Evaluate $int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx$
up vote
6
down vote
favorite
I want to evaluate $$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx$$
First,I tried to evaluate like this:
$$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx=int_{0}^{frac{pi}{2}}x^2left(frac{1+cos x}{sin x}right)frac{dx}{1+cos x}=int_{0}^{frac{pi}{2}}x^2left(frac{1+cos x}{sin x}right)dleft(frac{sin x}{1+cos x}right)$$
$$=int_{0}^{frac{pi}{2}}x^2dlogleft(frac{sin x}{1+cos x}right)=x^2logleft(frac{sin x}{1+cos x}right)|_{0}^{frac{pi}{2}}-2int_{0}^{frac{pi}{2}}xlogleft(frac{sin x}{1+cos x}right)dx$$
$$=0+2int_{0}^{frac{pi}{2}}xlogleft(frac{1+cos x}{sin x}right)dx=2int_{0}^{frac{pi}{2}}xlogleft(1+cos xright)dx-2int_{0}^{frac{pi}{2}}xlogleft(sin xright)dx$$
$$=2int_{0}^{frac{pi}{2}}xlogcot left(frac{x}{2}right)dx=8int_{0}^{frac{pi}{4}}xlogcot xdx$$
but I can't proceed next step,help me,thanks.
calculus integration analysis
|
show 7 more comments
up vote
6
down vote
favorite
I want to evaluate $$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx$$
First,I tried to evaluate like this:
$$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx=int_{0}^{frac{pi}{2}}x^2left(frac{1+cos x}{sin x}right)frac{dx}{1+cos x}=int_{0}^{frac{pi}{2}}x^2left(frac{1+cos x}{sin x}right)dleft(frac{sin x}{1+cos x}right)$$
$$=int_{0}^{frac{pi}{2}}x^2dlogleft(frac{sin x}{1+cos x}right)=x^2logleft(frac{sin x}{1+cos x}right)|_{0}^{frac{pi}{2}}-2int_{0}^{frac{pi}{2}}xlogleft(frac{sin x}{1+cos x}right)dx$$
$$=0+2int_{0}^{frac{pi}{2}}xlogleft(frac{1+cos x}{sin x}right)dx=2int_{0}^{frac{pi}{2}}xlogleft(1+cos xright)dx-2int_{0}^{frac{pi}{2}}xlogleft(sin xright)dx$$
$$=2int_{0}^{frac{pi}{2}}xlogcot left(frac{x}{2}right)dx=8int_{0}^{frac{pi}{4}}xlogcot xdx$$
but I can't proceed next step,help me,thanks.
calculus integration analysis
you mean the first step? Give some thoughts, please. (Although to be fair, it appears we have a doozy wolframalpha.com/input/… ... I'm guessing this means you should try a series approach.)
– spaceisdarkgreen
Mar 30 at 3:40
@FofX Do you want an exact answer, or just a numerical approximation?
– Toby Mak
Mar 30 at 3:41
@spaceisdarkgreen I think use integration by parts?
– FofX
Mar 30 at 3:42
@TobyMak I want an exact answer,hh,thank you.
– FofX
Mar 30 at 3:44
@FofX Don't think so, though I can't say I know it won't simplify things... whatever it is the indefinite integral doesn't seem to come out nice (see my wolfram alpha link). You can use the Taylor series for csc to turn it into an infinite sum that I don't find particularly inviting, but has some features that make the wolfram alpha answer involving the zeta function and Catalan constant plausible, like bernoulli numbers
– spaceisdarkgreen
Mar 30 at 3:56
|
show 7 more comments
up vote
6
down vote
favorite
up vote
6
down vote
favorite
I want to evaluate $$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx$$
First,I tried to evaluate like this:
$$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx=int_{0}^{frac{pi}{2}}x^2left(frac{1+cos x}{sin x}right)frac{dx}{1+cos x}=int_{0}^{frac{pi}{2}}x^2left(frac{1+cos x}{sin x}right)dleft(frac{sin x}{1+cos x}right)$$
$$=int_{0}^{frac{pi}{2}}x^2dlogleft(frac{sin x}{1+cos x}right)=x^2logleft(frac{sin x}{1+cos x}right)|_{0}^{frac{pi}{2}}-2int_{0}^{frac{pi}{2}}xlogleft(frac{sin x}{1+cos x}right)dx$$
$$=0+2int_{0}^{frac{pi}{2}}xlogleft(frac{1+cos x}{sin x}right)dx=2int_{0}^{frac{pi}{2}}xlogleft(1+cos xright)dx-2int_{0}^{frac{pi}{2}}xlogleft(sin xright)dx$$
$$=2int_{0}^{frac{pi}{2}}xlogcot left(frac{x}{2}right)dx=8int_{0}^{frac{pi}{4}}xlogcot xdx$$
but I can't proceed next step,help me,thanks.
calculus integration analysis
I want to evaluate $$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx$$
First,I tried to evaluate like this:
$$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx=int_{0}^{frac{pi}{2}}x^2left(frac{1+cos x}{sin x}right)frac{dx}{1+cos x}=int_{0}^{frac{pi}{2}}x^2left(frac{1+cos x}{sin x}right)dleft(frac{sin x}{1+cos x}right)$$
$$=int_{0}^{frac{pi}{2}}x^2dlogleft(frac{sin x}{1+cos x}right)=x^2logleft(frac{sin x}{1+cos x}right)|_{0}^{frac{pi}{2}}-2int_{0}^{frac{pi}{2}}xlogleft(frac{sin x}{1+cos x}right)dx$$
$$=0+2int_{0}^{frac{pi}{2}}xlogleft(frac{1+cos x}{sin x}right)dx=2int_{0}^{frac{pi}{2}}xlogleft(1+cos xright)dx-2int_{0}^{frac{pi}{2}}xlogleft(sin xright)dx$$
$$=2int_{0}^{frac{pi}{2}}xlogcot left(frac{x}{2}right)dx=8int_{0}^{frac{pi}{4}}xlogcot xdx$$
but I can't proceed next step,help me,thanks.
calculus integration analysis
calculus integration analysis
edited Mar 30 at 6:49
asked Mar 30 at 3:36
FofX
1706
1706
you mean the first step? Give some thoughts, please. (Although to be fair, it appears we have a doozy wolframalpha.com/input/… ... I'm guessing this means you should try a series approach.)
– spaceisdarkgreen
Mar 30 at 3:40
@FofX Do you want an exact answer, or just a numerical approximation?
– Toby Mak
Mar 30 at 3:41
@spaceisdarkgreen I think use integration by parts?
– FofX
Mar 30 at 3:42
@TobyMak I want an exact answer,hh,thank you.
– FofX
Mar 30 at 3:44
@FofX Don't think so, though I can't say I know it won't simplify things... whatever it is the indefinite integral doesn't seem to come out nice (see my wolfram alpha link). You can use the Taylor series for csc to turn it into an infinite sum that I don't find particularly inviting, but has some features that make the wolfram alpha answer involving the zeta function and Catalan constant plausible, like bernoulli numbers
– spaceisdarkgreen
Mar 30 at 3:56
|
show 7 more comments
you mean the first step? Give some thoughts, please. (Although to be fair, it appears we have a doozy wolframalpha.com/input/… ... I'm guessing this means you should try a series approach.)
– spaceisdarkgreen
Mar 30 at 3:40
@FofX Do you want an exact answer, or just a numerical approximation?
– Toby Mak
Mar 30 at 3:41
@spaceisdarkgreen I think use integration by parts?
– FofX
Mar 30 at 3:42
@TobyMak I want an exact answer,hh,thank you.
– FofX
Mar 30 at 3:44
@FofX Don't think so, though I can't say I know it won't simplify things... whatever it is the indefinite integral doesn't seem to come out nice (see my wolfram alpha link). You can use the Taylor series for csc to turn it into an infinite sum that I don't find particularly inviting, but has some features that make the wolfram alpha answer involving the zeta function and Catalan constant plausible, like bernoulli numbers
– spaceisdarkgreen
Mar 30 at 3:56
you mean the first step? Give some thoughts, please. (Although to be fair, it appears we have a doozy wolframalpha.com/input/… ... I'm guessing this means you should try a series approach.)
– spaceisdarkgreen
Mar 30 at 3:40
you mean the first step? Give some thoughts, please. (Although to be fair, it appears we have a doozy wolframalpha.com/input/… ... I'm guessing this means you should try a series approach.)
– spaceisdarkgreen
Mar 30 at 3:40
@FofX Do you want an exact answer, or just a numerical approximation?
– Toby Mak
Mar 30 at 3:41
@FofX Do you want an exact answer, or just a numerical approximation?
– Toby Mak
Mar 30 at 3:41
@spaceisdarkgreen I think use integration by parts?
– FofX
Mar 30 at 3:42
@spaceisdarkgreen I think use integration by parts?
– FofX
Mar 30 at 3:42
@TobyMak I want an exact answer,hh,thank you.
– FofX
Mar 30 at 3:44
@TobyMak I want an exact answer,hh,thank you.
– FofX
Mar 30 at 3:44
@FofX Don't think so, though I can't say I know it won't simplify things... whatever it is the indefinite integral doesn't seem to come out nice (see my wolfram alpha link). You can use the Taylor series for csc to turn it into an infinite sum that I don't find particularly inviting, but has some features that make the wolfram alpha answer involving the zeta function and Catalan constant plausible, like bernoulli numbers
– spaceisdarkgreen
Mar 30 at 3:56
@FofX Don't think so, though I can't say I know it won't simplify things... whatever it is the indefinite integral doesn't seem to come out nice (see my wolfram alpha link). You can use the Taylor series for csc to turn it into an infinite sum that I don't find particularly inviting, but has some features that make the wolfram alpha answer involving the zeta function and Catalan constant plausible, like bernoulli numbers
– spaceisdarkgreen
Mar 30 at 3:56
|
show 7 more comments
7 Answers
7
active
oldest
votes
up vote
3
down vote
accepted
At the price of special functions, the antiderivative could be computed
$$I=intfrac{x^2}{ sin x},dx=-4 i x text{Li}_2left(e^{i x}right)+i x text{Li}_2left(e^{2 i x}right)+4
text{Li}_3left(e^{i x}right)-frac{1}{2} text{Li}_3left(e^{2 i x}right)-2
x^2 tanh ^{-1}left(e^{i x}right)$$ where appear the polylogarithm functions.
$$lim_{xto frac{pi }{2}} , I=2 pi Cqquad text{and} qquadlim_{xto 0} , I=frac{7 }{2}zeta (3)implies int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx=2 pi C-frac{7 }{2}zeta (3)$$ as given by Wolfram Alpha. This evaluates a $approx 1.54798$.
For a fast approximation, we could use the superb approximation $$sin(x) simeq frac{16 (pi -x) x}{5 pi ^2-4 (pi -x) x}qquad (0leq xleqpi)$$ which was proposed by Mahabhaskariya of Bhaskara I, a seventh-century Indian mathematician (have a look here).
This would make
$$I approx J= - int left(frac{x^2}{4}+frac{5 pi ^3}{16 (x-pi )}+frac{5 pi ^2}{16} right),dx=-frac{x^3}{12}-frac{5 pi ^2 x}{16}-frac{5}{16} pi ^3 log (pi -x)+frac{19
pi ^3}{48}$$
$$lim_{xto frac{pi }{2}} , J=frac{pi ^3}{48} left(11-15 log left(frac{pi }{2}right)right)qquad text{and} qquadlim_{xto 0} , J
=frac{pi ^3}{48} (19-15 log (pi ))$$ leading to the approximation
$$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dxapprox frac{pi ^3}{48} (15 log (2)-8)approx 1.54851$$ which is not too bad.
Tha advantage of such approximation is that it allows a fast evaluation of
$$K(t)=int_{0}^{t}frac{x^2}{ sin x}dx$$ The table below compares the approximation to the exact result
$$left(
begin{array}{ccc}
t & text{approximation} & text{exact} \
frac{pi }{20} & 0.01221 & 0.01236 \
frac{pi }{10} & 0.04936 & 0.04976 \
frac{3 pi }{20} & 0.11258 & 0.11312 \
frac{pi }{5} & 0.20358 & 0.20409 \
frac{pi }{4} & 0.32475 & 0.32508 \
frac{3 pi }{10} & 0.47939 & 0.47945 \
frac{7 pi }{20} & 0.67196 & 0.67176 \
frac{2 pi }{5} & 0.90847 & 0.90807 \
frac{9 pi }{20} & 1.19701 & 1.19650 \
frac{pi }{2} & 1.54851 & 1.54798 \
frac{11 pi }{20} & 1.97802 & 1.97746 \
frac{3 pi }{5} & 2.50657 & 2.50583 \
frac{13 pi }{20} & 3.16447 & 3.16315 \
frac{7 pi }{10} & 3.99696 & 3.99445 \
frac{3 pi }{4} & 5.07529 & 5.07091 \
frac{4 pi }{5} & 6.52008 & 6.51359 \
frac{17 pi }{20} & 8.55922 & 8.55230 \
frac{9 pi }{10} & 11.7067 & 11.7077 \
frac{19 pi }{20} & 17.6067 & 17.6510
end{array}
right)$$
Thank you for your link ,thats very interesting!
– FofX
Mar 30 at 6:17
1
One may also accelerate the series $$ sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2} $$ to derive accurate numerical approximations of $I$.
– Jack D'Aurizio
Mar 30 at 14:30
add a comment |
up vote
3
down vote
$$mathcal{J}=int_{0}^{pi/2}frac{x^2}{sin x},dx = int_{0}^{1}frac{arcsin^2(x)}{xsqrt{1-x^2}},dx=sum_{ngeq 1}frac{2^{2n-1}}{n^2binom{2n}{n}}int_{0}^{1}frac{x^{2n-1}}{sqrt{1-x^2}},dx tag{1}$$
by the Maclaurin series of $arcsin^2(x)$. Euler's Beta function then leads to
$$ mathcal{J}=sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2}=phantom{}_4 F_3left(1,1,1,1;tfrac{3}{2},tfrac{3}{2},2;1right)tag{2} $$
where the RHS is a manageable hypergeometric function (similar objects are evaluated both here and here) and as already shown by Claude Leibovici, $mathcal{J}=4int_{0}^{1}frac{arctan^2(u)}{u},du $ is simply given by a combination of a dilogarithm and a trilogarithm. Indeed
$$ int_{0}^{pi/2}int_{0}^{theta}frac{u}{sin u},du,dtheta =-pi G+frac{7}{2}zeta(3)tag{3}$$
leading to $mathcal{J}=2pi G-frac{7}{2}zeta(3)$, has already been a key lemma in this historical thread.
