_0^infty frac{1}{x^{1-a}e^{i0(1-a)}(1+x)}dx = I $$



$$ int_{C_4} f(z) dz to -int_0^infty frac{1}{x^{1-a}e^{i2pi(1-a)}(1+x)}dx = -e^{i2pi a}I $$



The remaining integrals should go to $0$



$$ leftvertint_{C_1} f(z)dzrightvert le frac{pi epsilon}{|z|^{1-a}|1+z|} le frac{pi epsilon^a}{1-epsilon} to 0 $$



$$ leftvertint_{C_3} f(z) dz rightvert le frac{2pi R}{|z|^{1-a}|1+z|} le frac{2pi R^a}{R-1} to 0 $$



Therefore



$$ (1-e^{i2pi a})I = 2pi ioperatorname*{Res}_{z=-1} f(z) = -2pi i e^{ipi a} $$



So finally



$$ I = frac{2pi ie^{ipi a}}{e^{2pi a}-1} = frac{2pi i}{e^{ipi a}-e^{-ipi a}} = frac{pi}{sin(api)} $$





Edit: Here's a simpler method, credit to @Poujh



$$ I = int_{-infty}^infty frac{e^{ax}}{e^x+1}dx = int_{-infty}^infty f(x) dx $$



Take a rectangular contour:





  • $C_1$: Straight line from $-R$ to $R$


  • $C_2$: Straight line from $R$ to $R + i2pi$


  • $C_3$: Straight line from $R + i2pi$ to $-R + i2pi$


  • $C_4$: Straight line from $-R + i2pi$ to $-R$




In the limit of $Rtoinfty$, we have



$$ leftvertint_{C_2} frac{e^{az}}{e^z+1} dzrightvert = leftvertint_0^{2pi} frac{e^{a(R+iy)}}{e^{R+iy}+1} dyrightvert le 2pi leftvertfrac{e^{aR}}{e^R-1}rightvert to 0 $$



$$ leftvertint_{C_4} frac{e^{az}}{e^z+1} dzrightvert = leftvertint_0^{2pi} frac{e^{a(-R+iy)}}{e^{-R+iy}+1} rightvert le 2pi leftvert frac{e^{-aR}}{1-e^{-R}} rightvert to 0 $$



For the horizontal lines



$$ int_{C_1} frac{e^{az}}{e^z+1} dz to int_{-infty}^infty frac{e^{ax}}{e^x+1} dx = I $$



$$ int_{C_3} frac{e^{az}}{e^z+1} dz to -int_{-infty}^infty frac{e^{ax}e^{i2pi a}}{e^x+1} = -e^{i2pi a} I $$



We end up with the same result



$$ (1-e^{i2pi a})I = 2pi ioperatorname*{Res}_{z=ipi} frac{e^{az}}{e^z+1} $$



where the residue is computed using the limit



$$ lim_{zto ipi} e^{az}frac{z-ipi}{e^z+1} = -e^{ipi a} $$







share|cite|improve this answer














share|cite|improve this answer



share|cite|improve this answer








edited Dec 23 '18 at 16:46

























answered Dec 20 '18 at 17:22









DylanDylan

12.9k31027




12.9k31027












  • $begingroup$
    What about the other poles in this second method? Of course $exp(z)+1$ has zeros at $(2n+1)pi i$.
    $endgroup$
    – John
    Dec 23 '18 at 12:22






  • 1




    $begingroup$
    @John The other poles are outside the contour
    $endgroup$
    – Dylan
    Dec 23 '18 at 13:16








  • 1




    $begingroup$
    yes I got it and the residue is $- exp(i pi a)$, you can edit it.
    $endgroup$
    – John
    Dec 23 '18 at 13:23


















  • $begingroup$
    What about the other poles in this second method? Of course $exp(z)+1$ has zeros at $(2n+1)pi i$.
    $endgroup$
    – John
    Dec 23 '18 at 12:22






  • 1




    $begingroup$
    @John The other poles are outside the contour
    $endgroup$
    – Dylan
    Dec 23 '18 at 13:16








  • 1




    $begingroup$
    yes I got it and the residue is $- exp(i pi a)$, you can edit it.
    $endgroup$
    – John
    Dec 23 '18 at 13:23
















$begingroup$
What about the other poles in this second method? Of course $exp(z)+1$ has zeros at $(2n+1)pi i$.
$endgroup$
– John
Dec 23 '18 at 12:22




$begingroup$
What about the other poles in this second method? Of course $exp(z)+1$ has zeros at $(2n+1)pi i$.
$endgroup$
– John
Dec 23 '18 at 12:22




1




1




$begingroup$
@John The other poles are outside the contour
$endgroup$
– Dylan
Dec 23 '18 at 13:16






$begingroup$
@John The other poles are outside the contour
$endgroup$
– Dylan
Dec 23 '18 at 13:16






1




1




$begingroup$
yes I got it and the residue is $- exp(i pi a)$, you can edit it.
$endgroup$
– John
Dec 23 '18 at 13:23




$begingroup$
yes I got it and the residue is $- exp(i pi a)$, you can edit it.
$endgroup$
– John
Dec 23 '18 at 13:23



Popular posts from this blog

Quarter-circle Tiles

build a pushdown automaton that recognizes the reverse language of a given pushdown automaton?

Mont Emei