Hi Brilliant! Here is the most awaited Integration contest! See Season 1 and Season 2.
The aims of the Integration contest are to improve skills in the computation of integrals, to learn from each other as much as possible, and of course to have fun. Anyone here may participate in this contest.
The rules are as follows:
I will start by posting the first problem. If there is a user solves it, then they must post a new one.
You may only post a solution of the problem below the thread of problem and post your proposed problem in a new thread. Put them separately.
Only make substantial comment that will contribute to the discussion.
Make sure you know how to solve your own problem before posting it in case there is no one can answer it within 48 hours, then you must post the solution and you have a right to post another problem.
If the one who solves the last problem does not post his/her own problem after solving it within a day, then the one who has a right to post a problem is the last solver before him/her.
The scope of questions is only computation of integrals either definite or indefinite integrals.
You are NOT allowed to post a multiple integrals problem.
It is NOT compulsory to post original problems. But make sure it has not been posted on brilliant.
Do not copy questions from last year's contest. If anyone found to do so he/she will be banned from taking further part in this contest
Direct problems on contour integration are not allowed. However, solutions which use contour integration are allowed.
Also don't forget to upvote good problems and solutions!
Format your post is as follows:
1 2 3 4 5 6 7 |
|
The comments will be easiest to follow if you sort by "Newest":
Easy Math Editor
This discussion board is a place to discuss our Daily Challenges and the math and science related to those challenges. Explanations are more than just a solution — they should explain the steps and thinking strategies that you used to obtain the solution. Comments should further the discussion of math and science.
When posting on Brilliant:
*italics*
or_italics_
**bold**
or__bold__
paragraph 1
paragraph 2
[example link](https://brilliant.org)
> This is a quote
\(
...\)
or\[
...\]
to ensure proper formatting.2 \times 3
2^{34}
a_{i-1}
\frac{2}{3}
\sqrt{2}
\sum_{i=1}^3
\sin \theta
\boxed{123}
Comments
Problem 6:
Prove that ∫0∞(x+a)2+b2lnxdx=b1tan−1ablna2+b2 for all a,b>0.
This problem has been solved by Aditya Sharma.
Log in to reply
Solution to problem 6 :
Let us denote the integral by I .
I=∫0∞(x+a)2+b2lnxdx=∫0∞(x+a+ib)(x+a−ib)lnx
With the substitution xy=a2+b2 we have,
I=ln(a2+b2)∫0∞y2+2ay+a2+b2dy−I∫0∞(y+a)2+b2lnydy
I=ln(a2+b2)∫0∞(x+a)2+b2dx=b1tan−1ablna2+b2.
As an alternative to Aditya's proof...
Consider the cut plane C\[0,∞), so that 0<Argz<2π for all z, and consider the integral of f(z)=(z+a)2+b2(logz)2 around the keyhole contour γ1+γ2−γ3−γ4, where
where we assume that 0<ε<a2+b2<R. Now ∫γ1f(z)dz∫γ3f(z)dz(∫γ1−∫γ3)f(z)dz=∫εR(x+a)2+b2(lnx)2dx=∫εR(x+a)2+b2(lnx+2πi)2dx=∫εR(x+a)2+b24π2−4πilnxdx We also note that ∫γ2f(z)dz=O(Rln2R)∫γ4f(z)dz=O(εln2ε) as R→∞ and ε→0. Letting R→∞ and ε→0, we deduce that ∫0∞(x+a)2+b24π2−4πilnxdx=2πi[Resz=−a+ibf(z)+Resz=−a−ibf(z)] Now f(z) has simple poles at each of −a±ib, and Resz=−a±ibf(z)Resz=−a+ibf(z)+Resz=−a−ibf(z)=±2ib(lna2+b2+iπ∓itan−1ab)2=2ib−4i(lna2+b2+iπ)tan−1ab=−b2tan−1ab(lna2+b2+iπ) and hence ∫0∞(x+a)2+b24π2−4πilnxdx=b4πtan−1ab(π−ilna2+b2) Taking imaginary parts of this equation gives the result.
Problem 16:
Evaluate the integral ∫01(x+p)a+bxa−1(1−x)b−1dx for all a,b,p>0.
This problem has been solved by Aditya Sharma.
Log in to reply
If we use the substitution, x=u+11 then
I=∫0∞(up+p+1)a+bub−1du
Again using the substitution, u=y(pp+1) the integral transforms to,
I=pb(1+p)a1∫0∞(1+y)a+byb−1dy=(pb(1+p)aβ(a,b))
Log in to reply
The substitution z=p+x(p+1)x Is even more direct. Not a hard problem, but a very elegant result.
Log in to reply
Problem 14 : Prove That
∫0∞lnx[ln(2x+1)−x+11−ψ(2x+1)]dx=2ln22+ln2⋅lnπ−1
Notation : ψ(x) denotes the Digamma Function.
This problem has been solved by Mark Hennings.
Log in to reply
This is a really nice use of previous results of this round of the contest!