An alternative way for proving this identity is just to write $frac{x}{sin x}$ and $|x|$ as Fourier cosine series.
The Shafer-Fink inequality leads to
$$ int_{0}^{pi/2}frac{x^2}{sin x},dx = 4 int_{0}^{1}frac{arctan^2(u)}{u},du approx frac{6}{7}(3sqrt{2}-5)+9logleft(frac{2sqrt{2}+1}{3}right)approx 1.54.tag{4}$$
@ Wow, thank you for your detailed play. I learned a lot.
– FofX
Mar 31 at 3:55
add a comment |
up vote
3
down vote
$begin{align}
J&=int_{0}^{pi/2}frac{x^2}{sin x},dx&
end{align}$
Perform the change of variable,
$displaystyle y=tanleft(frac{x}{2}right)$,
$begin{align}
J&=4int_0^1 frac{arctan^2 x}{x},dx\
&=4Big[ln xarctan ^2 xBig]_0^1-8int_0^1 frac{arctan xln x}{1+x^2},dx\
&=-8int_0^1 frac{arctan xln x}{1+x^2},dx\
end{align}$
For $xin [0;1]$, define $F$,
$begin{align} F(x)&=int_0^x frac{ln t}{1+t^2},dt\
&=int_0^1 frac{xln(xt)}{1+x^2t^2},dt
end{align}$
Observe that,
$displaystyle F(0)=0$ and, $displaystyle F(1)=-text{G}$.
$text{G}$ is the Catalan constant.
$begin{align}J&=-8Big[F(x)arctan xBig]_0^1+8int_0^1 int_0^1 frac{xln(tx)}{(1+t^2x^2)(1+x^2)},dt,dx\
&=2Gpi+8int_0^1 int_0^1 frac{xln x}{(1+t^2x^2)(1+x^2)},dt,dx+8int_0^1 int_0^1 frac{xln t}{(1+t^2x^2)(1+x^2)},dt,dx\
&=2Gpi+8int_0^1 Big[frac{arctan(tx)ln x}{1+x^2}Big]_{t=0}^{t=1},dx+4int_0^1 Big[frac{(ln(1+t^2x^2)-ln(1+x^2))ln t}{t^2-1}Big]_{x=0}^{x=1},dt\
&=2Gpi+8int_0^1 frac{arctan xln x}{1+x^2},dx+4int_0^1 frac{(ln(1+t^2)-ln 2)ln t }{t^2-1},dt\
&=2Gpi-J+4int_0^1 frac{(ln 2-ln(1+t^2))ln t }{1-t^2},dt\
end{align}$
Therefore,
$displaystyle J=text{G}pi+2int_0^1 frac{(ln 2-ln(1+x^2))ln x }{1-x^2},dx$
For $xin[0;1]$, define,
$begin{align}H(x)&=int_0^x frac{ln t}{1-t^2},dt\
&=int_0^1 frac{xln(tx)}{1-t^2x^2},dt\
end{align}$
Observe that,
$displaystyle H(0)=0$ and $displaystyle H(1)=-frac{pi^2}{8}$.
$begin{align}J&=text{G}pi+2Big[(ln 2-ln(1+x^2))H(x)Big]_0^1+4int_0^1int_0^1frac{x^2ln(tx)}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1int_0^1frac{x^2ln(tx)}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1int_0^1frac{x^2ln t}{(1+x^2)(1-t^2x^2)},dt,dx+4int_0^1int_0^1frac{x^2ln x}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1Big[frac{ln t}{1+t^2}left(frac{ln(1+tx)}{2t}-frac{ln(1-tx)}{2t}-arctan xright)Big]_{x=0}^{x=1},dt+\
&2int_0^1 Big[frac{xln x}{1+x^2}lnleft(frac{1+tx}{1-tx}right)Big]_{t=0}^{t=1},dx\
&=text{G}pi+2int_0^1 frac{ln t}{t(1+t^2)}lnleft(frac{1+t}{1-t}right),dt-piint_0^1 frac{ln t}{1+t^2},dt+2int_0^1 frac{xln x}{1+x^2}lnleft(frac{1+x}{1-x}right),dx\
&=2text{G}pi+2int_0^1 frac{ln x}{x}lnleft(frac{1+x}{1-x}right),dx\
end{align}$
But, for $0leq x<1$,
$displaystyle frac{1}{x}lnleft(frac{1+x}{1-x}right)=2sum_{n=0}^{infty}frac{x^{2n}}{2n+1}$
Therefore,
$begin{align}int_0^1 frac{ln x}{x}lnleft(frac{1+x}{1-x}right),dx&=2int_0^1 left(sum_{n=0}^{infty}frac{x^{2n}}{2n+1}right)ln x,dx\
&=2 sum_{n=0}^{infty}int_0^1 frac{x^{2n}ln x}{2n+1},dx\
&=-2sum_{n=0}^{infty}frac{1}{(2n+1)^3}\
&=-2left(sum_{n=1}^{infty} frac{1}{n^3}-sum_{n=1}^{infty} frac{1}{(2n)^3}right)\
&=-2left(zeta(3)-frac{1}{8}zeta(3)right)\
&=-frac{7}{4}zeta(3)\
end{align}$
Therefore,
$ boxed{J=2text{G}pi-frac{7}{2}zeta(3)}$
add a comment |
up vote
3
down vote
Observe we have
begin{align}
I=int^{pi/2}_0 frac{x^2}{sin x} dx = int^{pi/2}_0 frac{x^2}{cosleft(frac{pi}{2}-x right)} dx = int^{pi/2}_0 frac{(frac{pi}{2}-u)^2}{cos u} du.
end{align}
Then using integration by parts, we see that
begin{align}
I&=left(frac{pi}{2}-uright)^2left{logleft|1 + sin uright|-log|cos u|right}bigg|^{pi/2}_0 + 2int^{pi/2}_0left(frac{pi}{2}-u right)log|sec u + tan u| du\
&= 2pi left(frac{1}{2}int_{0}^{pi/2}log|sec u+tan u| du right)-frac{7}{2}left(frac{4}{7}int^{pi/2}_0 u log|sec u+tan u| du right)\
&= 2pi G - frac{7}{2}zeta(3).
end{align}
Here, I have used the facts that
begin{align}
G= frac{1}{2}int_{0}^{pi/2}log|sec u+tan u| du
end{align}
and
begin{align}
zeta(3) = frac{4}{7}int^{pi/2}_0 u log|sec u+tan u| du.
end{align}
See here for reference.
Thanks for your link,it is helpful for me.
– FofX
Mar 30 at 6:08
add a comment |
up vote
2
down vote
$newcommand{bbx}[1]{,bbox[15px,border:1px groove navy]{displaystyle{#1}},}
newcommand{braces}[1]{leftlbrace,{#1},rightrbrace}
newcommand{bracks}[1]{leftlbrack,{#1},rightrbrack}
newcommand{dd}{mathrm{d}}
newcommand{ds}[1]{displaystyle{#1}}
newcommand{expo}[1]{,mathrm{e}^{#1},}
newcommand{ic}{mathrm{i}}
newcommand{mc}[1]{mathcal{#1}}
newcommand{mrm}[1]{mathrm{#1}}
newcommand{pars}[1]{left(,{#1},right)}
newcommand{partiald}[3]{frac{partial^{#1} #2}{partial #3^{#1}}}
newcommand{root}[2]{,sqrt[#1]{,{#2},},}
newcommand{totald}[3]{frac{mathrm{d}^{#1} #2}{mathrm{d} #3^{#1}}}
newcommand{verts}[1]{leftvert,{#1},rightvert}$
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
left.Reint_{x = 0}^{x = pi/2}{bracks{-iclnpars{z}}^{2} over
pars{z - 1/z}/pars{2ic}},{dd z over ic z}
,rightvert_{ z = exppars{ic x}}
\[5mm] & =
left.2,Reint_{x = 0}^{x = pi/2}{ln^{2}pars{z} over 1 - z^{2}},dd z
,rightvert_{ z = exppars{ic x}}
end{align}
$ds{ln}$ is the $ds{log}$-principal branch. Integration of $ds{{ln^{2}pars{z} over 1 - z^{2}}}$ along the path
$ds{C_{x}cup C_{R}cup C_{y}}$ vanishes out such that
$ds{int_{large C_{R}}{ln^{2}pars{z} over 1 - z^{2}},dd z =
-int_{large C_{y}}{ln^{2}pars{z} over 1 - z^{2}},dd z -
int_{large C_{x}}{ln^{2}pars{z} over 1 - z^{2}},dd z}$
Then,
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
overbrace{-2,Reint_{1}^{0}{bracks{lnpars{y} + piic/2}^{, 2} over
1 + y^{2}},ic,dd y}^{ds{mbox{along} C_{y}}} -
overbrace{2,Reint_{0}^{1}{ln^{2}pars{x} over 1 - x^{2}},dd x}
^{ds{mbox{along} C_{x}}}
\[5mm] & =
-2pi,int_{0}^{1}{lnpars{y} over 1 + y^{2}},dd y -
2int_{0}^{1}{ln^{2}pars{x} over 1 - x^{2}},dd x
end{align}
However, $ds{int_{0}^{1}{lnpars{y} over 1 + y^{2}},dd y = -G}$ where
$ds{G}$ is the Catalan Constant such that
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
2pi G - 2sum_{n = 0}^{infty}
overbrace{int_{0}^{1}ln^{2}pars{x}x^{2n},dd x}
^{ds{2 over pars{2n + 1}^{3}}}
\[5mm] & =
2pi G - 4bracks{sum_{n = 1}^{infty}{1 over n^{3}} -
sum_{n = 1}^{infty}{1 over pars{2n}^{3}}} =
2pi G - {7 over 2}sum_{n = 1}^{infty}{1 over n^{3}}
\[5mm] & = bbx{2pi G - {7 over 2},zetapars{3}} approx 1.5480
end{align}
add a comment |
up vote
1
down vote
As pointed out within the other answers we want to prove that
$$mathfrak{I}=int_0^{pi/2}frac{x^2}{sin x}~dx=2pi G-frac72zeta(3)$$
As the OP showed $mathfrak{I}$ can be reduced to a linear combination of $x$ and the function $log(cot x)$
$$mathfrak{I}=int_0^{pi/2}frac{x^2}{sin x}~dx=8int_0^{pi/4}xlog(cot x)~dx$$
By applying the definition of the cotangent function followed up by the usage of the well-known Fourier series expansions of $log(cos x)$ and $log(sin x)$ this can be further simplified. Therefore we get
$$smallbegin{align}
mathfrak{I}=8int_0^{pi/4}xlog(cot x)~dx&=8left[int_0^{pi/4}xlog(cos x)~dx-int_0^{pi/4}xlog(sin x)~dxright]\
&=8left[int_0^{pi/4}xleft(-log(2)-sum_{n=1}^{infty}(-1)^nfrac{cos(2nx)}{n}right)~dx-int_0^{pi/4}xleft(-log(2)-sum_{n=1}^{infty}frac{cos(2nx)}{n}right)~dxright]\
&=8left[-sum_{n=1}^{infty}frac{(-1)^n}{n}underbrace{int_0^{pi/4}xcos(2nx)~dx}_I+sum_{n=1}^{infty}frac{1}{n}underbrace{int_0^{pi/4}xcos(2nx)~dx}_Iright]\
end{align}$$
The inner integral $I$ can be easily evaluated using IBP which leads to
$$I=int_0^{pi/4}xcos(2nx)~dx=frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}$$
Plugging this into our original formula and followed by a little bit of algebraic manipulation we get
$$smallbegin{align}
mathfrak{I}&=8left[-sum_{n=1}^{infty}frac{(-1)^n}{n}left(frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}right)+sum_{n=1}^{infty}frac{1}{n}left(frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}right)right]\
&=2left[sum_{n=1}^{infty}frac{(-1)^n}{n^3}-sum_{n=1}^{infty}frac{1}{n^3}right]+pileft[sum_{n=1}^{infty}frac{1}{n^2}sinleft(nfrac{pi}2right)-sum_{n=1}^{infty}frac{(-1)^n}{n^2}sinleft(nfrac{pi}2right)right]
end{align}$$
The first terms can be evaluated in terms of the Riemann Zeta Function $zeta(s)$ and the Dirichlet Eta Function $eta(s)$ whereas for the second term we have to consider some basic properties of the Sine function. For $ninmathbb{N}>0$ the function $sinleft(nfrac{pi}2right)$ will be zero for all even $n$ and $-1$ and $1$ respectively for odd $n$ starting with $sinleft(frac{pi}2right)=1$ for $n=1$. Therefore all even terms vanish while the odd ones will remain with a oscillating negative sign. This leads to
$$smallbegin{align}
mathfrak{I}&=2left[sum_{n=1}^{infty}frac{(-1)^n}{n^3}-sum_{n=1}^{infty}frac{1}{n^3}right]+pileft[sum_{n=1}^{infty}frac{1}{n^2}sinleft(nfrac{pi}2right)-sum_{n=1}^{infty}frac{(-1)^n}{n^2}sinleft(nfrac{pi}2right)right]\
&=2[-eta(3)-zeta(3)]+pileft[sum_{n=0}^{infty}frac{1}{(2n+1)^2}(-1)^n-sum_{n=0}^{infty}frac{(-1)^{2n+1}}{(2n+1)^2}(-1)^nright]\
&=-2[(1-2^{-2})zeta(3)+zeta(3)]+2pisum_{n=0}^{infty}frac{(-1)^n}{(2n+1)^2}\
Leftrightarrowmathfrak{I}&=-frac72zeta(3)+2pi G
end{align}$$
where within the last step the functional relation between the Riemann Zeta Function and the Dirichlet Eta Function aswell as the series defintions of Catalan's Constant $G$ where used.
