1
$\begingroup$

As a part of my research (in array processing - the specifics are not too exciting :-)), I cant resolve one specific integral.

Assume $r\in\left[0 ,1\right]$, $N\in\mathbb{N}$.

The basic form is $$ \int_{\cos{(x)}=-1}^{\cos{(x)}=1}{ \frac{ 1-\cos{\left(Nx\right)} }{ N^{2}\left(1-\cos{(x)}\right) +r^{2}\left(1-\cos{\left(Nx\right)}\right) -Nr\left(1-\cos{(x)}-\cos{\left(Nx\right)}+\cos{\left(\left(N-1\right)x\right)}\right) } dx }. $$ Searching for hints, I came across the very informative "How to expand $\cos{(nx)}$ with $\cos{(x)}$?" discussion which lead me to investigate the expression using Chebyshev polynomials of the first kind, in light of the fact that $\cos(nu)=T_n(\cos(u)),$ but still no luck.

Anyhow, as can be seen in the wikipedia page (Chebyshev polynomials ), $$ T_n(x)= \sum_{k=0}^{\lfloor n/2\rfloor} \binom{n}{2k} (x^2-1)^k x^{n-2k}, $$ and plugging it into the basic integral results in $$ \int_{-1}^{1}{ \frac{ 1-T_N(x) }{ N^{2}\left(1-x\right) +r^{2}\left(1-T_N(x)\right) -Nr\left(1-x-T_N(x)+T_{N-1}(x)\right) } \frac{-1}{\sqrt{1-x^{2}}} dx }. $$ Considering also the Chebyshev polynomials of the second kind, $$ U_{N}\left(x\right)=\frac{\left(x+\sqrt{x^{2}-1}\right)^{N+1}-\left(x-\sqrt{x^{2}-1}\right)^{N+1}}{2\sqrt{x^{2}-1}}, $$ and the equality $$ T_{N+1}(x) = xT_{N}(x)-\left(1-x^{2}\right)U_{N-1}(x) $$ results in $$ \int_{-1}^{1}{ \frac{ 1-T_N(x) }{ N^{2}\left(1-x\right) +r^{2}\left(1-T_N(x)\right) -Nr \left( 1-x+\sqrt{1-x^{2}} \left(\left(x+\sqrt{x^{2}-1}\right)^{N}-\left(x-\sqrt{x^{2}-1}\right)^{N}\right) \right) } \frac{-1}{\sqrt{1-x^{2}}} dx }, $$ which has (as the headline states) "almost" quadratic form of the denominator. This is where I am stuck and if someone has an idea for solving it I will be thrilled.

Thank you.

$\endgroup$
1
$\begingroup$

A numerically achieved analytical answer:

sweeping many $[N,r]$ sets in MATHEMATICA, revealed that the finite integral (i.e. integration from $x=0$ to $x=2\pi$) can be expressed as $$ \int_{0}^{2\pi}(...) = \frac{2\pi}{r^{2}-\left(N-1\right)r+N} $$ which is very nice and perfectly fits the fact that for $r=0$ it should be equal to $2\pi/N$.

Now that I have the answer, should someone have an idea for analytical way of solving the integral, I will be very glad to hear about it.

Thanks Itay.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.