Starting with the identity ψ(x)=lnx−2x1−2∫0∞(t2+x2)(e2πt−1)tdt we deduce that ln(2x+1)+x+11−ψ(2x+1)=2∫0∞(t2+41(x+1)2)(e2πt−1)tdt=8∫0∞(4t2+(x+1)2)(e2πt−1)tdt=2∫0∞(t2+(x+1)2)(eπt−1)tdt and hence ∫0∞lnx(ln(2x+1)+x+11−ψ(2x+1))dx=2∫0∞(∫0∞(x+1)2+t2lnxdx)eπt−1tdt=2∫0∞(t1tan−1tlnt2+1)eπt−1tdt=∫0∞eπt−1tan−1tln(t2+1)dt=21ln22+ln2lnπ−1 using the results of Problems 5 and 6.
Ishan, please set another one!
Log in to reply
Yes indeed! This integral arose from my attempt to find an alternate solution of Problem 5 :) The first identity can be proved (using real analysis) by differentiating the identity ∫0∞1+x2log(1−e−2aπx)dx=π[21log(2aπ)+a(loga−1)−log(Γ(a+1))] wrt a. I have proved it here.
I have posted a question on Math Stack Exchange asking for alternate methods to evaluate the above integral (independently of Problems 5 and 6) here. If we can evaluate it using other methods, it'll also give us an alternate solution for Problem 5.
Problem 1:
Show that: ∫01xcosh(alog(x))log(1+x)dx=2a1(πcsc(πa)−a1), a<1
This problem has been solved by Aditya Sharma.
Log in to reply
If we denote the integral by I & using 2cosh(alnx)=xa+x−a & ln(1+x)=n≥1∑n(−1)n−1xn the integral can be transformed into,
I=n≥1∑n(−1)n−1∫01(xn+a−1+xn−a−1)=n≥1∑n(−1)n−1(n−a1+n+a1)=2a1n≥1∑(−1)n−1(n−a1−n+a1)
For removing the alternating factor (−1)n−1 this can be written as ,
2aI=n≥1∑(n−a1−n+a1)−2n≥1∑(2n−a1−2n+a1)
which simplifies to ,
2aI=n≥1∑(n−a1−n+a1−n−2a1+n+2a1)
Using Ha=n≥1∑(n1−n+a1) we have,
I=2a1(Ha−H−a−H2a+H−2a)
Again using reflection formula for harmonic numbers as a<1
i.e. H1−a−Ha=πcot(πa)−a1+1−a1 it simplifies to,
I=2a1(−πcot(πa)+πcot(π2a)−a1)=2a1(πcsc(πa)−a1)
Problem 5:
Prove that:
∫0∞eπx−1ln(x2+1)arctanxdx=2ln22+ln2⋅lnπ−1.
This problem has been solved by Mark Hennings.
Log in to reply
Solution of Problem 5
Starting with the Hermite formula for the generalized Riemann zeta function: ζ(s,a)=21a−s+a1−s(s−1)−1+2∫0∞e2πy−1(a2+y2)−21ssin(stan−1ay)dy we have ζ(s,21)=2s−1(1+s−11)+2∫0∞e2πy−1(41+y2)−21ssin(stan−12y)dy=2s−1s−1s+2s+1∫0∞e2πy−1(1+4y2)−21ssin(tan−12y)dy=2s−1s−1s+2s∫0∞eπt−1(1+t2)−21ssin(stan−1t)dt so that F(s)=∫0∞eπt−1(1+t2)−21ssin(stan−1t)dt=2−sζ(s,21)−2(s−1)s Now since F′′(s)=∂s2∂2∫0∞eπt−1(1+t2)−21ssin(stan−1t)dt=∫0∞eπt−1(1+t2)−21s{41(ln(1+t2))2sin(stan−1t)−ln(1+t2)tan−1tcos(stan−1t)−(tan−1t)2sin(stan−1t)}dt we deduce that F′′(0)=−∫0∞eπt−1ln(1+t2)tan−1tdt and hence that the desired integral is ∫0∞eπt−1ln(1+t2)tan−1tdt=−F′′(0)=∂s2∂2[2(s−1)s−2−sζ(s,21)]∣∣∣s=0 which simplifies to give ∫0∞eπt−1ln(1+t2)tan−1tdt=−1−21ln22+ln2⋅ln2π=−1+21ln22+ln2lnπ
Problem 8:
Show that ∫021πln2(asinθ)dθ=241π3+61π[ln2(a2)−2ln(a2)ln(2a)] for any a>0.
This problem has been solved by Jasper Braun.
Log in to reply
Solution to Problem 8: I=∫02πln2a+2lnalnsinx+ln2sinxdx=2πln2a−lnaπln2+∫02πln2sinxdx
∫02πln2sinxdx=∫011−x2ln2xdx(by u=sinx)
Now using B(m,n)=2∫01x2m−1(1−x2)n−1dx and thus ∫011−x2ln2xdx=dm2d22∗4B(m,21)∣∣21
Using Γ′(m)=Γ(m)ψ(m),Γ′′(m)=Γ(m)ψ(m)2+Γ(m)ψ(1)(m)
8πdm2d2Γ(m+21)Γ(m)∣∣∣21=2πln22+4πζ(2)
I=24π3+2πln22+2πln2a−πlnaln2
Problem 13:
Evaluate ∫02πecosθcos(nθ−sinθ)dθ for any integer n.
This problem has been solved by Ishan Singh.
Log in to reply
It's simplest using contour integration...