add a comment |
up vote
1
down vote
We can adapt the formula derived in $(2)$ of this answer:
$$
log(2cos(x/2))=sum_{k=1}^inftyfrac{(-1)^{k-1}}kcos(kx)tag{1a}
$$
Substituting $xmapstopi-x$ in $text{(1a)}$, we get
$$
log(2sin(x/2))=sum_{k=1}^inftyfrac{-1}kcos(kx)tag{1b}
$$
Subtracting $text{(1a)}$ from $text{(1b)}$, the even terms cancel and we get
$$
bbox[5px,border:2px solid #C0A000]{log(tan(x/2))=sum_{k=0}^inftyfrac{-2}{2k+1}cos((2k+1)x)}tag2
$$
Therefore,
$$
begin{align}
int_0^{pi/2}frac{x^2}{sin(x)},mathrm{d}x
&=int_0^{pi/2}x^2,mathrm{d}log(tan(x/2))tag3\
&=-2int_0^{pi/2}xlog(tan(x/2)),mathrm{d}xtag4\
&=sum_{k=0}^inftyfrac4{2k+1}int_0^{pi/2}xcos((2k+1)x),mathrm{d}xtag5\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}int_0^{pi/2}x,mathrm{d}sin((2k+1)x)tag6\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}left[xsin((2k+1)x)+frac{cos((2k+1)x)}{2k+1}right]_0^{pi/2}tag7\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}left[fracpi2(-1)^k-frac1{2k+1}right]tag8\
&=bbox[5px,border:2px solid #C0A000]{2pimathrm{G}-frac72zeta(3)}tag9
end{align}
$$
Explanation:
$(3)$: prepare to integrate by parts
$(4)$: integrate by parts
$(5)$: apply $(2)$
$(6)$: prepare to integrate by parts
$(7)$: integrate by parts
$(8)$: apply the limits of integration
$(9)$: evaluate, where $mathrm{G}$ is Catalan's Constant
add a comment |
7 Answers
7
active
oldest
votes
7 Answers
7
active
oldest
votes
active
oldest
votes
active
oldest
votes
up vote
3
down vote
accepted
At the price of special functions, the antiderivative could be computed
$$I=intfrac{x^2}{ sin x},dx=-4 i x text{Li}_2left(e^{i x}right)+i x text{Li}_2left(e^{2 i x}right)+4
text{Li}_3left(e^{i x}right)-frac{1}{2} text{Li}_3left(e^{2 i x}right)-2
x^2 tanh ^{-1}left(e^{i x}right)$$ where appear the polylogarithm functions.
$$lim_{xto frac{pi }{2}} , I=2 pi Cqquad text{and} qquadlim_{xto 0} , I=frac{7 }{2}zeta (3)implies int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx=2 pi C-frac{7 }{2}zeta (3)$$ as given by Wolfram Alpha. This evaluates a $approx 1.54798$.
For a fast approximation, we could use the superb approximation $$sin(x) simeq frac{16 (pi -x) x}{5 pi ^2-4 (pi -x) x}qquad (0leq xleqpi)$$ which was proposed by Mahabhaskariya of Bhaskara I, a seventh-century Indian mathematician (have a look here).
This would make
$$I approx J= - int left(frac{x^2}{4}+frac{5 pi ^3}{16 (x-pi )}+frac{5 pi ^2}{16} right),dx=-frac{x^3}{12}-frac{5 pi ^2 x}{16}-frac{5}{16} pi ^3 log (pi -x)+frac{19
pi ^3}{48}$$
$$lim_{xto frac{pi }{2}} , J=frac{pi ^3}{48} left(11-15 log left(frac{pi }{2}right)right)qquad text{and} qquadlim_{xto 0} , J
=frac{pi ^3}{48} (19-15 log (pi ))$$ leading to the approximation
$$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dxapprox frac{pi ^3}{48} (15 log (2)-8)approx 1.54851$$ which is not too bad.
Tha advantage of such approximation is that it allows a fast evaluation of
$$K(t)=int_{0}^{t}frac{x^2}{ sin x}dx$$ The table below compares the approximation to the exact result
$$left(
begin{array}{ccc}
t & text{approximation} & text{exact} \
frac{pi }{20} & 0.01221 & 0.01236 \
frac{pi }{10} & 0.04936 & 0.04976 \
frac{3 pi }{20} & 0.11258 & 0.11312 \
frac{pi }{5} & 0.20358 & 0.20409 \
frac{pi }{4} & 0.32475 & 0.32508 \
frac{3 pi }{10} & 0.47939 & 0.47945 \
frac{7 pi }{20} & 0.67196 & 0.67176 \
frac{2 pi }{5} & 0.90847 & 0.90807 \
frac{9 pi }{20} & 1.19701 & 1.19650 \
frac{pi }{2} & 1.54851 & 1.54798 \
frac{11 pi }{20} & 1.97802 & 1.97746 \
frac{3 pi }{5} & 2.50657 & 2.50583 \
frac{13 pi }{20} & 3.16447 & 3.16315 \
frac{7 pi }{10} & 3.99696 & 3.99445 \
frac{3 pi }{4} & 5.07529 & 5.07091 \
frac{4 pi }{5} & 6.52008 & 6.51359 \
frac{17 pi }{20} & 8.55922 & 8.55230 \
frac{9 pi }{10} & 11.7067 & 11.7077 \
frac{19 pi }{20} & 17.6067 & 17.6510
end{array}
right)$$
Thank you for your link ,thats very interesting!
– FofX
Mar 30 at 6:17
1
One may also accelerate the series $$ sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2} $$ to derive accurate numerical approximations of $I$.
– Jack D'Aurizio
Mar 30 at 14:30
add a comment |
up vote
3
down vote
accepted
At the price of special functions, the antiderivative could be computed
$$I=intfrac{x^2}{ sin x},dx=-4 i x text{Li}_2left(e^{i x}right)+i x text{Li}_2left(e^{2 i x}right)+4
text{Li}_3left(e^{i x}right)-frac{1}{2} text{Li}_3left(e^{2 i x}right)-2
x^2 tanh ^{-1}left(e^{i x}right)$$ where appear the polylogarithm functions.
$$lim_{xto frac{pi }{2}} , I=2 pi Cqquad text{and} qquadlim_{xto 0} , I=frac{7 }{2}zeta (3)implies int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx=2 pi C-frac{7 }{2}zeta (3)$$ as given by Wolfram Alpha. This evaluates a $approx 1.54798$.
For a fast approximation, we could use the superb approximation $$sin(x) simeq frac{16 (pi -x) x}{5 pi ^2-4 (pi -x) x}qquad (0leq xleqpi)$$ which was proposed by Mahabhaskariya of Bhaskara I, a seventh-century Indian mathematician (have a look here).
This would make
$$I approx J= - int left(frac{x^2}{4}+frac{5 pi ^3}{16 (x-pi )}+frac{5 pi ^2}{16} right),dx=-frac{x^3}{12}-frac{5 pi ^2 x}{16}-frac{5}{16} pi ^3 log (pi -x)+frac{19
pi ^3}{48}$$
$$lim_{xto frac{pi }{2}} , J=frac{pi ^3}{48} left(11-15 log left(frac{pi }{2}right)right)qquad text{and} qquadlim_{xto 0} , J
=frac{pi ^3}{48} (19-15 log (pi ))$$ leading to the approximation
$$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dxapprox frac{pi ^3}{48} (15 log (2)-8)approx 1.54851$$ which is not too bad.
Tha advantage of such approximation is that it allows a fast evaluation of
$$K(t)=int_{0}^{t}frac{x^2}{ sin x}dx$$ The table below compares the approximation to the exact result
$$left(
begin{array}{ccc}
t & text{approximation} & text{exact} \
frac{pi }{20} & 0.01221 & 0.01236 \
frac{pi }{10} & 0.04936 & 0.04976 \
frac{3 pi }{20} & 0.11258 & 0.11312 \
frac{pi }{5} & 0.20358 & 0.20409 \
frac{pi }{4} & 0.32475 & 0.32508 \
frac{3 pi }{10} & 0.47939 & 0.47945 \
frac{7 pi }{20} & 0.67196 & 0.67176 \
frac{2 pi }{5} & 0.90847 & 0.90807 \
frac{9 pi }{20} & 1.19701 & 1.19650 \
frac{pi }{2} & 1.54851 & 1.54798 \
frac{11 pi }{20} & 1.97802 & 1.97746 \
frac{3 pi }{5} & 2.50657 & 2.50583 \
frac{13 pi }{20} & 3.16447 & 3.16315 \
frac{7 pi }{10} & 3.99696 & 3.99445 \
frac{3 pi }{4} & 5.07529 & 5.07091 \
frac{4 pi }{5} & 6.52008 & 6.51359 \
frac{17 pi }{20} & 8.55922 & 8.55230 \
frac{9 pi }{10} & 11.7067 & 11.7077 \
frac{19 pi }{20} & 17.6067 & 17.6510
end{array}
right)$$
Thank you for your link ,thats very interesting!
– FofX
Mar 30 at 6:17
1
One may also accelerate the series $$ sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2} $$ to derive accurate numerical approximations of $I$.
– Jack D'Aurizio
Mar 30 at 14:30
add a comment |
up vote
3
down vote
accepted
up vote
3
down vote
accepted
At the price of special functions, the antiderivative could be computed
$$I=intfrac{x^2}{ sin x},dx=-4 i x text{Li}_2left(e^{i x}right)+i x text{Li}_2left(e^{2 i x}right)+4
text{Li}_3left(e^{i x}right)-frac{1}{2} text{Li}_3left(e^{2 i x}right)-2
x^2 tanh ^{-1}left(e^{i x}right)$$ where appear the polylogarithm functions.
$$lim_{xto frac{pi }{2}} , I=2 pi Cqquad text{and} qquadlim_{xto 0} , I=frac{7 }{2}zeta (3)implies int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx=2 pi C-frac{7 }{2}zeta (3)$$ as given by Wolfram Alpha. This evaluates a $approx 1.54798$.
For a fast approximation, we could use the superb approximation $$sin(x) simeq frac{16 (pi -x) x}{5 pi ^2-4 (pi -x) x}qquad (0leq xleqpi)$$ which was proposed by Mahabhaskariya of Bhaskara I, a seventh-century Indian mathematician (have a look here).
This would make
$$I approx J= - int left(frac{x^2}{4}+frac{5 pi ^3}{16 (x-pi )}+frac{5 pi ^2}{16} right),dx=-frac{x^3}{12}-frac{5 pi ^2 x}{16}-frac{5}{16} pi ^3 log (pi -x)+frac{19
pi ^3}{48}$$
$$lim_{xto frac{pi }{2}} , J=frac{pi ^3}{48} left(11-15 log left(frac{pi }{2}right)right)qquad text{and} qquadlim_{xto 0} , J
=frac{pi ^3}{48} (19-15 log (pi ))$$ leading to the approximation
$$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dxapprox frac{pi ^3}{48} (15 log (2)-8)approx 1.54851$$ which is not too bad.
Tha advantage of such approximation is that it allows a fast evaluation of
$$K(t)=int_{0}^{t}frac{x^2}{ sin x}dx$$ The table below compares the approximation to the exact result
$$left(
begin{array}{ccc}
t & text{approximation} & text{exact} \
frac{pi }{20} & 0.01221 & 0.01236 \
frac{pi }{10} & 0.04936 & 0.04976 \
frac{3 pi }{20} & 0.11258 & 0.11312 \
frac{pi }{5} & 0.20358 & 0.20409 \
frac{pi }{4} & 0.32475 & 0.32508 \
frac{3 pi }{10} & 0.47939 & 0.47945 \
frac{7 pi }{20} & 0.67196 & 0.67176 \
frac{2 pi }{5} & 0.90847 & 0.90807 \
frac{9 pi }{20} & 1.19701 & 1.19650 \
frac{pi }{2} & 1.54851 & 1.54798 \
frac{11 pi }{20} & 1.97802 & 1.97746 \
frac{3 pi }{5} & 2.50657 & 2.50583 \
frac{13 pi }{20} & 3.16447 & 3.16315 \
frac{7 pi }{10} & 3.99696 & 3.99445 \
frac{3 pi }{4} & 5.07529 & 5.07091 \
frac{4 pi }{5} & 6.52008 & 6.51359 \
frac{17 pi }{20} & 8.55922 & 8.55230 \
frac{9 pi }{10} & 11.7067 & 11.7077 \
frac{19 pi }{20} & 17.6067 & 17.6510
end{array}
right)$$
At the price of special functions, the antiderivative could be computed
$$I=intfrac{x^2}{ sin x},dx=-4 i x text{Li}_2left(e^{i x}right)+i x text{Li}_2left(e^{2 i x}right)+4
text{Li}_3left(e^{i x}right)-frac{1}{2} text{Li}_3left(e^{2 i x}right)-2
x^2 tanh ^{-1}left(e^{i x}right)$$ where appear the polylogarithm functions.
$$lim_{xto frac{pi }{2}} , I=2 pi Cqquad text{and} qquadlim_{xto 0} , I=frac{7 }{2}zeta (3)implies int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dx=2 pi C-frac{7 }{2}zeta (3)$$ as given by Wolfram Alpha. This evaluates a $approx 1.54798$.
For a fast approximation, we could use the superb approximation $$sin(x) simeq frac{16 (pi -x) x}{5 pi ^2-4 (pi -x) x}qquad (0leq xleqpi)$$ which was proposed by Mahabhaskariya of Bhaskara I, a seventh-century Indian mathematician (have a look here).
This would make
$$I approx J= - int left(frac{x^2}{4}+frac{5 pi ^3}{16 (x-pi )}+frac{5 pi ^2}{16} right),dx=-frac{x^3}{12}-frac{5 pi ^2 x}{16}-frac{5}{16} pi ^3 log (pi -x)+frac{19
pi ^3}{48}$$
$$lim_{xto frac{pi }{2}} , J=frac{pi ^3}{48} left(11-15 log left(frac{pi }{2}right)right)qquad text{and} qquadlim_{xto 0} , J
=frac{pi ^3}{48} (19-15 log (pi ))$$ leading to the approximation
$$int_{0}^{frac{pi}{2}}frac{x^2}{ sin x}dxapprox frac{pi ^3}{48} (15 log (2)-8)approx 1.54851$$ which is not too bad.
