Friday, 27 June 2014

calculus - Evaluating inti0nftyfraclog10x1+x3dx




How one would evaluate the following integral?




0log10(x)1+x3dx




I have tried substitution with no success as well as differentiation under integral sign.
Can anyone help me please. I prefer not to use contour integration.


Answer




Caveat. I wrote the following answer not having seen the request
by the OP that he would prefer not to use contour integration. Perhaps
what follows can help make the case for and showcase contour
integration for this integral which belongs to a class that has
frequently appeared here at MSE.



Observe that if we are allowed to use a CAS (which would appear
necessary for this problem) then we may compute



Qn=0lognxx3+1dx=0fn(x)dx



where



fn(z)=lognzz3+1



by computing all Qn recursively by integrating fn+1(z),fn(z), and so on around a keyhole contour with the slot on the
positive real axis and the branch cut of the logarithm on that axis as
well, with argument from 0 to 2π. The poles of fn(z) are at

ρk=exp(πi/3+2πik/3) with k=0,1,2. We obtain for
the residues



Resz=ρkfn(z)=Resz=ρklognzz3+1=lognz3z2|z=ρk=zlognz3z3|z=ρk=13zlognz|z=ρk=13exp(πi/3+2πik/3)(πi/3+2πik/3)n=αn,k.




We obtain by integrating fn(z)



0lognzz3+1dz+0(2πi+logz)nz3+1dz=2πikResz=ρkfn(z)=2πikαn,k.



This yields




\sum_{p=0}^{n-1} {n\choose p} (2\pi i)^{n-p} \int_0^\infty \frac{\log^p z}{z^3+1} \; dz = - 2\pi i \sum_k \alpha_{n,k}



which is



\sum_{p=0}^{n-1} {n\choose p} (2\pi i)^{n-p-1} Q_p = - \sum_k \alpha_{n,k}



or




\sum_{p=0}^{n} {n+1\choose p} (2\pi i)^{n-p} Q_p = - \sum_k \alpha_{n+1,k}



Therefore to compute Q_n we use the recurrence



Q_n = - \frac{1}{n+1} \sum_k \alpha_{n+1,k} - \frac{1}{n+1} \sum_{p=0}^{n-1} {n+1\choose p} (2\pi i)^{n-p} Q_p




We just need the base case Q_0 which we compute using a pizza slice
resting on the positive real axis and having argument 2\pi/3 so that
it only contains \alpha_{0,0}. Parameterizing with z=\exp(2\pi i/3) t we get



Q_0 - \exp(2\pi i/3) Q_0 = 2\pi i \alpha_{0,0}



which yields



Q_0 = - \frac{1}{3} 2\pi i \frac{\exp(\pi i/3)}{1-\exp(2\pi i/3)} = - \frac{1}{3} 2\pi i \frac{1}{\exp(-\pi i/3)-\exp(\pi i/3)} \\ = \frac{1}{3} \pi \frac{1}{\sin(\pi/3)} = \frac{2}{9}\sqrt{3}\pi.



With everything in place we obtain e.g. the sequence up to n=10



-{\frac {2\,{\pi }^{2}}{27}},{\frac {10\,{\pi }^{3} \sqrt {3}}{243}},-{\frac {14\,{\pi }^{4}}{243}},{\frac {34\,{\pi }^{5}\sqrt {3}}{729}},\\-{\frac {806\,{\pi }^{6 }}{6561}},{\frac {910\,{\pi }^{7}\sqrt {3}}{6561}},-{ \frac {10414\,{\pi }^{8}}{19683}},{\frac {415826\,{\pi }^{9}\sqrt {3}}{531441}},\\-{\frac {685762\,{\pi }^{10}}{ 177147}},{\frac {3786350\,{\pi }^{11}\sqrt {3}}{531441}}, \ldots



The Maple code for this is extremely simple, consisting of a few
lines.




alpha := (n,k) ->

-1/3 * exp(Pi*I/3+2*Pi*I*k/3) * (Pi*I/3 + 2*Pi*I*k/3)^n;

Q :=
proc(n)
option remember;
local res;

if n = 0 then return 2/9*sqrt(3)*Pi fi;

res :=

-1/(n+1)*add(alpha(n+1,k), k=0..2)
-1/(n+1)*add(binomial(n+1, p)*(2*Pi*I)^(n-p)*Q(p),
p=0..n-1);

simplify(res);
end;

VERIF := n -> int((log(x))^n/(x^3+1), x=0..infinity);



Observe that this method generalizes quite nicely. Suppose we are
interested in



K_n = \int_0^\infty \frac{\log^n x}{x^3-2x+4} \; dx.



The same computation goes through except now we have the following
three poles and their logarithms



\begin{array}{|l|l|} \hline \text{pole} & \text{logarithm} \\ \hline \rho_0 = 1+i & \log \rho_0 = \frac{1}{2} \log 2 + \frac{1}{4}i\pi \\ \hline \rho_1 = 1-i & \log \rho_1 = \frac{1}{2} \log 2 + \frac{7}{4}i\pi \\ \hline \rho_2 = -2 & \log \rho_2 = \log 2 + i\pi.\\ \hline \end{array}



The rest is unchanged. We obtain e.g.




K_4 = {\frac {357\,{\pi }^{5}}{10240}}-{\frac {31\, \left( \ln \left( 2 \right) \right) ^{5}}{1600}}-{\frac {139\, \left( \ln \left( 2 \right) \right) ^{3}{\pi }^{2}}{1920 }} \\ -{\frac {4897\,\ln \left( 2 \right) {\pi }^{4}}{76800}}+ {\frac {9\, \left( \ln \left( 2 \right) \right) ^{4}\pi }{640}}+{\frac {63\, \left( \ln \left( 2 \right) \right) ^{2}{\pi }^{3}}{1280}}.



The Maple code is very similar to the first version.





alpha_sum :=
proc(n)
local poles;

poles :=
[[1+I, 1/2*log(2) + I*Pi/4],
[1-I, 1/2*log(2) + 7*I*Pi/4],
[-2, log(2) + I*Pi]];


add(residue(1/(x^3-2*x+4), x=p[1])*p[2]^n,
p in poles);
end;

Q :=
proc(n)
option remember;
local res;

if n = 0 then

return
simplify(int(1/(x^3-2*x+4), x=0..infinity));
fi;

res :=
-1/(n+1)*alpha_sum(n+1)
-1/(n+1)*add(binomial(n+1, p)*(2*Pi*I)^(n-p)*Q(p),
p=0..n-1);

simplify(res);

end;

VERIF := n -> int((log(x))^n/(x^3-2*x+4), x=0..infinity);

No comments:

Post a Comment

real analysis - How to find lim_{hrightarrow 0}frac{sin(ha)}{h}

How to find \lim_{h\rightarrow 0}\frac{\sin(ha)}{h} without lhopital rule? I know when I use lhopital I easy get $$ \lim_{h\rightarrow 0}...