Note that ∫02πecosθcos(sinθ−nθ)dθ=Re(∫02πecosθei(sinθ−nθ)dθ)=Re(∫02πeeiθ−inθdθ) With the substitution z=eiθ, this becomes the contour integral Re(∫∣z∣=1ezz−nizdz)=Re(i1∫∣z∣=1zn+1ezdz)=2πRe(Resz=0zn+1ez)={n!2π0n≥0n<0
I=∫02πecostcos(nt−sint) dt=ℜ(∫02πeint⋅e(cost−isint) dt)
=ℜ(∫02πeint⋅ee−it dt)
=n!2π+ℜ⎝⎛r≥0 ; r=n∑(r!1∫02πeit(n−r) dt)⎠⎞
Since last integral is 0, we have,
I=n!2π
If n is negative, we can write n=−k ; k∈Z+, so that,
J=∫02πecostcos(kt+sint) dt
By the above method, we can similarly show that J=0.
Therefore,
∫02πecostcos(nt−sint) dt={n!2π0n≥0n<0
Problem 2:
Evaluate ∫01x(x2+1)tan−13xdx.
This problem has been solved by Aditya Sharma.
Log in to reply
The integral can be written as : I=I1∫01xtan−1(3x)dx+I2∫03xtan−1xdx
A simple integration by parts would yield I1=95tan−13−61
I2 is the Inverse Tangent Integral and we represent it by dilogarithms.
Ti2(3)=2i1(Li2(3i)−Li2(−3i))
So, I=95tan−13−61+2i1(Li2(3i)−Li2(−3i))
Problem 03:
Show that ∫01xlnxln2(1+x)dx=24π4−4Li4(21)−27ζ(3)ln2+6π2ln22−6ln42.
This problem has been solved by Jasper Braun.
Log in to reply
Solution to Problem 3:
Hn(k)=j=1∑∞inHn(k)xiH(x)=1−x−ln(1−x)
I=∫0∞xln2(1+x)lnxdx=−∫011+xln(1+x)ln2xdx=by parts
−∫12xln2(x−1)lnxdx=(by u=1+x)
−∫211xln2(x1−1)lnx1dx=∫211xlnxln2(1−x)−x2ln2xln(1−t)+xln3xdx(by u=1/x)
∫xlnxln2(1−x)=2ln2xln2(1−x)+∫1−xln2xln(1−x)dx
∫1−xln2xln(1−x)dx=−n=1∑∞Hndn2d2∫xndx=−n=1∑∞Hn(n+1xn+1ln2x−(1+n)22xn+1lnx+(1+n)32xn+1)=
Now use Hn=Hn+1−n+11 to obtain:
−(H1(x)ln2x−Li2(x)ln2x−2H2(x)lnx+2Li3(x)lnx+2H3(x)−2Li4(x))
Next −2∫xln2xln(1−x)dx=2n=1∑∞n1dn2d2∫xn−1=
2n=1∑∞n2xnln2x−n32xnlnx+n42xn=2(Li2(x)ln2x−2Li3(x)lnx+2Li4(x))
I=4ln4x∣∣∣211+2ln2xln2(1−x)∣∣∣211−2H3(1)+2Li4(1)+
+H1(21)ln22−Li2(21)ln22+2H2(21)ln2−2Li3(21)ln2+2H3(21)−2Li4(21)+
4Li4(1)−2Li2(21)ln22−4Li3(21)ln2−4Li4(21)=
4−3ln42−2H3(1)+H1(21)ln22+2H2(21)ln2+2H3(21)−6Li4(21)+6Li4(1)−3Li2(21)ln22−6Li3(21)ln2
Now using this for the evaluation of Hn(x)
−2H3(1)=36−π4,ln22H1(21)=2ln42+ln22Li221,2ln2H221=2ln2ζ(3)−ζ(2)ln22,2H3(21)=360π4+12ln42−4ζ(3)ln2+2Li4(21)
Adding all this together and using formulas for Li3 and Li2(I have checked on a calculator) gets the desired answer. Anyone may post the next problem.
Problem 4:
Prove that ∫01x2Li32(−x)dx=−169ζ2(3)−43ζ(2)ζ(3)−41ζ2(2)+47ζ(4)−3ζ(3)ln2+6ζ(3)−6ζ(2)ln2+6ζ(2)−12ln22
This problem has been solved by Aditya Kumar.