Tha advantage of such approximation is that it allows a fast evaluation of
$$K(t)=int_{0}^{t}frac{x^2}{ sin x}dx$$ The table below compares the approximation to the exact result
$$left(
begin{array}{ccc}
t & text{approximation} & text{exact} \
frac{pi }{20} & 0.01221 & 0.01236 \
frac{pi }{10} & 0.04936 & 0.04976 \
frac{3 pi }{20} & 0.11258 & 0.11312 \
frac{pi }{5} & 0.20358 & 0.20409 \
frac{pi }{4} & 0.32475 & 0.32508 \
frac{3 pi }{10} & 0.47939 & 0.47945 \
frac{7 pi }{20} & 0.67196 & 0.67176 \
frac{2 pi }{5} & 0.90847 & 0.90807 \
frac{9 pi }{20} & 1.19701 & 1.19650 \
frac{pi }{2} & 1.54851 & 1.54798 \
frac{11 pi }{20} & 1.97802 & 1.97746 \
frac{3 pi }{5} & 2.50657 & 2.50583 \
frac{13 pi }{20} & 3.16447 & 3.16315 \
frac{7 pi }{10} & 3.99696 & 3.99445 \
frac{3 pi }{4} & 5.07529 & 5.07091 \
frac{4 pi }{5} & 6.52008 & 6.51359 \
frac{17 pi }{20} & 8.55922 & 8.55230 \
frac{9 pi }{10} & 11.7067 & 11.7077 \
frac{19 pi }{20} & 17.6067 & 17.6510
end{array}
right)$$
edited Mar 30 at 5:11
answered Mar 30 at 4:47
Claude Leibovici
116k1156131
116k1156131
Thank you for your link ,thats very interesting!
– FofX
Mar 30 at 6:17
1
One may also accelerate the series $$ sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2} $$ to derive accurate numerical approximations of $I$.
– Jack D'Aurizio
Mar 30 at 14:30
add a comment |
Thank you for your link ,thats very interesting!
– FofX
Mar 30 at 6:17
1
One may also accelerate the series $$ sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2} $$ to derive accurate numerical approximations of $I$.
– Jack D'Aurizio
Mar 30 at 14:30
Thank you for your link ,thats very interesting!
– FofX
Mar 30 at 6:17
Thank you for your link ,thats very interesting!
– FofX
Mar 30 at 6:17
1
1
One may also accelerate the series $$ sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2} $$ to derive accurate numerical approximations of $I$.
– Jack D'Aurizio
Mar 30 at 14:30
One may also accelerate the series $$ sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2} $$ to derive accurate numerical approximations of $I$.
– Jack D'Aurizio
Mar 30 at 14:30
add a comment |
up vote
3
down vote
$$mathcal{J}=int_{0}^{pi/2}frac{x^2}{sin x},dx = int_{0}^{1}frac{arcsin^2(x)}{xsqrt{1-x^2}},dx=sum_{ngeq 1}frac{2^{2n-1}}{n^2binom{2n}{n}}int_{0}^{1}frac{x^{2n-1}}{sqrt{1-x^2}},dx tag{1}$$
by the Maclaurin series of $arcsin^2(x)$. Euler's Beta function then leads to
$$ mathcal{J}=sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2}=phantom{}_4 F_3left(1,1,1,1;tfrac{3}{2},tfrac{3}{2},2;1right)tag{2} $$
where the RHS is a manageable hypergeometric function (similar objects are evaluated both here and here) and as already shown by Claude Leibovici, $mathcal{J}=4int_{0}^{1}frac{arctan^2(u)}{u},du $ is simply given by a combination of a dilogarithm and a trilogarithm. Indeed
$$ int_{0}^{pi/2}int_{0}^{theta}frac{u}{sin u},du,dtheta =-pi G+frac{7}{2}zeta(3)tag{3}$$
leading to $mathcal{J}=2pi G-frac{7}{2}zeta(3)$, has already been a key lemma in this historical thread.
An alternative way for proving this identity is just to write $frac{x}{sin x}$ and $|x|$ as Fourier cosine series.
The Shafer-Fink inequality leads to
$$ int_{0}^{pi/2}frac{x^2}{sin x},dx = 4 int_{0}^{1}frac{arctan^2(u)}{u},du approx frac{6}{7}(3sqrt{2}-5)+9logleft(frac{2sqrt{2}+1}{3}right)approx 1.54.tag{4}$$
@ Wow, thank you for your detailed play. I learned a lot.
– FofX
Mar 31 at 3:55
add a comment |
up vote
3
down vote
$$mathcal{J}=int_{0}^{pi/2}frac{x^2}{sin x},dx = int_{0}^{1}frac{arcsin^2(x)}{xsqrt{1-x^2}},dx=sum_{ngeq 1}frac{2^{2n-1}}{n^2binom{2n}{n}}int_{0}^{1}frac{x^{2n-1}}{sqrt{1-x^2}},dx tag{1}$$
by the Maclaurin series of $arcsin^2(x)$. Euler's Beta function then leads to
$$ mathcal{J}=sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2}=phantom{}_4 F_3left(1,1,1,1;tfrac{3}{2},tfrac{3}{2},2;1right)tag{2} $$
where the RHS is a manageable hypergeometric function (similar objects are evaluated both here and here) and as already shown by Claude Leibovici, $mathcal{J}=4int_{0}^{1}frac{arctan^2(u)}{u},du $ is simply given by a combination of a dilogarithm and a trilogarithm. Indeed
$$ int_{0}^{pi/2}int_{0}^{theta}frac{u}{sin u},du,dtheta =-pi G+frac{7}{2}zeta(3)tag{3}$$
leading to $mathcal{J}=2pi G-frac{7}{2}zeta(3)$, has already been a key lemma in this historical thread.
An alternative way for proving this identity is just to write $frac{x}{sin x}$ and $|x|$ as Fourier cosine series.
The Shafer-Fink inequality leads to
$$ int_{0}^{pi/2}frac{x^2}{sin x},dx = 4 int_{0}^{1}frac{arctan^2(u)}{u},du approx frac{6}{7}(3sqrt{2}-5)+9logleft(frac{2sqrt{2}+1}{3}right)approx 1.54.tag{4}$$
@ Wow, thank you for your detailed play. I learned a lot.
– FofX
Mar 31 at 3:55
add a comment |
up vote
3
down vote
up vote
3
down vote
$$mathcal{J}=int_{0}^{pi/2}frac{x^2}{sin x},dx = int_{0}^{1}frac{arcsin^2(x)}{xsqrt{1-x^2}},dx=sum_{ngeq 1}frac{2^{2n-1}}{n^2binom{2n}{n}}int_{0}^{1}frac{x^{2n-1}}{sqrt{1-x^2}},dx tag{1}$$
by the Maclaurin series of $arcsin^2(x)$. Euler's Beta function then leads to
$$ mathcal{J}=sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2}=phantom{}_4 F_3left(1,1,1,1;tfrac{3}{2},tfrac{3}{2},2;1right)tag{2} $$
where the RHS is a manageable hypergeometric function (similar objects are evaluated both here and here) and as already shown by Claude Leibovici, $mathcal{J}=4int_{0}^{1}frac{arctan^2(u)}{u},du $ is simply given by a combination of a dilogarithm and a trilogarithm. Indeed
$$ int_{0}^{pi/2}int_{0}^{theta}frac{u}{sin u},du,dtheta =-pi G+frac{7}{2}zeta(3)tag{3}$$
leading to $mathcal{J}=2pi G-frac{7}{2}zeta(3)$, has already been a key lemma in this historical thread.
An alternative way for proving this identity is just to write $frac{x}{sin x}$ and $|x|$ as Fourier cosine series.
The Shafer-Fink inequality leads to
$$ int_{0}^{pi/2}frac{x^2}{sin x},dx = 4 int_{0}^{1}frac{arctan^2(u)}{u},du approx frac{6}{7}(3sqrt{2}-5)+9logleft(frac{2sqrt{2}+1}{3}right)approx 1.54.tag{4}$$
$$mathcal{J}=int_{0}^{pi/2}frac{x^2}{sin x},dx = int_{0}^{1}frac{arcsin^2(x)}{xsqrt{1-x^2}},dx=sum_{ngeq 1}frac{2^{2n-1}}{n^2binom{2n}{n}}int_{0}^{1}frac{x^{2n-1}}{sqrt{1-x^2}},dx tag{1}$$
by the Maclaurin series of $arcsin^2(x)$. Euler's Beta function then leads to
$$ mathcal{J}=sum_{ngeq 1}frac{16^n}{4n^3 binom{2n}{n}^2}=phantom{}_4 F_3left(1,1,1,1;tfrac{3}{2},tfrac{3}{2},2;1right)tag{2} $$
where the RHS is a manageable hypergeometric function (similar objects are evaluated both here and here) and as already shown by Claude Leibovici, $mathcal{J}=4int_{0}^{1}frac{arctan^2(u)}{u},du $ is simply given by a combination of a dilogarithm and a trilogarithm. Indeed
$$ int_{0}^{pi/2}int_{0}^{theta}frac{u}{sin u},du,dtheta =-pi G+frac{7}{2}zeta(3)tag{3}$$
leading to $mathcal{J}=2pi G-frac{7}{2}zeta(3)$, has already been a key lemma in this historical thread.
An alternative way for proving this identity is just to write $frac{x}{sin x}$ and $|x|$ as Fourier cosine series.
The Shafer-Fink inequality leads to
$$ int_{0}^{pi/2}frac{x^2}{sin x},dx = 4 int_{0}^{1}frac{arctan^2(u)}{u},du approx frac{6}{7}(3sqrt{2}-5)+9logleft(frac{2sqrt{2}+1}{3}right)approx 1.54.tag{4}$$
edited Mar 30 at 14:48
answered Mar 30 at 14:22
Jack D'Aurizio
283k33275653
283k33275653
@ Wow, thank you for your detailed play. I learned a lot.
– FofX
Mar 31 at 3:55
add a comment |
@ Wow, thank you for your detailed play. I learned a lot.
– FofX
Mar 31 at 3:55
@ Wow, thank you for your detailed play. I learned a lot.
– FofX
Mar 31 at 3:55
@ Wow, thank you for your detailed play. I learned a lot.
– FofX
Mar 31 at 3:55
add a comment |
up vote
3
down vote
$begin{align}
J&=int_{0}^{pi/2}frac{x^2}{sin x},dx&
end{align}$
Perform the change of variable,
$displaystyle y=tanleft(frac{x}{2}right)$,
$begin{align}
J&=4int_0^1 frac{arctan^2 x}{x},dx\
&=4Big[ln xarctan ^2 xBig]_0^1-8int_0^1 frac{arctan xln x}{1+x^2},dx\
&=-8int_0^1 frac{arctan xln x}{1+x^2},dx\
end{align}$
For $xin [0;1]$, define $F$,
$begin{align} F(x)&=int_0^x frac{ln t}{1+t^2},dt\
&=int_0^1 frac{xln(xt)}{1+x^2t^2},dt
end{align}$
Observe that,
$displaystyle F(0)=0$ and, $displaystyle F(1)=-text{G}$.
$text{G}$ is the Catalan constant.
$begin{align}J&=-8Big[F(x)arctan xBig]_0^1+8int_0^1 int_0^1 frac{xln(tx)}{(1+t^2x^2)(1+x^2)},dt,dx\
&=2Gpi+8int_0^1 int_0^1 frac{xln x}{(1+t^2x^2)(1+x^2)},dt,dx+8int_0^1 int_0^1 frac{xln t}{(1+t^2x^2)(1+x^2)},dt,dx\
&=2Gpi+8int_0^1 Big[frac{arctan(tx)ln x}{1+x^2}Big]_{t=0}^{t=1},dx+4int_0^1 Big[frac{(ln(1+t^2x^2)-ln(1+x^2))ln t}{t^2-1}Big]_{x=0}^{x=1},dt\
&=2Gpi+8int_0^1 frac{arctan xln x}{1+x^2},dx+4int_0^1 frac{(ln(1+t^2)-ln 2)ln t }{t^2-1},dt\
&=2Gpi-J+4int_0^1 frac{(ln 2-ln(1+t^2))ln t }{1-t^2},dt\
end{align}$
Therefore,
$displaystyle J=text{G}pi+2int_0^1 frac{(ln 2-ln(1+x^2))ln x }{1-x^2},dx$
For $xin[0;1]$, define,
$begin{align}H(x)&=int_0^x frac{ln t}{1-t^2},dt\
&=int_0^1 frac{xln(tx)}{1-t^2x^2},dt\
end{align}$
Observe that,
$displaystyle H(0)=0$ and $displaystyle H(1)=-frac{pi^2}{8}$.
$begin{align}J&=text{G}pi+2Big[(ln 2-ln(1+x^2))H(x)Big]_0^1+4int_0^1int_0^1frac{x^2ln(tx)}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1int_0^1frac{x^2ln(tx)}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1int_0^1frac{x^2ln t}{(1+x^2)(1-t^2x^2)},dt,dx+4int_0^1int_0^1frac{x^2ln x}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1Big[frac{ln t}{1+t^2}left(frac{ln(1+tx)}{2t}-frac{ln(1-tx)}{2t}-arctan xright)Big]_{x=0}^{x=1},dt+\
&2int_0^1 Big[frac{xln x}{1+x^2}lnleft(frac{1+tx}{1-tx}right)Big]_{t=0}^{t=1},dx\
&=text{G}pi+2int_0^1 frac{ln t}{t(1+t^2)}lnleft(frac{1+t}{1-t}right),dt-piint_0^1 frac{ln t}{1+t^2},dt+2int_0^1 frac{xln x}{1+x^2}lnleft(frac{1+x}{1-x}right),dx\
&=2text{G}pi+2int_0^1 frac{ln x}{x}lnleft(frac{1+x}{1-x}right),dx\
end{align}$
But, for $0leq x<1$,
$displaystyle frac{1}{x}lnleft(frac{1+x}{1-x}right)=2sum_{n=0}^{infty}frac{x^{2n}}{2n+1}$
Therefore,
$begin{align}int_0^1 frac{ln x}{x}lnleft(frac{1+x}{1-x}right),dx&=2int_0^1 left(sum_{n=0}^{infty}frac{x^{2n}}{2n+1}right)ln x,dx\
&=2 sum_{n=0}^{infty}int_0^1 frac{x^{2n}ln x}{2n+1},dx\
&=-2sum_{n=0}^{infty}frac{1}{(2n+1)^3}\
&=-2left(sum_{n=1}^{infty} frac{1}{n^3}-sum_{n=1}^{infty} frac{1}{(2n)^3}right)\
&=-2left(zeta(3)-frac{1}{8}zeta(3)right)\
&=-frac{7}{4}zeta(3)\
end{align}$
Therefore,
$ boxed{J=2text{G}pi-frac{7}{2}zeta(3)}$
add a comment |
up vote
3
down vote
$begin{align}
J&=int_{0}^{pi/2}frac{x^2}{sin x},dx&
end{align}$
Perform the change of variable,
$displaystyle y=tanleft(frac{x}{2}right)$,
$begin{align}
J&=4int_0^1 frac{arctan^2 x}{x},dx\
&=4Big[ln xarctan ^2 xBig]_0^1-8int_0^1 frac{arctan xln x}{1+x^2},dx\
&=-8int_0^1 frac{arctan xln x}{1+x^2},dx\
end{align}$
For $xin [0;1]$, define $F$,
$begin{align} F(x)&=int_0^x frac{ln t}{1+t^2},dt\
&=int_0^1 frac{xln(xt)}{1+x^2t^2},dt
end{align}$
Observe that,
$displaystyle F(0)=0$ and, $displaystyle F(1)=-text{G}$.
$text{G}$ is the Catalan constant.
$begin{align}J&=-8Big[F(x)arctan xBig]_0^1+8int_0^1 int_0^1 frac{xln(tx)}{(1+t^2x^2)(1+x^2)},dt,dx\
&=2Gpi+8int_0^1 int_0^1 frac{xln x}{(1+t^2x^2)(1+x^2)},dt,dx+8int_0^1 int_0^1 frac{xln t}{(1+t^2x^2)(1+x^2)},dt,dx\
&=2Gpi+8int_0^1 Big[frac{arctan(tx)ln x}{1+x^2}Big]_{t=0}^{t=1},dx+4int_0^1 Big[frac{(ln(1+t^2x^2)-ln(1+x^2))ln t}{t^2-1}Big]_{x=0}^{x=1},dt\
&=2Gpi+8int_0^1 frac{arctan xln x}{1+x^2},dx+4int_0^1 frac{(ln(1+t^2)-ln 2)ln t }{t^2-1},dt\
&=2Gpi-J+4int_0^1 frac{(ln 2-ln(1+t^2))ln t }{1-t^2},dt\
end{align}$
Therefore,
$displaystyle J=text{G}pi+2int_0^1 frac{(ln 2-ln(1+x^2))ln x }{1-x^2},dx$
For $xin[0;1]$, define,
$begin{align}H(x)&=int_0^x frac{ln t}{1-t^2},dt\
&=int_0^1 frac{xln(tx)}{1-t^2x^2},dt\
end{align}$
Observe that,
$displaystyle H(0)=0$ and $displaystyle H(1)=-frac{pi^2}{8}$.
$begin{align}J&=text{G}pi+2Big[(ln 2-ln(1+x^2))H(x)Big]_0^1+4int_0^1int_0^1frac{x^2ln(tx)}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1int_0^1frac{x^2ln(tx)}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1int_0^1frac{x^2ln t}{(1+x^2)(1-t^2x^2)},dt,dx+4int_0^1int_0^1frac{x^2ln x}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1Big[frac{ln t}{1+t^2}left(frac{ln(1+tx)}{2t}-frac{ln(1-tx)}{2t}-arctan xright)Big]_{x=0}^{x=1},dt+\
&2int_0^1 Big[frac{xln x}{1+x^2}lnleft(frac{1+tx}{1-tx}right)Big]_{t=0}^{t=1},dx\
&=text{G}pi+2int_0^1 frac{ln t}{t(1+t^2)}lnleft(frac{1+t}{1-t}right),dt-piint_0^1 frac{ln t}{1+t^2},dt+2int_0^1 frac{xln x}{1+x^2}lnleft(frac{1+x}{1-x}right),dx\
&=2text{G}pi+2int_0^1 frac{ln x}{x}lnleft(frac{1+x}{1-x}right),dx\
end{align}$
But, for $0leq x<1$,
$displaystyle frac{1}{x}lnleft(frac{1+x}{1-x}right)=2sum_{n=0}^{infty}frac{x^{2n}}{2n+1}$
Therefore,
$begin{align}int_0^1 frac{ln x}{x}lnleft(frac{1+x}{1-x}right),dx&=2int_0^1 left(sum_{n=0}^{infty}frac{x^{2n}}{2n+1}right)ln x,dx\
&=2 sum_{n=0}^{infty}int_0^1 frac{x^{2n}ln x}{2n+1},dx\
&=-2sum_{n=0}^{infty}frac{1}{(2n+1)^3}\
&=-2left(sum_{n=1}^{infty} frac{1}{n^3}-sum_{n=1}^{infty} frac{1}{(2n)^3}right)\
&=-2left(zeta(3)-frac{1}{8}zeta(3)right)\
&=-frac{7}{4}zeta(3)\
end{align}$
Therefore,
$ boxed{J=2text{G}pi-frac{7}{2}zeta(3)}$
add a comment |
up vote
3
down vote
up vote
3
down vote
$begin{align}
J&=int_{0}^{pi/2}frac{x^2}{sin x},dx&
end{align}$
Perform the change of variable,
$displaystyle y=tanleft(frac{x}{2}right)$,
$begin{align}
J&=4int_0^1 frac{arctan^2 x}{x},dx\
&=4Big[ln xarctan ^2 xBig]_0^1-8int_0^1 frac{arctan xln x}{1+x^2},dx\
&=-8int_0^1 frac{arctan xln x}{1+x^2},dx\
end{align}$
For $xin [0;1]$, define $F$,
$begin{align} F(x)&=int_0^x frac{ln t}{1+t^2},dt\
&=int_0^1 frac{xln(xt)}{1+x^2t^2},dt
end{align}$
Observe that,
$displaystyle F(0)=0$ and, $displaystyle F(1)=-text{G}$.
$text{G}$ is the Catalan constant.
$begin{align}J&=-8Big[F(x)arctan xBig]_0^1+8int_0^1 int_0^1 frac{xln(tx)}{(1+t^2x^2)(1+x^2)},dt,dx\
&=2Gpi+8int_0^1 int_0^1 frac{xln x}{(1+t^2x^2)(1+x^2)},dt,dx+8int_0^1 int_0^1 frac{xln t}{(1+t^2x^2)(1+x^2)},dt,dx\
&=2Gpi+8int_0^1 Big[frac{arctan(tx)ln x}{1+x^2}Big]_{t=0}^{t=1},dx+4int_0^1 Big[frac{(ln(1+t^2x^2)-ln(1+x^2))ln t}{t^2-1}Big]_{x=0}^{x=1},dt\
&=2Gpi+8int_0^1 frac{arctan xln x}{1+x^2},dx+4int_0^1 frac{(ln(1+t^2)-ln 2)ln t }{t^2-1},dt\
&=2Gpi-J+4int_0^1 frac{(ln 2-ln(1+t^2))ln t }{1-t^2},dt\
end{align}$
Therefore,
$displaystyle J=text{G}pi+2int_0^1 frac{(ln 2-ln(1+x^2))ln x }{1-x^2},dx$
For $xin[0;1]$, define,
$begin{align}H(x)&=int_0^x frac{ln t}{1-t^2},dt\
&=int_0^1 frac{xln(tx)}{1-t^2x^2},dt\
end{align}$
Observe that,
$displaystyle H(0)=0$ and $displaystyle H(1)=-frac{pi^2}{8}$.
$begin{align}J&=text{G}pi+2Big[(ln 2-ln(1+x^2))H(x)Big]_0^1+4int_0^1int_0^1frac{x^2ln(tx)}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1int_0^1frac{x^2ln(tx)}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1int_0^1frac{x^2ln t}{(1+x^2)(1-t^2x^2)},dt,dx+4int_0^1int_0^1frac{x^2ln x}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1Big[frac{ln t}{1+t^2}left(frac{ln(1+tx)}{2t}-frac{ln(1-tx)}{2t}-arctan xright)Big]_{x=0}^{x=1},dt+\
&2int_0^1 Big[frac{xln x}{1+x^2}lnleft(frac{1+tx}{1-tx}right)Big]_{t=0}^{t=1},dx\
&=text{G}pi+2int_0^1 frac{ln t}{t(1+t^2)}lnleft(frac{1+t}{1-t}right),dt-piint_0^1 frac{ln t}{1+t^2},dt+2int_0^1 frac{xln x}{1+x^2}lnleft(frac{1+x}{1-x}right),dx\
&=2text{G}pi+2int_0^1 frac{ln x}{x}lnleft(frac{1+x}{1-x}right),dx\
end{align}$
But, for $0leq x<1$,
$displaystyle frac{1}{x}lnleft(frac{1+x}{1-x}right)=2sum_{n=0}^{infty}frac{x^{2n}}{2n+1}$
Therefore,
$begin{align}int_0^1 frac{ln x}{x}lnleft(frac{1+x}{1-x}right),dx&=2int_0^1 left(sum_{n=0}^{infty}frac{x^{2n}}{2n+1}right)ln x,dx\
&=2 sum_{n=0}^{infty}int_0^1 frac{x^{2n}ln x}{2n+1},dx\
&=-2sum_{n=0}^{infty}frac{1}{(2n+1)^3}\
&=-2left(sum_{n=1}^{infty} frac{1}{n^3}-sum_{n=1}^{infty} frac{1}{(2n)^3}right)\
&=-2left(zeta(3)-frac{1}{8}zeta(3)right)\
&=-frac{7}{4}zeta(3)\
end{align}$
Therefore,
$ boxed{J=2text{G}pi-frac{7}{2}zeta(3)}$
$begin{align}
J&=int_{0}^{pi/2}frac{x^2}{sin x},dx&
end{align}$
Perform the change of variable,
$displaystyle y=tanleft(frac{x}{2}right)$,
$begin{align}
J&=4int_0^1 frac{arctan^2 x}{x},dx\
&=4Big[ln xarctan ^2 xBig]_0^1-8int_0^1 frac{arctan xln x}{1+x^2},dx\
&=-8int_0^1 frac{arctan xln x}{1+x^2},dx\
end{align}$
For $xin [0;1]$, define $F$,
$begin{align} F(x)&=int_0^x frac{ln t}{1+t^2},dt\
&=int_0^1 frac{xln(xt)}{1+x^2t^2},dt
end{align}$
Observe that,
$displaystyle F(0)=0$ and, $displaystyle F(1)=-text{G}$.
$text{G}$ is the Catalan constant.
$begin{align}J&=-8Big[F(x)arctan xBig]_0^1+8int_0^1 int_0^1 frac{xln(tx)}{(1+t^2x^2)(1+x^2)},dt,dx\
&=2Gpi+8int_0^1 int_0^1 frac{xln x}{(1+t^2x^2)(1+x^2)},dt,dx+8int_0^1 int_0^1 frac{xln t}{(1+t^2x^2)(1+x^2)},dt,dx\
&=2Gpi+8int_0^1 Big[frac{arctan(tx)ln x}{1+x^2}Big]_{t=0}^{t=1},dx+4int_0^1 Big[frac{(ln(1+t^2x^2)-ln(1+x^2))ln t}{t^2-1}Big]_{x=0}^{x=1},dt\
&=2Gpi+8int_0^1 frac{arctan xln x}{1+x^2},dx+4int_0^1 frac{(ln(1+t^2)-ln 2)ln t }{t^2-1},dt\
&=2Gpi-J+4int_0^1 frac{(ln 2-ln(1+t^2))ln t }{1-t^2},dt\
end{align}$
Therefore,
$displaystyle J=text{G}pi+2int_0^1 frac{(ln 2-ln(1+x^2))ln x }{1-x^2},dx$
For $xin[0;1]$, define,
$begin{align}H(x)&=int_0^x frac{ln t}{1-t^2},dt\
&=int_0^1 frac{xln(tx)}{1-t^2x^2},dt\
end{align}$
Observe that,
$displaystyle H(0)=0$ and $displaystyle H(1)=-frac{pi^2}{8}$.
$begin{align}J&=text{G}pi+2Big[(ln 2-ln(1+x^2))H(x)Big]_0^1+4int_0^1int_0^1frac{x^2ln(tx)}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1int_0^1frac{x^2ln(tx)}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1int_0^1frac{x^2ln t}{(1+x^2)(1-t^2x^2)},dt,dx+4int_0^1int_0^1frac{x^2ln x}{(1+x^2)(1-t^2x^2)},dt,dx\
&=text{G}pi+4int_0^1Big[frac{ln t}{1+t^2}left(frac{ln(1+tx)}{2t}-frac{ln(1-tx)}{2t}-arctan xright)Big]_{x=0}^{x=1},dt+\
&2int_0^1 Big[frac{xln x}{1+x^2}lnleft(frac{1+tx}{1-tx}right)Big]_{t=0}^{t=1},dx\
&=text{G}pi+2int_0^1 frac{ln t}{t(1+t^2)}lnleft(frac{1+t}{1-t}right),dt-piint_0^1 frac{ln t}{1+t^2},dt+2int_0^1 frac{xln x}{1+x^2}lnleft(frac{1+x}{1-x}right),dx\
&=2text{G}pi+2int_0^1 frac{ln x}{x}lnleft(frac{1+x}{1-x}right),dx\
end{align}$
But, for $0leq x<1$,
$displaystyle frac{1}{x}lnleft(frac{1+x}{1-x}right)=2sum_{n=0}^{infty}frac{x^{2n}}{2n+1}$
Therefore,
$begin{align}int_0^1 frac{ln x}{x}lnleft(frac{1+x}{1-x}right),dx&=2int_0^1 left(sum_{n=0}^{infty}frac{x^{2n}}{2n+1}right)ln x,dx\
&=2 sum_{n=0}^{infty}int_0^1 frac{x^{2n}ln x}{2n+1},dx\
&=-2sum_{n=0}^{infty}frac{1}{(2n+1)^3}\
&=-2left(sum_{n=1}^{infty} frac{1}{n^3}-sum_{n=1}^{infty} frac{1}{(2n)^3}right)\
&=-2left(zeta(3)-frac{1}{8}zeta(3)right)\
&=-frac{7}{4}zeta(3)\
end{align}$
Therefore,
$ boxed{J=2text{G}pi-frac{7}{2}zeta(3)}$
answered Mar 31 at 22:25
FDP
4,44411221
4,44411221
add a comment |
add a comment |
up vote
3
down vote
Observe we have
begin{align}
I=int^{pi/2}_0 frac{x^2}{sin x} dx = int^{pi/2}_0 frac{x^2}{cosleft(frac{pi}{2}-x right)} dx = int^{pi/2}_0 frac{(frac{pi}{2}-u)^2}{cos u} du.
end{align}
Then using integration by parts, we see that
begin{align}
I&=left(frac{pi}{2}-uright)^2left{logleft|1 + sin uright|-log|cos u|right}bigg|^{pi/2}_0 + 2int^{pi/2}_0left(frac{pi}{2}-u right)log|sec u + tan u| du\
&= 2pi left(frac{1}{2}int_{0}^{pi/2}log|sec u+tan u| du right)-frac{7}{2}left(frac{4}{7}int^{pi/2}_0 u log|sec u+tan u| du right)\
&= 2pi G - frac{7}{2}zeta(3).
end{align}
Here, I have used the facts that
begin{align}
G= frac{1}{2}int_{0}^{pi/2}log|sec u+tan u| du
end{align}
and
begin{align}
zeta(3) = frac{4}{7}int^{pi/2}_0 u log|sec u+tan u| du.
end{align}
See here for reference.