Log in to reply
Solution to Problem 4:
I=∫01x2Li32(−x)dx
I=−xLi32(−x)∣∣∣∣∣01+∫01x22Li2(−x)Li3(−x)dx=−169ζ2(3)−x2Li2(−x)Li3(−x)∣∣∣∣∣01+∫01x22Li22(−x)dx−∫01x22Li3(−x)ln(1+x)dx
I=−169ζ2(3)−43ζ(2)ζ(3)−x2Li22(−x)∣∣∣∣∣01−∫01x24Li2(−x)ln(1+x)dx+x2Li3(−x)ln(1+x)∣∣∣∣∣01−∫01x(1+x)2Li3(−x)dx−∫01x22Li2(−x)ln(1+x)dx
I=−169ζ2(3)−43ζ(2)ζ(3)−21ζ2(2)+47ζ(4)−23ζ(3)ln2+∫011+x2Li3(−x)dx+x6Li2(−x) ln(1+x)∣∣∣∣∣01−∫01x(1+x)6Li2(−x)dx+∫01x26ln2(1+x)dx
I=−169ζ2(3)−43ζ(2)ζ(3)−21ζ2(2)+47ζ(4)−23ζ(3)ln2+ 92ζ(3)−3ζ(2)ln2+∫011+x6Li2(−x)dx+∫011+x2Li3(−x)dx−x6ln2(1+x)∣∣∣∣∣01+∫01x(1+x)12ln(1+x)dx
I=−169ζ2(3)−43ζ(2)ζ(3)−21ζ2(2)+47ζ(4)−23ζ(3)ln2+29ζ(3)−3ζ(2)ln2+6ζ(2)−12ln22+∫011+x6Li2(−x)dx+∫011+x2Li3(−x)dx
Now,
I1=∫011+x6Li2(−x)dx=6Li2(−x)ln(1+x)∣∣∣∣∣01+∫01x6ln2(1+x)dx
I1=23ζ(3)−3ζ(2)ln2
I2=∫011+x2Li3(−x)dx=2Li3(−x)ln(1+x)∣∣∣∣∣01−∫012Li2(−x)dLi2(−x)x2Li2(−x)ln(1+x)dx
I2=41ζ2(2)−23ζ(3)ln2
Therefore,
∫01x2Li32(−x)dx=−169ζ2(3)−43ζ(2)ζ(3)−41ζ2(2)+47ζ(4)−3ζ(3)ln2+6ζ(3)−6ζ(2)ln2+6ζ(2)−12ln22
Problem 10:
Show that ∫01(x−1)e−xlnxdx=ee−1
This problem has been solved by Jasper Braun.
Log in to reply
Solution to Problem 10:
∫01xe−xlnx−e−xlnxdx=∫01e−xlnxdx+∫01e−xdx−∫01e−xlnxdx=1−e1 (by IBP)
Log in to reply
You can also apply IBP this way (though almost same way),
−∫01(1−x)e−xlnx dx=−∫01lnx d(xe−x)
@Aditya Kumar @Aditya Sharma @Ishan Singh @Mark Hennings @Rohith M.Athreya and all other participants:
I have made this to make seeing old problems easier. I will complete it by tomorrow.
Log in to reply
That's very nice of you to compile the past problems! Thanks :)
Log in to reply
Sure! I hope it will make things easier!
really nice :)
thank you
Problem 11:
Show that ∫01(1+t)lnttα−1−tβ−1dt=ln(Γ(21+21β)Γ(21α)Γ(21+21α)Γ(21β)) for all α,β>0.
This problem has been solved by Aditya Sharma.
Log in to reply
Results used : n≥0∑n+a1=−ψ(a), ∫ψ(ax)=a1ln(Γ(ax))
Let F(a)=∫01(1+x)lnxxa−1dx then by differentiating w.r.t to a,
F′(a)=∫011+xxa−1dx=n≥0∑(−1)nn+a1
Now , n≥0∑(−1)nn+a1=21(n≥0∑n+2a1−n≥0∑n+2a+11)=21(ψ(2a+1)−ψ(2a))
So , F′(α)−F′(β)=21(ψ(2α+1)−ψ(2α))−21(ψ(2β+1)−ψ(2β))
Integrating we have, F(α)−F(β)=(ln(Γ(21+β)Γ(2α)Γ(21+α)Γ(2β)))+C
Now since, F(1)=0⟹C=0
So F(α)−F(β)=(lnΓ(21+β)Γ(2α)Γ(21+α)Γ(2β))
Problem 18 : Prove That
∫0∞log(a2+2acosqx+1a2+2acospx+1) xdx=2log(pq)log(1+a) ; −1<a≤1
This problem has been solved by Mark Hennings.
Log in to reply
There is some delicate interchanging of integrals and infinite sums needed here, so I will set this out in full. Fix p,q>0. For 0<ε<X we have ∫εXxcoskpx−coskqxdx=(∫pεpX−∫qεqX)xcoskxdx=∫qXpXxcoskxdx−∫qεpεxcoskxdx which implies that ∣∣∣∣∣∫εXxcoskpq−coskqxdx∣∣∣∣∣≤min(p,q)2∣p−q∣0<ε<X Thus, if we define the functions fk(X)=∫0Xxcoskpx−coskqxdxk∈N,X>0 then we have shown that ∣∣fk(X)∣∣≤min(p,q)2∣p−q∣k∈N,X>0 and hence the functions fk(X) are uniformly bounded in both k and X. Moreover, the well-known Frullani integrals ∫0∞xcosax−cosbxdx=ln(ab)∫0∞xsinax−sinbxdx=0 which hold whenever a and b are nonzero with the same sign, imply that ∫0∞xeirpx−eirqxdx=ln(pq)r=0 and hence, since fk(X)=2−k∫0Xx(eipx+eipx)k−(eiqx+e−iqx)kdx=2−kr=0∑k(rk)∫0Xxei(2r−k)px−ei(2r−k)qxdx we deduce that φk=X→∞limfk(X)=⎩⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎧2−kr=0∑k(rk)ln(pq)2−kr=21k0≤r≤k∑(rk)ln(pq)k oddk even and hence that φk={ln(pq)ln(pq)(1−2−k(21kk))k oddk even Now, for −1<a<1, ∫0Xln(1+2acosqx+a21+2acospx+a2)xdx=∫0X{ln(1+1+a22acospx)−ln(1+1+a22acosqx)}xdx=∫0Xk=1∑∞k(−1)k−1(a2+12a)k[coskpx−coskqx]xdx=k=1∑∞k(−1)k−1(a2+12a)kfk(X) where the interchange of the sum and integral in the last line is valid because ∣∣a2+12a∣∣<1, and hence the series is uniformly convergent in x. We can now use the uniform boundedness of the functions fk(X) to deduce that ∫0∞ln(1+2acosqx+a21+2acospx+a2)xdx=X→∞lim∫0Xln(1+2acosqx+a21+2acospx+a2)xdx=k=1∑∞k(−1)k−1(a2+12a)kφk Thus the integral is equal to ln(pq)[k=1∑∞k(−1)k−1(a2+12a)k+k=1∑∞2k1(a2+12a)2k2−2k(k2k)]=ln(pq)[ln(1+a2+12a)+21k=1∑∞k1(k2k)((a2+1)2a2)k]=ln(pq)[ln(a2+1(1+a)2)−ln(21[1+1−4(a2+1)2a2])]=2ln(pq)ln(1+a) as required.