Thanks for your link,it is helpful for me.
– FofX
Mar 30 at 6:08
add a comment |
up vote
3
down vote
Observe we have
begin{align}
I=int^{pi/2}_0 frac{x^2}{sin x} dx = int^{pi/2}_0 frac{x^2}{cosleft(frac{pi}{2}-x right)} dx = int^{pi/2}_0 frac{(frac{pi}{2}-u)^2}{cos u} du.
end{align}
Then using integration by parts, we see that
begin{align}
I&=left(frac{pi}{2}-uright)^2left{logleft|1 + sin uright|-log|cos u|right}bigg|^{pi/2}_0 + 2int^{pi/2}_0left(frac{pi}{2}-u right)log|sec u + tan u| du\
&= 2pi left(frac{1}{2}int_{0}^{pi/2}log|sec u+tan u| du right)-frac{7}{2}left(frac{4}{7}int^{pi/2}_0 u log|sec u+tan u| du right)\
&= 2pi G - frac{7}{2}zeta(3).
end{align}
Here, I have used the facts that
begin{align}
G= frac{1}{2}int_{0}^{pi/2}log|sec u+tan u| du
end{align}
and
begin{align}
zeta(3) = frac{4}{7}int^{pi/2}_0 u log|sec u+tan u| du.
end{align}
See here for reference.
Thanks for your link,it is helpful for me.
– FofX
Mar 30 at 6:08
add a comment |
up vote
3
down vote
up vote
3
down vote
Observe we have
begin{align}
I=int^{pi/2}_0 frac{x^2}{sin x} dx = int^{pi/2}_0 frac{x^2}{cosleft(frac{pi}{2}-x right)} dx = int^{pi/2}_0 frac{(frac{pi}{2}-u)^2}{cos u} du.
end{align}
Then using integration by parts, we see that
begin{align}
I&=left(frac{pi}{2}-uright)^2left{logleft|1 + sin uright|-log|cos u|right}bigg|^{pi/2}_0 + 2int^{pi/2}_0left(frac{pi}{2}-u right)log|sec u + tan u| du\
&= 2pi left(frac{1}{2}int_{0}^{pi/2}log|sec u+tan u| du right)-frac{7}{2}left(frac{4}{7}int^{pi/2}_0 u log|sec u+tan u| du right)\
&= 2pi G - frac{7}{2}zeta(3).
end{align}
Here, I have used the facts that
begin{align}
G= frac{1}{2}int_{0}^{pi/2}log|sec u+tan u| du
end{align}
and
begin{align}
zeta(3) = frac{4}{7}int^{pi/2}_0 u log|sec u+tan u| du.
end{align}
See here for reference.
Observe we have
begin{align}
I=int^{pi/2}_0 frac{x^2}{sin x} dx = int^{pi/2}_0 frac{x^2}{cosleft(frac{pi}{2}-x right)} dx = int^{pi/2}_0 frac{(frac{pi}{2}-u)^2}{cos u} du.
end{align}
Then using integration by parts, we see that
begin{align}
I&=left(frac{pi}{2}-uright)^2left{logleft|1 + sin uright|-log|cos u|right}bigg|^{pi/2}_0 + 2int^{pi/2}_0left(frac{pi}{2}-u right)log|sec u + tan u| du\
&= 2pi left(frac{1}{2}int_{0}^{pi/2}log|sec u+tan u| du right)-frac{7}{2}left(frac{4}{7}int^{pi/2}_0 u log|sec u+tan u| du right)\
&= 2pi G - frac{7}{2}zeta(3).
end{align}
Here, I have used the facts that
begin{align}
G= frac{1}{2}int_{0}^{pi/2}log|sec u+tan u| du
end{align}
and
begin{align}
zeta(3) = frac{4}{7}int^{pi/2}_0 u log|sec u+tan u| du.
end{align}
See here for reference.
edited Nov 14 at 11:00
DavidG
805513
805513
answered Mar 30 at 4:43
Jacky Chong
17.2k21027
17.2k21027
Thanks for your link,it is helpful for me.
– FofX
Mar 30 at 6:08
add a comment |
Thanks for your link,it is helpful for me.
– FofX
Mar 30 at 6:08
Thanks for your link,it is helpful for me.
– FofX
Mar 30 at 6:08
Thanks for your link,it is helpful for me.
– FofX
Mar 30 at 6:08
add a comment |
up vote
2
down vote
$newcommand{bbx}[1]{,bbox[15px,border:1px groove navy]{displaystyle{#1}},}
newcommand{braces}[1]{leftlbrace,{#1},rightrbrace}
newcommand{bracks}[1]{leftlbrack,{#1},rightrbrack}
newcommand{dd}{mathrm{d}}
newcommand{ds}[1]{displaystyle{#1}}
newcommand{expo}[1]{,mathrm{e}^{#1},}
newcommand{ic}{mathrm{i}}
newcommand{mc}[1]{mathcal{#1}}
newcommand{mrm}[1]{mathrm{#1}}
newcommand{pars}[1]{left(,{#1},right)}
newcommand{partiald}[3]{frac{partial^{#1} #2}{partial #3^{#1}}}
newcommand{root}[2]{,sqrt[#1]{,{#2},},}
newcommand{totald}[3]{frac{mathrm{d}^{#1} #2}{mathrm{d} #3^{#1}}}
newcommand{verts}[1]{leftvert,{#1},rightvert}$
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
left.Reint_{x = 0}^{x = pi/2}{bracks{-iclnpars{z}}^{2} over
pars{z - 1/z}/pars{2ic}},{dd z over ic z}
,rightvert_{ z = exppars{ic x}}
\[5mm] & =
left.2,Reint_{x = 0}^{x = pi/2}{ln^{2}pars{z} over 1 - z^{2}},dd z
,rightvert_{ z = exppars{ic x}}
end{align}
$ds{ln}$ is the $ds{log}$-principal branch. Integration of $ds{{ln^{2}pars{z} over 1 - z^{2}}}$ along the path
$ds{C_{x}cup C_{R}cup C_{y}}$ vanishes out such that
$ds{int_{large C_{R}}{ln^{2}pars{z} over 1 - z^{2}},dd z =
-int_{large C_{y}}{ln^{2}pars{z} over 1 - z^{2}},dd z -
int_{large C_{x}}{ln^{2}pars{z} over 1 - z^{2}},dd z}$
Then,
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
overbrace{-2,Reint_{1}^{0}{bracks{lnpars{y} + piic/2}^{, 2} over
1 + y^{2}},ic,dd y}^{ds{mbox{along} C_{y}}} -
overbrace{2,Reint_{0}^{1}{ln^{2}pars{x} over 1 - x^{2}},dd x}
^{ds{mbox{along} C_{x}}}
\[5mm] & =
-2pi,int_{0}^{1}{lnpars{y} over 1 + y^{2}},dd y -
2int_{0}^{1}{ln^{2}pars{x} over 1 - x^{2}},dd x
end{align}
However, $ds{int_{0}^{1}{lnpars{y} over 1 + y^{2}},dd y = -G}$ where
$ds{G}$ is the Catalan Constant such that
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
2pi G - 2sum_{n = 0}^{infty}
overbrace{int_{0}^{1}ln^{2}pars{x}x^{2n},dd x}
^{ds{2 over pars{2n + 1}^{3}}}
\[5mm] & =
2pi G - 4bracks{sum_{n = 1}^{infty}{1 over n^{3}} -
sum_{n = 1}^{infty}{1 over pars{2n}^{3}}} =
2pi G - {7 over 2}sum_{n = 1}^{infty}{1 over n^{3}}
\[5mm] & = bbx{2pi G - {7 over 2},zetapars{3}} approx 1.5480
end{align}
add a comment |
up vote
2
down vote
$newcommand{bbx}[1]{,bbox[15px,border:1px groove navy]{displaystyle{#1}},}
newcommand{braces}[1]{leftlbrace,{#1},rightrbrace}
newcommand{bracks}[1]{leftlbrack,{#1},rightrbrack}
newcommand{dd}{mathrm{d}}
newcommand{ds}[1]{displaystyle{#1}}
newcommand{expo}[1]{,mathrm{e}^{#1},}
newcommand{ic}{mathrm{i}}
newcommand{mc}[1]{mathcal{#1}}
newcommand{mrm}[1]{mathrm{#1}}
newcommand{pars}[1]{left(,{#1},right)}
newcommand{partiald}[3]{frac{partial^{#1} #2}{partial #3^{#1}}}
newcommand{root}[2]{,sqrt[#1]{,{#2},},}
newcommand{totald}[3]{frac{mathrm{d}^{#1} #2}{mathrm{d} #3^{#1}}}
newcommand{verts}[1]{leftvert,{#1},rightvert}$
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
left.Reint_{x = 0}^{x = pi/2}{bracks{-iclnpars{z}}^{2} over
pars{z - 1/z}/pars{2ic}},{dd z over ic z}
,rightvert_{ z = exppars{ic x}}
\[5mm] & =
left.2,Reint_{x = 0}^{x = pi/2}{ln^{2}pars{z} over 1 - z^{2}},dd z
,rightvert_{ z = exppars{ic x}}
end{align}
$ds{ln}$ is the $ds{log}$-principal branch. Integration of $ds{{ln^{2}pars{z} over 1 - z^{2}}}$ along the path
$ds{C_{x}cup C_{R}cup C_{y}}$ vanishes out such that
$ds{int_{large C_{R}}{ln^{2}pars{z} over 1 - z^{2}},dd z =
-int_{large C_{y}}{ln^{2}pars{z} over 1 - z^{2}},dd z -
int_{large C_{x}}{ln^{2}pars{z} over 1 - z^{2}},dd z}$
Then,
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
overbrace{-2,Reint_{1}^{0}{bracks{lnpars{y} + piic/2}^{, 2} over
1 + y^{2}},ic,dd y}^{ds{mbox{along} C_{y}}} -
overbrace{2,Reint_{0}^{1}{ln^{2}pars{x} over 1 - x^{2}},dd x}
^{ds{mbox{along} C_{x}}}
\[5mm] & =
-2pi,int_{0}^{1}{lnpars{y} over 1 + y^{2}},dd y -
2int_{0}^{1}{ln^{2}pars{x} over 1 - x^{2}},dd x
end{align}
However, $ds{int_{0}^{1}{lnpars{y} over 1 + y^{2}},dd y = -G}$ where
$ds{G}$ is the Catalan Constant such that
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
2pi G - 2sum_{n = 0}^{infty}
overbrace{int_{0}^{1}ln^{2}pars{x}x^{2n},dd x}
^{ds{2 over pars{2n + 1}^{3}}}
\[5mm] & =
2pi G - 4bracks{sum_{n = 1}^{infty}{1 over n^{3}} -
sum_{n = 1}^{infty}{1 over pars{2n}^{3}}} =
2pi G - {7 over 2}sum_{n = 1}^{infty}{1 over n^{3}}
\[5mm] & = bbx{2pi G - {7 over 2},zetapars{3}} approx 1.5480
end{align}
add a comment |
up vote
2
down vote
up vote
2
down vote
$newcommand{bbx}[1]{,bbox[15px,border:1px groove navy]{displaystyle{#1}},}
newcommand{braces}[1]{leftlbrace,{#1},rightrbrace}
newcommand{bracks}[1]{leftlbrack,{#1},rightrbrack}
newcommand{dd}{mathrm{d}}
newcommand{ds}[1]{displaystyle{#1}}
newcommand{expo}[1]{,mathrm{e}^{#1},}
newcommand{ic}{mathrm{i}}
newcommand{mc}[1]{mathcal{#1}}
newcommand{mrm}[1]{mathrm{#1}}
newcommand{pars}[1]{left(,{#1},right)}
newcommand{partiald}[3]{frac{partial^{#1} #2}{partial #3^{#1}}}
newcommand{root}[2]{,sqrt[#1]{,{#2},},}
newcommand{totald}[3]{frac{mathrm{d}^{#1} #2}{mathrm{d} #3^{#1}}}
newcommand{verts}[1]{leftvert,{#1},rightvert}$
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
left.Reint_{x = 0}^{x = pi/2}{bracks{-iclnpars{z}}^{2} over
pars{z - 1/z}/pars{2ic}},{dd z over ic z}
,rightvert_{ z = exppars{ic x}}
\[5mm] & =
left.2,Reint_{x = 0}^{x = pi/2}{ln^{2}pars{z} over 1 - z^{2}},dd z
,rightvert_{ z = exppars{ic x}}
end{align}
$ds{ln}$ is the $ds{log}$-principal branch. Integration of $ds{{ln^{2}pars{z} over 1 - z^{2}}}$ along the path
$ds{C_{x}cup C_{R}cup C_{y}}$ vanishes out such that
$ds{int_{large C_{R}}{ln^{2}pars{z} over 1 - z^{2}},dd z =
-int_{large C_{y}}{ln^{2}pars{z} over 1 - z^{2}},dd z -
int_{large C_{x}}{ln^{2}pars{z} over 1 - z^{2}},dd z}$
Then,
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
overbrace{-2,Reint_{1}^{0}{bracks{lnpars{y} + piic/2}^{, 2} over
1 + y^{2}},ic,dd y}^{ds{mbox{along} C_{y}}} -
overbrace{2,Reint_{0}^{1}{ln^{2}pars{x} over 1 - x^{2}},dd x}
^{ds{mbox{along} C_{x}}}
\[5mm] & =
-2pi,int_{0}^{1}{lnpars{y} over 1 + y^{2}},dd y -
2int_{0}^{1}{ln^{2}pars{x} over 1 - x^{2}},dd x
end{align}
However, $ds{int_{0}^{1}{lnpars{y} over 1 + y^{2}},dd y = -G}$ where
$ds{G}$ is the Catalan Constant such that
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
2pi G - 2sum_{n = 0}^{infty}
overbrace{int_{0}^{1}ln^{2}pars{x}x^{2n},dd x}
^{ds{2 over pars{2n + 1}^{3}}}
\[5mm] & =
2pi G - 4bracks{sum_{n = 1}^{infty}{1 over n^{3}} -
sum_{n = 1}^{infty}{1 over pars{2n}^{3}}} =
2pi G - {7 over 2}sum_{n = 1}^{infty}{1 over n^{3}}
\[5mm] & = bbx{2pi G - {7 over 2},zetapars{3}} approx 1.5480
end{align}
$newcommand{bbx}[1]{,bbox[15px,border:1px groove navy]{displaystyle{#1}},}
newcommand{braces}[1]{leftlbrace,{#1},rightrbrace}
newcommand{bracks}[1]{leftlbrack,{#1},rightrbrack}
newcommand{dd}{mathrm{d}}
newcommand{ds}[1]{displaystyle{#1}}
newcommand{expo}[1]{,mathrm{e}^{#1},}
newcommand{ic}{mathrm{i}}
newcommand{mc}[1]{mathcal{#1}}
newcommand{mrm}[1]{mathrm{#1}}
newcommand{pars}[1]{left(,{#1},right)}
newcommand{partiald}[3]{frac{partial^{#1} #2}{partial #3^{#1}}}
newcommand{root}[2]{,sqrt[#1]{,{#2},},}
newcommand{totald}[3]{frac{mathrm{d}^{#1} #2}{mathrm{d} #3^{#1}}}
newcommand{verts}[1]{leftvert,{#1},rightvert}$
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
left.Reint_{x = 0}^{x = pi/2}{bracks{-iclnpars{z}}^{2} over
pars{z - 1/z}/pars{2ic}},{dd z over ic z}