Log in to reply
(+1) Nice solution! There is a slight typo in the Frullani Integral for cosine, the integral is for a and b whereas the logarithm is for p and q.
Problem 19:
Show that ∫0∞xe−xp−e−xqdx=pqp−qγ for p,q>0, where γ is the Euler-Mascheroni constant.
The problem has been solved first by Aditya Sharma and then by Ishan Singh.
Log in to reply
Solution to Problem 19
I=∫0∞(e−xp−e−xq)d(lnx)
I=[(e−xp−e−xq)lnx]0∞+p∫0∞e−xpxp−1lnx−q∫0∞e−xqxq−1lnxdx
The substitutions xp=u,xq=v makes the last two integrals as,
p∫0∞e−xpxp−1lnx−q∫0∞e−xqxq−1lnxdx=∫0∞e−xxlnxdx(qpq−p)
=∫0∞e−xlnxdx(qpq−p)=γ(qpp−q)
Now [(e−xp−e−xq)lnx]0∞=x→∞lime−xp−e−xq1lnx−x→0lime−xp−e−xq1lnx
Both of these can be evaluated by applying L-Hospital's rule succesively and both of them are 0.
Hence ∫0∞x(e−xp−e−xq)dx=γ(pqp−q)
Solution to Problem 19 :
Proof : Let,
J=∫0∞(e−x−x+11)xdx
=∫0∞∫0∞e−x(a+1)−x+1e−axdx da
=∫0∞a+11da−∫0∞∫0∞x+1e−axdx da
Substitute ax↦x in the second integral,
⟹J=∫0∞a+11da−∫0∞∫0∞x+ae−xdx da
Since ∫0∞e−xdx=1
⟹J=∫0∞∫0∞a+1e−x−x+ae−xdx da
=∫0∞e−x∫0∞a+11−x+a1da dx
=∫0∞e−x[ln(a+xa+1)]a=0a→∞dx
=∫0∞e−xlnx dx
=−γ□
Proof : ∫01x1−e−xdx−∫1∞xe−xdx=∫0∞∫01(1−e−x)e−yxdx dy−∫0∞∫1∞e−x(y+1)dx dy
=∫0∞(y1−e−y+y+1e−(y+1)−1)dy−∫0∞y+1e(y+1)dy
=−∫0∞(e−y−y+11)ydy
=γ (Using Proposition 1) □
Now,
I=∫0∞xe−xp−e−xq dx
=(∫01x1−e−xq dx−∫1∞xe−xq dx)−(∫01x1−e−xp dx−∫1∞xe−xp dx)
Substitute xq↦x in the first integral and xp↦x in the second integral to get,
I=q1(∫01x1−e−x dx−∫1∞xe−x dx)−p1(∫01x1−e−x dx−∫1∞xe−x dx)
Using Proposition 2, we get,
I=(q1−p1)γ
∴I=pqp−qγ□
Problem 20 :
Show that ∫011+xxlnxarctanxdx=21ln2−21Gln2+64π3+48π2−4π.
This problem has been solved by Mark Hennings and then by Fdp Dpf.
Log in to reply
Note that I=∫011+xxlnxtan−1xdx=∫011+xxlnx(∫0x1+u2du)dx=∫01(∫u11+xxlnxdx)1+u2du=∫01[xlnx−x−lnxln(1+x)−Li2(−x)]x=u11+u2du=∫01(−1+121π2−ulnu+u+lnuln(1+u)+Li2(−u)]1+u2du=−41π+481π2+481π3+21ln2+∫011+u2lnuln(1+u)+Li2(−u)du It is possible (integrating by parts and so forth) to calculate the indefinite integrals of both parts of the remaining integral in terms of polylogarithms, but the end result is extremely complicated! My expression for the indefinite integral ∫1+u2lnuln(1+u)du has over 30 terms, and the indefinite integral for ∫1+u2Li2(−u)du is even more complicated. The definite integrals are ∫011+u2lnuln(1+u)du∫011+u2Li2(−u)du=1281(11π3−256Gln2+10iπ2ln2+12π(ln2)2−8i(ln2)3−394iLi3(21+21i)+210ζ(3))=3841(−35π3−30iπ2ln2−36π(ln2)2+576Gln2+24i(ln2)3−768iLi3(21+21i)+630iζ(3)) so that ∫011+x2lnxln(1+x)+Li2(−x)dx=−1921π3−21Gln2 and hence I=−41π+481π2+641π3+21ln2−21Gln2 as required.
That was exhausting! Someone else can set the next one...
SOLUTION OF PROBLEM 20:
Alternative solution.
Let R(x)=∫0x1+ttlntdt=∫011+xttx2ln(tx)dt
Note that,
R(1)=12π2−1
(Taylor expansion of integrand)
I=∫011+xxlnxarctanxdx=[R(x)arctanx]01−∫011+x2R(x)dx=48π3−4π−∫01∫01(1+tx)(1+x2)tx2ln(tx)dtdx=48π3−4π−∫01∫01(1+tx)(1+x2)tx2lntdtdx−∫01∫01(1+tx)(1+x2)tx2lnxdtdx=48π3−4π−∫01[tlnt(2(1+t2)tln(1+x2)+t(t2+1)ln(1+tx)−1+t2arctanx)]x=0x=1dt−∫01⎣⎡1+x2lnx(tx−ln(1+tx))⎦⎤t=0t=1dx=48π3−4π−2ln2∫011+t2t2lntdt−∫011+t2lntln(1+t)dt+4π∫011+t2tlntdt−∫011+x2xlnxdx+∫011+x2lnxln(1+x)dx=48π3−4π−2ln2∫011+t2t2lntdt+(4π−1)∫011+t2tlntdt
Using Taylor expansion of1+x2x2lnx and 1+x2xlnx one obtains,
∫011+x2x2lnxdx=G−1
G being the Catalan constant.
∫011+x2xlnxdx=−48π2
Therefore,
I=64π3+48π2−4π−21Gln2+21ln2
Log in to reply
(+1) Nice solution! You might be interested in Part 2 of the current season.
Problem 7:
Show that for n≥2, ∫0∞xn+1lnxdx=−n2π2cot(nπ)csc(nπ)
This problem has been solved by Mark Hennings.
Log in to reply
Solution to Problem 7:
Consider the integrals of f(z)=zn+11g(z)=zn+1logz around the contour γ1+γ2−γ3, where
where R>1. Now ∫γ1f(z)dz∫γ3f(z)dz(∫γ1−∫γ3)f(z)dz∫γ1g(z)dz∫γ3g(z)dz(∫γ1−∫γ3)g(z)dz=∫0Rxn+11dx=en2πi∫0Rxn+11dx=−2ienπisinnπ∫0Rxn+11dx=∫0Rxn+1lnxdx=en2πi∫0Rxn+1lnx+inπdx=−2ienπisinnπ∫0Rxn+1lnxdx−inπen2πi∫0Rxn+11dx Since ∫γ2f(z)dz=O(R1−n)∫γ2g(z)dz=O(lnRR1−n) as R→∞, we deduce that (since both f(z) and g(z) just have simple poles at enπi inside this contour) −2ienπisinnπ∫0∞xn+11dx−2ienπisinnπ∫0Rxn+1lnxdx−inπen2πi∫0Rxn+11dx=2πiResz=enπif(z)=−n2πienπi=2πiResz=enπig(z)=n22π2enπi and so ∫0∞xn+11dx=nπcosecnπ∫0∞xn+1lnxdx=−n2π2cotnπcosecnπ
Problem 9:
Prove ∫01Li22(x)dx=36π4−4ζ(3)−2ζ(2)+6
This problem has been solved by Mark Hennings.
Log in to reply
I will be using the Euler result that u=1∑∞u2Hu=2ζ(3) Integrating by parts, ∫01Li22(x)dx=ζ(2)2−2∫01Li1(x)Li2(x)dx=ζ(2)2−2r,s=1∑∞rs21∫01xr+sdx=ζ(2)2−2r,s=1∑∞rs2(r+s+1)1=ζ(2)2−2r,s=1∑∞s2(s+1)1(r1−r+s+11)=ζ(2)2−2s=1∑∞s2(s+1)Hs+1=ζ(2)2−2s=1∑∞Hs+1(s21−s(s+1)1)=ζ(2)2−2s=1∑∞s2Hs+1+2s=1∑∞s(s+1)Hs+1=ζ(2)2−2s=1∑∞s2Hs−2s=1∑∞s2(s+1)1+2s=1∑∞Hs+1(s1−s+11)=ζ(2)2−4ζ(3)−2s=1∑∞(s21−s(s+1)1)+2S→∞lim(s=1∑SsHs+1−s=2∑S+1sHs)=ζ(2)2−4ζ(3)−2ζ(2)+2+2S→∞lim(23+s=2∑Ss(s+1)1−SHS+1)=ζ(2)2−4ζ(3)−2ζ(2)+6 as required.