,rightvert_{ z = exppars{ic x}}
\[5mm] & =
left.2,Reint_{x = 0}^{x = pi/2}{ln^{2}pars{z} over 1 - z^{2}},dd z
,rightvert_{ z = exppars{ic x}}
end{align}
$ds{ln}$ is the $ds{log}$-principal branch. Integration of $ds{{ln^{2}pars{z} over 1 - z^{2}}}$ along the path
$ds{C_{x}cup C_{R}cup C_{y}}$ vanishes out such that
$ds{int_{large C_{R}}{ln^{2}pars{z} over 1 - z^{2}},dd z =
-int_{large C_{y}}{ln^{2}pars{z} over 1 - z^{2}},dd z -
int_{large C_{x}}{ln^{2}pars{z} over 1 - z^{2}},dd z}$
Then,
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
overbrace{-2,Reint_{1}^{0}{bracks{lnpars{y} + piic/2}^{, 2} over
1 + y^{2}},ic,dd y}^{ds{mbox{along} C_{y}}} -
overbrace{2,Reint_{0}^{1}{ln^{2}pars{x} over 1 - x^{2}},dd x}
^{ds{mbox{along} C_{x}}}
\[5mm] & =
-2pi,int_{0}^{1}{lnpars{y} over 1 + y^{2}},dd y -
2int_{0}^{1}{ln^{2}pars{x} over 1 - x^{2}},dd x
end{align}
However, $ds{int_{0}^{1}{lnpars{y} over 1 + y^{2}},dd y = -G}$ where
$ds{G}$ is the Catalan Constant such that
begin{align}
int_{0}^{pi/2}{x^{2} over sinpars{x}},dd x & =
2pi G - 2sum_{n = 0}^{infty}
overbrace{int_{0}^{1}ln^{2}pars{x}x^{2n},dd x}
^{ds{2 over pars{2n + 1}^{3}}}
\[5mm] & =
2pi G - 4bracks{sum_{n = 1}^{infty}{1 over n^{3}} -
sum_{n = 1}^{infty}{1 over pars{2n}^{3}}} =
2pi G - {7 over 2}sum_{n = 1}^{infty}{1 over n^{3}}
\[5mm] & = bbx{2pi G - {7 over 2},zetapars{3}} approx 1.5480
end{align}
edited Apr 5 at 0:06
answered Mar 31 at 6:49
Felix Marin
65.8k7107138
65.8k7107138
add a comment |
add a comment |
up vote
1
down vote
As pointed out within the other answers we want to prove that
$$mathfrak{I}=int_0^{pi/2}frac{x^2}{sin x}~dx=2pi G-frac72zeta(3)$$
As the OP showed $mathfrak{I}$ can be reduced to a linear combination of $x$ and the function $log(cot x)$
$$mathfrak{I}=int_0^{pi/2}frac{x^2}{sin x}~dx=8int_0^{pi/4}xlog(cot x)~dx$$
By applying the definition of the cotangent function followed up by the usage of the well-known Fourier series expansions of $log(cos x)$ and $log(sin x)$ this can be further simplified. Therefore we get
$$smallbegin{align}
mathfrak{I}=8int_0^{pi/4}xlog(cot x)~dx&=8left[int_0^{pi/4}xlog(cos x)~dx-int_0^{pi/4}xlog(sin x)~dxright]\
&=8left[int_0^{pi/4}xleft(-log(2)-sum_{n=1}^{infty}(-1)^nfrac{cos(2nx)}{n}right)~dx-int_0^{pi/4}xleft(-log(2)-sum_{n=1}^{infty}frac{cos(2nx)}{n}right)~dxright]\
&=8left[-sum_{n=1}^{infty}frac{(-1)^n}{n}underbrace{int_0^{pi/4}xcos(2nx)~dx}_I+sum_{n=1}^{infty}frac{1}{n}underbrace{int_0^{pi/4}xcos(2nx)~dx}_Iright]\
end{align}$$
The inner integral $I$ can be easily evaluated using IBP which leads to
$$I=int_0^{pi/4}xcos(2nx)~dx=frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}$$
Plugging this into our original formula and followed by a little bit of algebraic manipulation we get
$$smallbegin{align}
mathfrak{I}&=8left[-sum_{n=1}^{infty}frac{(-1)^n}{n}left(frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}right)+sum_{n=1}^{infty}frac{1}{n}left(frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}right)right]\
&=2left[sum_{n=1}^{infty}frac{(-1)^n}{n^3}-sum_{n=1}^{infty}frac{1}{n^3}right]+pileft[sum_{n=1}^{infty}frac{1}{n^2}sinleft(nfrac{pi}2right)-sum_{n=1}^{infty}frac{(-1)^n}{n^2}sinleft(nfrac{pi}2right)right]
end{align}$$
The first terms can be evaluated in terms of the Riemann Zeta Function $zeta(s)$ and the Dirichlet Eta Function $eta(s)$ whereas for the second term we have to consider some basic properties of the Sine function. For $ninmathbb{N}>0$ the function $sinleft(nfrac{pi}2right)$ will be zero for all even $n$ and $-1$ and $1$ respectively for odd $n$ starting with $sinleft(frac{pi}2right)=1$ for $n=1$. Therefore all even terms vanish while the odd ones will remain with a oscillating negative sign. This leads to
$$smallbegin{align}
mathfrak{I}&=2left[sum_{n=1}^{infty}frac{(-1)^n}{n^3}-sum_{n=1}^{infty}frac{1}{n^3}right]+pileft[sum_{n=1}^{infty}frac{1}{n^2}sinleft(nfrac{pi}2right)-sum_{n=1}^{infty}frac{(-1)^n}{n^2}sinleft(nfrac{pi}2right)right]\
&=2[-eta(3)-zeta(3)]+pileft[sum_{n=0}^{infty}frac{1}{(2n+1)^2}(-1)^n-sum_{n=0}^{infty}frac{(-1)^{2n+1}}{(2n+1)^2}(-1)^nright]\
&=-2[(1-2^{-2})zeta(3)+zeta(3)]+2pisum_{n=0}^{infty}frac{(-1)^n}{(2n+1)^2}\
Leftrightarrowmathfrak{I}&=-frac72zeta(3)+2pi G
end{align}$$
where within the last step the functional relation between the Riemann Zeta Function and the Dirichlet Eta Function aswell as the series defintions of Catalan's Constant $G$ where used.
add a comment |
up vote
1
down vote
As pointed out within the other answers we want to prove that
$$mathfrak{I}=int_0^{pi/2}frac{x^2}{sin x}~dx=2pi G-frac72zeta(3)$$
As the OP showed $mathfrak{I}$ can be reduced to a linear combination of $x$ and the function $log(cot x)$
$$mathfrak{I}=int_0^{pi/2}frac{x^2}{sin x}~dx=8int_0^{pi/4}xlog(cot x)~dx$$
By applying the definition of the cotangent function followed up by the usage of the well-known Fourier series expansions of $log(cos x)$ and $log(sin x)$ this can be further simplified. Therefore we get
$$smallbegin{align}
mathfrak{I}=8int_0^{pi/4}xlog(cot x)~dx&=8left[int_0^{pi/4}xlog(cos x)~dx-int_0^{pi/4}xlog(sin x)~dxright]\
&=8left[int_0^{pi/4}xleft(-log(2)-sum_{n=1}^{infty}(-1)^nfrac{cos(2nx)}{n}right)~dx-int_0^{pi/4}xleft(-log(2)-sum_{n=1}^{infty}frac{cos(2nx)}{n}right)~dxright]\
&=8left[-sum_{n=1}^{infty}frac{(-1)^n}{n}underbrace{int_0^{pi/4}xcos(2nx)~dx}_I+sum_{n=1}^{infty}frac{1}{n}underbrace{int_0^{pi/4}xcos(2nx)~dx}_Iright]\
end{align}$$
The inner integral $I$ can be easily evaluated using IBP which leads to
$$I=int_0^{pi/4}xcos(2nx)~dx=frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}$$
Plugging this into our original formula and followed by a little bit of algebraic manipulation we get
$$smallbegin{align}
mathfrak{I}&=8left[-sum_{n=1}^{infty}frac{(-1)^n}{n}left(frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}right)+sum_{n=1}^{infty}frac{1}{n}left(frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}right)right]\
&=2left[sum_{n=1}^{infty}frac{(-1)^n}{n^3}-sum_{n=1}^{infty}frac{1}{n^3}right]+pileft[sum_{n=1}^{infty}frac{1}{n^2}sinleft(nfrac{pi}2right)-sum_{n=1}^{infty}frac{(-1)^n}{n^2}sinleft(nfrac{pi}2right)right]
end{align}$$
The first terms can be evaluated in terms of the Riemann Zeta Function $zeta(s)$ and the Dirichlet Eta Function $eta(s)$ whereas for the second term we have to consider some basic properties of the Sine function. For $ninmathbb{N}>0$ the function $sinleft(nfrac{pi}2right)$ will be zero for all even $n$ and $-1$ and $1$ respectively for odd $n$ starting with $sinleft(frac{pi}2right)=1$ for $n=1$. Therefore all even terms vanish while the odd ones will remain with a oscillating negative sign. This leads to
$$smallbegin{align}
mathfrak{I}&=2left[sum_{n=1}^{infty}frac{(-1)^n}{n^3}-sum_{n=1}^{infty}frac{1}{n^3}right]+pileft[sum_{n=1}^{infty}frac{1}{n^2}sinleft(nfrac{pi}2right)-sum_{n=1}^{infty}frac{(-1)^n}{n^2}sinleft(nfrac{pi}2right)right]\
&=2[-eta(3)-zeta(3)]+pileft[sum_{n=0}^{infty}frac{1}{(2n+1)^2}(-1)^n-sum_{n=0}^{infty}frac{(-1)^{2n+1}}{(2n+1)^2}(-1)^nright]\
&=-2[(1-2^{-2})zeta(3)+zeta(3)]+2pisum_{n=0}^{infty}frac{(-1)^n}{(2n+1)^2}\
Leftrightarrowmathfrak{I}&=-frac72zeta(3)+2pi G
end{align}$$
where within the last step the functional relation between the Riemann Zeta Function and the Dirichlet Eta Function aswell as the series defintions of Catalan's Constant $G$ where used.
add a comment |
up vote
1
down vote
up vote
1
down vote
As pointed out within the other answers we want to prove that
$$mathfrak{I}=int_0^{pi/2}frac{x^2}{sin x}~dx=2pi G-frac72zeta(3)$$
As the OP showed $mathfrak{I}$ can be reduced to a linear combination of $x$ and the function $log(cot x)$
$$mathfrak{I}=int_0^{pi/2}frac{x^2}{sin x}~dx=8int_0^{pi/4}xlog(cot x)~dx$$
By applying the definition of the cotangent function followed up by the usage of the well-known Fourier series expansions of $log(cos x)$ and $log(sin x)$ this can be further simplified. Therefore we get
$$smallbegin{align}
mathfrak{I}=8int_0^{pi/4}xlog(cot x)~dx&=8left[int_0^{pi/4}xlog(cos x)~dx-int_0^{pi/4}xlog(sin x)~dxright]\
&=8left[int_0^{pi/4}xleft(-log(2)-sum_{n=1}^{infty}(-1)^nfrac{cos(2nx)}{n}right)~dx-int_0^{pi/4}xleft(-log(2)-sum_{n=1}^{infty}frac{cos(2nx)}{n}right)~dxright]\
&=8left[-sum_{n=1}^{infty}frac{(-1)^n}{n}underbrace{int_0^{pi/4}xcos(2nx)~dx}_I+sum_{n=1}^{infty}frac{1}{n}underbrace{int_0^{pi/4}xcos(2nx)~dx}_Iright]\
end{align}$$
The inner integral $I$ can be easily evaluated using IBP which leads to
$$I=int_0^{pi/4}xcos(2nx)~dx=frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}$$
Plugging this into our original formula and followed by a little bit of algebraic manipulation we get
$$smallbegin{align}
mathfrak{I}&=8left[-sum_{n=1}^{infty}frac{(-1)^n}{n}left(frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}right)+sum_{n=1}^{infty}frac{1}{n}left(frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}right)right]\
&=2left[sum_{n=1}^{infty}frac{(-1)^n}{n^3}-sum_{n=1}^{infty}frac{1}{n^3}right]+pileft[sum_{n=1}^{infty}frac{1}{n^2}sinleft(nfrac{pi}2right)-sum_{n=1}^{infty}frac{(-1)^n}{n^2}sinleft(nfrac{pi}2right)right]
end{align}$$
The first terms can be evaluated in terms of the Riemann Zeta Function $zeta(s)$ and the Dirichlet Eta Function $eta(s)$ whereas for the second term we have to consider some basic properties of the Sine function. For $ninmathbb{N}>0$ the function $sinleft(nfrac{pi}2right)$ will be zero for all even $n$ and $-1$ and $1$ respectively for odd $n$ starting with $sinleft(frac{pi}2right)=1$ for $n=1$. Therefore all even terms vanish while the odd ones will remain with a oscillating negative sign. This leads to
$$smallbegin{align}
mathfrak{I}&=2left[sum_{n=1}^{infty}frac{(-1)^n}{n^3}-sum_{n=1}^{infty}frac{1}{n^3}right]+pileft[sum_{n=1}^{infty}frac{1}{n^2}sinleft(nfrac{pi}2right)-sum_{n=1}^{infty}frac{(-1)^n}{n^2}sinleft(nfrac{pi}2right)right]\
&=2[-eta(3)-zeta(3)]+pileft[sum_{n=0}^{infty}frac{1}{(2n+1)^2}(-1)^n-sum_{n=0}^{infty}frac{(-1)^{2n+1}}{(2n+1)^2}(-1)^nright]\
&=-2[(1-2^{-2})zeta(3)+zeta(3)]+2pisum_{n=0}^{infty}frac{(-1)^n}{(2n+1)^2}\
Leftrightarrowmathfrak{I}&=-frac72zeta(3)+2pi G
end{align}$$
where within the last step the functional relation between the Riemann Zeta Function and the Dirichlet Eta Function aswell as the series defintions of Catalan's Constant $G$ where used.