Problem 12
Show that ∫0∞(xlnxarctanx)2dx=π34ln2−12π3+2πln2−4πζ(3)
This problem has been solved by Mark Hennings
Log in to reply
Consider the principal branch of the logarithm, so that the plane is cut along the negative real axis. Suppose that a>0. If we integrate f(z)=z2+a2(logz)2 around the contour γ1+γ2−γ3−γ4, where
where 0<ε<a<R. Then ∫γ1f(z)dz=∫εRx2+a2ln2xdx∫γ3f(z)dz=−∫εRx2+a2(lnx+iπ)2dx while ε→0lim∫γ4f(z)dz=R→∞lim∫γ2f(z)dz=0 and hence we deduce that ∫0∞x2+a2ln2x+(lnx+iπ)2dx=2πiResz=iaf(z)=aπ(lna+21πi)2 so that ∫0∞x2+a2ln2xdx=2aπln2a+8aπ3∫0∞x2+a2lnxdx=2aπlna
Now xtan−1x=∫011+x2u2du and hence ∫0∞(xtan−1xlnx)2dx=∫01∫01(∫0∞(1+x2u2)(1+x2v2)ln2xdx)dudv=∫01∫01(∫0∞ln2x{x2+u−21−x2+v−21}dx)u2−v2dudv=∫01∫01{21πuln2u+81π3u−21πvln2v−81π3v}u2−v2dudv=21π∫01∫01u2−v2uln2u−vln2vdudv+81π3∫01∫01u+vdudv=21π{−61π2+4ln2−21ζ(3)}+81π3×2ln2=π(41π2ln2−121π2+2ln2−41ζ(3)) I initially relied on Mathematica to evaluate the double integral ∫01∫01u2−v2uln2u−vln2vdudv but here is a proper derivation.
∫01∫01u2−v2uln2u−vln2vdudv=2∫01du∫0udvu2−v2uln2u−vln2v=2∫01du∫0udvu2−v21∫vu(2lnw+ln2w)dw=∫∫∫0<v<w<u<1u2lnw+ln2w(u+v1+u−v1)=∬0<w<u<1u2lnw+ln2wln(u−wu+w)dwdu Since it is easy to check (by differentiating back again) that ∫(2lnw+ln2w)ln(u−wu+w)dw=uln2wln(1+uw)+wln2wln(u−wu+w)+uln2wln(1−uw)+2ulnwLi2(−uw)+2ulnwLi2(uw)−2uLi3(−uw)−2uLi3(uw)+c it follows that ∫0u(2lnw+ln2w)ln(u−wu+w)dw=61u(π2lnu+12ln2ln2u−3ζ(3)) and so ∫01∫01u2−v2uln2u−vln2vdudv=61∫01(π2lnu+12ln2ln2u−3ζ(3))du=−61π2+4ln2−21ζ(3) as required.
Problem 15 : Prove That
∫0∞cosh2mxcos(2nx)dx=2πΓ(m)Γ(m+21)∣Γ(m+in)∣2 ; (m,n)∈R+
This problem has been solved by Mark Hennings.
Log in to reply
Note that B(z,w)=∫0∞(1+t)z+wtz−1dt=(∫01+∫1∞)(1+t)z+wtz−1dt=∫01(1+t)z+wtz−1dt+∫10(1+t−1)z+wt1−z(−t−2dt)=∫01(1+t)z+wtz−1+tw−1dt for all Rez,Rew>0, so that (with the substitution t=e−2x) B(m+in,m−in)=∫01(1+t)2mtm+in−1+tm−in−1dt=∫01(1+t)2mtm−1(tin+t−in)dt=∫∞0(1+e−2x)2me2(1−m)x(e−2inx+e2inx)(−2e−2x)dx=4∫0∞(1+e−2x)2me−2mxcos2nxdx=4∫0∞(ex+e−x)2mcos2nxdx=41−m∫0∞cosh2mxcos2nxdx for any real n and m>0, so that ∫0∞cosh2mxcos2nxdx=4m−1B(m+in,m−in)=Γ(2m)4m−1Γ(m+in)Γ(m−in)=2Γ(m)Γ(m+21)π∣∣Γ(m+in)∣∣2 for any real n and any m>0, using the duplication formula Γ(2m)=2π122m−21Γ(m)Γ(m+21)
Problem 17 :
Find a closed form for ∫0∞xsin2n+1xdx for n≥0
This problem has been solved by Ishan SIngh
Log in to reply
Since,
sinx=2ieix−e−ix
We have,
sin2n+1x=22n(−1)nr=0∑n(−1)r(r2n+1)sin(2r+1)x
⟹∫0∞xsin2n+1x dx=22n(−1)nr=0∑n(−1)r(r2n+1)∫0∞xsin(2r+1)x dx
=22n+1(−1)nπr=0∑n(−1)r(r2n+1)
=22n+1(−1)nπr=0∑n((−1)r(r2n)−(−1)r−1(r−12n))
Clearly, the above sum telescopes. Evaluating it, we have,
∫0∞xsin2n+1x dx=22n+1(−1)nπ(−1)n(n2n)
∴∫0∞xsin2n+1x dx=22n+1π(n2n) ∀ n∈Z+
Log in to reply
As an alternative,
I=∫0∞xsin2n+1xdx=∫0∞∫0∞e−xysin2n+1xdxdy
Now by IBP two times we can create a recurrence relation and thus evaluate,
Jn=∫0∞e−axsin2n+1xdx=∏r=1n(a2+(2r+1)2)(2n+1)!