As pointed out within the other answers we want to prove that
$$mathfrak{I}=int_0^{pi/2}frac{x^2}{sin x}~dx=2pi G-frac72zeta(3)$$
As the OP showed $mathfrak{I}$ can be reduced to a linear combination of $x$ and the function $log(cot x)$
$$mathfrak{I}=int_0^{pi/2}frac{x^2}{sin x}~dx=8int_0^{pi/4}xlog(cot x)~dx$$
By applying the definition of the cotangent function followed up by the usage of the well-known Fourier series expansions of $log(cos x)$ and $log(sin x)$ this can be further simplified. Therefore we get
$$smallbegin{align}
mathfrak{I}=8int_0^{pi/4}xlog(cot x)~dx&=8left[int_0^{pi/4}xlog(cos x)~dx-int_0^{pi/4}xlog(sin x)~dxright]\
&=8left[int_0^{pi/4}xleft(-log(2)-sum_{n=1}^{infty}(-1)^nfrac{cos(2nx)}{n}right)~dx-int_0^{pi/4}xleft(-log(2)-sum_{n=1}^{infty}frac{cos(2nx)}{n}right)~dxright]\
&=8left[-sum_{n=1}^{infty}frac{(-1)^n}{n}underbrace{int_0^{pi/4}xcos(2nx)~dx}_I+sum_{n=1}^{infty}frac{1}{n}underbrace{int_0^{pi/4}xcos(2nx)~dx}_Iright]\
end{align}$$
The inner integral $I$ can be easily evaluated using IBP which leads to
$$I=int_0^{pi/4}xcos(2nx)~dx=frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}$$
Plugging this into our original formula and followed by a little bit of algebraic manipulation we get
$$smallbegin{align}
mathfrak{I}&=8left[-sum_{n=1}^{infty}frac{(-1)^n}{n}left(frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}right)+sum_{n=1}^{infty}frac{1}{n}left(frac{pi}{8n}sinleft(nfrac{pi}2right)-frac1{(2n)^2}right)right]\
&=2left[sum_{n=1}^{infty}frac{(-1)^n}{n^3}-sum_{n=1}^{infty}frac{1}{n^3}right]+pileft[sum_{n=1}^{infty}frac{1}{n^2}sinleft(nfrac{pi}2right)-sum_{n=1}^{infty}frac{(-1)^n}{n^2}sinleft(nfrac{pi}2right)right]
end{align}$$
The first terms can be evaluated in terms of the Riemann Zeta Function $zeta(s)$ and the Dirichlet Eta Function $eta(s)$ whereas for the second term we have to consider some basic properties of the Sine function. For $ninmathbb{N}>0$ the function $sinleft(nfrac{pi}2right)$ will be zero for all even $n$ and $-1$ and $1$ respectively for odd $n$ starting with $sinleft(frac{pi}2right)=1$ for $n=1$. Therefore all even terms vanish while the odd ones will remain with a oscillating negative sign. This leads to
$$smallbegin{align}
mathfrak{I}&=2left[sum_{n=1}^{infty}frac{(-1)^n}{n^3}-sum_{n=1}^{infty}frac{1}{n^3}right]+pileft[sum_{n=1}^{infty}frac{1}{n^2}sinleft(nfrac{pi}2right)-sum_{n=1}^{infty}frac{(-1)^n}{n^2}sinleft(nfrac{pi}2right)right]\
&=2[-eta(3)-zeta(3)]+pileft[sum_{n=0}^{infty}frac{1}{(2n+1)^2}(-1)^n-sum_{n=0}^{infty}frac{(-1)^{2n+1}}{(2n+1)^2}(-1)^nright]\
&=-2[(1-2^{-2})zeta(3)+zeta(3)]+2pisum_{n=0}^{infty}frac{(-1)^n}{(2n+1)^2}\
Leftrightarrowmathfrak{I}&=-frac72zeta(3)+2pi G
end{align}$$
where within the last step the functional relation between the Riemann Zeta Function and the Dirichlet Eta Function aswell as the series defintions of Catalan's Constant $G$ where used.
edited Nov 12 at 22:42
answered Nov 12 at 22:30
mrtaurho
2,3691726
2,3691726
add a comment |
add a comment |
up vote
1
down vote
We can adapt the formula derived in $(2)$ of this answer:
$$
log(2cos(x/2))=sum_{k=1}^inftyfrac{(-1)^{k-1}}kcos(kx)tag{1a}
$$
Substituting $xmapstopi-x$ in $text{(1a)}$, we get
$$
log(2sin(x/2))=sum_{k=1}^inftyfrac{-1}kcos(kx)tag{1b}
$$
Subtracting $text{(1a)}$ from $text{(1b)}$, the even terms cancel and we get
$$
bbox[5px,border:2px solid #C0A000]{log(tan(x/2))=sum_{k=0}^inftyfrac{-2}{2k+1}cos((2k+1)x)}tag2
$$
Therefore,
$$
begin{align}
int_0^{pi/2}frac{x^2}{sin(x)},mathrm{d}x
&=int_0^{pi/2}x^2,mathrm{d}log(tan(x/2))tag3\
&=-2int_0^{pi/2}xlog(tan(x/2)),mathrm{d}xtag4\
&=sum_{k=0}^inftyfrac4{2k+1}int_0^{pi/2}xcos((2k+1)x),mathrm{d}xtag5\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}int_0^{pi/2}x,mathrm{d}sin((2k+1)x)tag6\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}left[xsin((2k+1)x)+frac{cos((2k+1)x)}{2k+1}right]_0^{pi/2}tag7\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}left[fracpi2(-1)^k-frac1{2k+1}right]tag8\
&=bbox[5px,border:2px solid #C0A000]{2pimathrm{G}-frac72zeta(3)}tag9
end{align}
$$
Explanation:
$(3)$: prepare to integrate by parts
$(4)$: integrate by parts
$(5)$: apply $(2)$
$(6)$: prepare to integrate by parts
$(7)$: integrate by parts
$(8)$: apply the limits of integration
$(9)$: evaluate, where $mathrm{G}$ is Catalan's Constant
add a comment |
up vote
1
down vote
We can adapt the formula derived in $(2)$ of this answer:
$$
log(2cos(x/2))=sum_{k=1}^inftyfrac{(-1)^{k-1}}kcos(kx)tag{1a}
$$
Substituting $xmapstopi-x$ in $text{(1a)}$, we get
$$
log(2sin(x/2))=sum_{k=1}^inftyfrac{-1}kcos(kx)tag{1b}
$$
Subtracting $text{(1a)}$ from $text{(1b)}$, the even terms cancel and we get
$$
bbox[5px,border:2px solid #C0A000]{log(tan(x/2))=sum_{k=0}^inftyfrac{-2}{2k+1}cos((2k+1)x)}tag2
$$
Therefore,
$$
begin{align}
int_0^{pi/2}frac{x^2}{sin(x)},mathrm{d}x
&=int_0^{pi/2}x^2,mathrm{d}log(tan(x/2))tag3\
&=-2int_0^{pi/2}xlog(tan(x/2)),mathrm{d}xtag4\
&=sum_{k=0}^inftyfrac4{2k+1}int_0^{pi/2}xcos((2k+1)x),mathrm{d}xtag5\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}int_0^{pi/2}x,mathrm{d}sin((2k+1)x)tag6\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}left[xsin((2k+1)x)+frac{cos((2k+1)x)}{2k+1}right]_0^{pi/2}tag7\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}left[fracpi2(-1)^k-frac1{2k+1}right]tag8\
&=bbox[5px,border:2px solid #C0A000]{2pimathrm{G}-frac72zeta(3)}tag9
end{align}
$$
Explanation:
$(3)$: prepare to integrate by parts
$(4)$: integrate by parts
$(5)$: apply $(2)$
$(6)$: prepare to integrate by parts
$(7)$: integrate by parts
$(8)$: apply the limits of integration
$(9)$: evaluate, where $mathrm{G}$ is Catalan's Constant
add a comment |
up vote
1
down vote
up vote
1
down vote
We can adapt the formula derived in $(2)$ of this answer:
$$
log(2cos(x/2))=sum_{k=1}^inftyfrac{(-1)^{k-1}}kcos(kx)tag{1a}
$$
Substituting $xmapstopi-x$ in $text{(1a)}$, we get
$$
log(2sin(x/2))=sum_{k=1}^inftyfrac{-1}kcos(kx)tag{1b}
$$
Subtracting $text{(1a)}$ from $text{(1b)}$, the even terms cancel and we get
$$
bbox[5px,border:2px solid #C0A000]{log(tan(x/2))=sum_{k=0}^inftyfrac{-2}{2k+1}cos((2k+1)x)}tag2
$$
Therefore,
$$
begin{align}
int_0^{pi/2}frac{x^2}{sin(x)},mathrm{d}x
&=int_0^{pi/2}x^2,mathrm{d}log(tan(x/2))tag3\
&=-2int_0^{pi/2}xlog(tan(x/2)),mathrm{d}xtag4\
&=sum_{k=0}^inftyfrac4{2k+1}int_0^{pi/2}xcos((2k+1)x),mathrm{d}xtag5\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}int_0^{pi/2}x,mathrm{d}sin((2k+1)x)tag6\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}left[xsin((2k+1)x)+frac{cos((2k+1)x)}{2k+1}right]_0^{pi/2}tag7\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}left[fracpi2(-1)^k-frac1{2k+1}right]tag8\
&=bbox[5px,border:2px solid #C0A000]{2pimathrm{G}-frac72zeta(3)}tag9
end{align}
$$
Explanation:
$(3)$: prepare to integrate by parts
$(4)$: integrate by parts
$(5)$: apply $(2)$
$(6)$: prepare to integrate by parts
$(7)$: integrate by parts
$(8)$: apply the limits of integration
$(9)$: evaluate, where $mathrm{G}$ is Catalan's Constant
We can adapt the formula derived in $(2)$ of this answer:
$$
log(2cos(x/2))=sum_{k=1}^inftyfrac{(-1)^{k-1}}kcos(kx)tag{1a}
$$
Substituting $xmapstopi-x$ in $text{(1a)}$, we get
$$
log(2sin(x/2))=sum_{k=1}^inftyfrac{-1}kcos(kx)tag{1b}
$$
Subtracting $text{(1a)}$ from $text{(1b)}$, the even terms cancel and we get
$$
bbox[5px,border:2px solid #C0A000]{log(tan(x/2))=sum_{k=0}^inftyfrac{-2}{2k+1}cos((2k+1)x)}tag2
$$
Therefore,
$$
begin{align}
int_0^{pi/2}frac{x^2}{sin(x)},mathrm{d}x
&=int_0^{pi/2}x^2,mathrm{d}log(tan(x/2))tag3\
&=-2int_0^{pi/2}xlog(tan(x/2)),mathrm{d}xtag4\
&=sum_{k=0}^inftyfrac4{2k+1}int_0^{pi/2}xcos((2k+1)x),mathrm{d}xtag5\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}int_0^{pi/2}x,mathrm{d}sin((2k+1)x)tag6\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}left[xsin((2k+1)x)+frac{cos((2k+1)x)}{2k+1}right]_0^{pi/2}tag7\
&=sum_{k=0}^inftyfrac4{(2k+1)^2}left[fracpi2(-1)^k-frac1{2k+1}right]tag8\
&=bbox[5px,border:2px solid #C0A000]{2pimathrm{G}-frac72zeta(3)}tag9
end{align}
$$
Explanation:
$(3)$: prepare to integrate by parts
$(4)$: integrate by parts
$(5)$: apply $(2)$
$(6)$: prepare to integrate by parts
$(7)$: integrate by parts
$(8)$: apply the limits of integration
$(9)$: evaluate, where $mathrm{G}$ is Catalan's Constant
edited Nov 15 at 3:31
answered Nov 14 at 7:22
robjohn♦
262k27300620
262k27300620
add a comment |
add a comment |
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f2714146%2fevaluate-int-0-frac-pi2-fracx2-sin-xdx%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
you mean the first step? Give some thoughts, please. (Although to be fair, it appears we have a doozy wolframalpha.com/input/… ... I'm guessing this means you should try a series approach.)
– spaceisdarkgreen
Mar 30 at 3:40
@FofX Do you want an exact answer, or just a numerical approximation?
– Toby Mak
Mar 30 at 3:41
@spaceisdarkgreen I think use integration by parts?
– FofX
Mar 30 at 3:42
@TobyMak I want an exact answer,hh,thank you.
– FofX
Mar 30 at 3:44
@FofX Don't think so, though I can't say I know it won't simplify things... whatever it is the indefinite integral doesn't seem to come out nice (see my wolfram alpha link). You can use the Taylor series for csc to turn it into an infinite sum that I don't find particularly inviting, but has some features that make the wolfram alpha answer involving the zeta function and Catalan constant plausible, like bernoulli numbers
– spaceisdarkgreen
Mar 30 at 3:56