Thus, I=(2n+1)!∫0∞∏r=1n(a2+(2r+1)2)dy
The expression ∏r=1n(a2+(2r+1)2)1=r=0∑ny2+(2r+1)2Ar
where the coefficients can be determined by the cover-up rule as Ak=22n(−1)k(n−k)!(n+k+1)!2k+1
So, I=(2n+1)!k=0∑n∫0∞y2+(2k+1)2Akdy=22n+1πk=0∑n(−1)k(n−k2n+1)=22n+1π(n2n)
Log in to reply
technique looks familiar!
ThatIt is interesting to note that the sum r=0∑m(−1)r(rn)=(−1)m(mn−1)0≤m<n can be proved by considering the coefficients of xm on both sides of the identity (1+x)n−1=(1+x)−1×(1+x)n∣x∣<1.
(+1) Nice ! I have another method so all total we could have atleast 2 real methods to evaluate this.
Problem 22:
Evaluate ∫041πln(1+tanx)dx
This problem has been solved by Rohinth M.Athreya.
Log in to reply
∫04πln(1+tan(4π−x)dx
∫04πln(1+tanx2dx
if we denote the integral by I. it is also 4πln2−I
and so the integral is 8πln2.
Problem 23:
Evaluate the indefinite integral, ∫((1+x2)1012(2+x2)3012)20121xdx.
This problem has been solved by Mark Hennings.
Log in to reply
Start with the substitution y=x2+1. Then Ip(x)=∫(1+x2)1−p(2+x2)1+pxdx=21∫y1−p(1+y)1+pdy=21∫(1+yy)p−1(1+y)2dy Then the substitution z=1+yy gives Ip(x)=21∫zp−1dz=2p1zp+c=2p1(x2+2x2+1)p+c In this case, we have p=20121000=503250, and hence I503250(x)=500503(x2+2x2+1)503250+c
Problem 21:
Evaluate ∫02πln2sinxln2cosxdx
This problem has been solved by Mark Hennings.
Log in to reply
Start with f(a,b)=∫021πsina−1xcosb−1xdx=21B(21a,21b)=2Γ(21a+21b)Γ(21a)Γ(21b) for a,b>0. Then ∂a∂f∂a2∂2f=21f(a,b)[ψ(21a)−ψ(21a+21b)]=41f(a,b)[ψ′(21a)−ψ′(21a+21b)+[ψ(21a)−ψ(21a+21b)]2] and so g(b)=∂a2∂2f(1,b)=81A(b)B(b) where A(b)=Γ(21+21b)Γ(21)Γ(21b)=Γ(21+21b)πΓ(21b)B(b)=ψ′(21)−ψ′(21+21b)+[ψ(21)−ψ(21+21b)]2 Thus A′(b)A′′(b)=21A(b)[ψ(21b)−ψ(21+21b)]=41A(b)[ψ′(21b)−ψ′(21+21b)+[ψ(21b)−ψ(21+21b)]2] and hence A(1)=πA′(1)=−πln2A′′(1)=121π3+π(ln2)2 Also B′(b)B′′(b)=−21ψ′′(21+21b)−[ψ(21)−ψ(21+21b)]ψ′(21+21b)=−41ψ′′′(21+21b)+21ψ′(21+21b)2−21[ψ(21)−ψ(21+21b)]ψ′′(21+21b) and hence B(1)=31π2+4(ln2)2B′(1)=31π2ln2+ζ(3)B′′(1)=−3601π4−2ζ(3)ln2 and hence ∫021πln2(sinx)ln2(cosx)dx=∂a2∂b2∂4f∣∣∣a=b=1=g′′(1)=81[A′′(1)B(1)+2A′(1)B′(1)+A(1)B′′(1)]=3201π5+21π(ln2)4−21πζ(3)ln2
PROBLEM 24:
Show that ∫0∞e−ax2−bx−2dx=21aπe−2ab for any a,b>0.
Log in to reply
Let,
I=∫0∞e−ax2−x2b dx
=21∫−∞∞e−ax2−x2b dx
Sunstituting x↦ax, we have,
I=2a1∫−∞∞e−x2−x2ab dx
=2ae−2ab∫−∞∞e−(x−xab)2 dx
Using Glasser's Master Theorem, we have,
I=2ae−2ab∫−∞∞e−x2 dx
=21aπe−2ab□
Log in to reply
You don't need a master theorem!
Completing the square ∫0∞e−ax2−bx−2dx=e−2ab∫0∞e−(ax−bx−1)2dx Now the substitution y=abx−1 yields ∫0∞e−(ax−bx−1)2dx=∫∞0e−(by−ay)2×(−aby−2)dy=ab∫0∞x−2e−(ax−bx−1)2dy=21∫0∞(1+abx−2)e−(ax−bx−1)2dx=2a1∫0∞(a+bx−2)e−(ax−bx−1)2dx The further substitution u=ax−bx−1 gives ∫0∞e−(ax−bx−1)2dx=2a1∫Re−u2du=21aπ which implies that ∫0∞e−ax2−bx−2dx=21aπe−2ab
Log in to reply
Solve LaTex: \displaysize \large \int _((cos \phi)^-1)^((cos \phi)^-1)\((e^(\delta*x))/(\sqrt (1-(x^2)))dx