3
$\begingroup$

Is there a closed form for the integral $$\int_0^{\pi/2}(P_\nu^\mu(\cos\theta))^2\,\mathrm d\theta,\quad\mu>\nu\gt-\frac12$$ where $P_\nu^\mu(x)$ is the associated Legendre function of the first kind?

I encountered this integral while trying to derive explicit solutions for a certain Sturm-Liouville problem. I am primarily interested in $\mu,\nu$ being nonnegative integers, but a result that is valid for real $\mu,\nu$ (subject to the above restriction) is very much welcome.

Neither Maple nor Mathematica seem to be able to make a dent on this integral, but I was able to at least confirm that for $\mu,\nu$ an integer, I get results that are rational multiples of $\pi$, which makes me believe there ought to be a (simple?) closed form, perhaps involving gamma functions.

I wasn't able to find anything in G&R or the DLMF that resembles this integral, so I am really stuck, and would appreciate any ideas on resolving this.

$\endgroup$
  • $\begingroup$ Using its recurrence formulas it boils down to evaluating the overlap $\int_{-1}^1P_{\nu}^\mu(x) P_{\nu-1}^{\mu\pm1}(x) \mathrm{d} x$ of two associated Legendre polynomials. Perhaps this source helps? $\endgroup$ – Timothy Budd Jul 20 at 17:16
  • $\begingroup$ Your link refers to formula 14.3.6, which holds for $x > 1$ (while $\cos\theta\leq 1$). Do you actually mean formula 14.3.1, that is Ferrers rather than Legendre functions? $\endgroup$ – Max Alekseyev Jul 21 at 2:50
  • $\begingroup$ Also, for integer parameters $\nu,\mu$, associated Legendre function $P_\nu^\mu(x)$ turns into the associated Legendre polynomial. However, such polynomial is zero when $\mu>\nu$. So, please clarify the meaning of $P_\nu^\mu(x)$ in your question. $\endgroup$ – Max Alekseyev Jul 21 at 3:27
  • 1
    $\begingroup$ The measure seems odd - $\sin \theta d\theta $ would look a lot more natural. It might be useful to have more details on how this integral comes about. $\endgroup$ – Michael Engelhardt Jul 24 at 17:30
1
$\begingroup$

The best evaluation, a single sum, that I can derive is

$$ \int_0^{\pi/2} \big( P_n^m(\cos{\theta}) \big)^2\ d\theta = \frac{\pi}{2} \frac{(2m)!}{m!^4} \Big( \frac{(n+m)!}{(n-m)!} \Big)^2 {}_4F_3 \bigl( \begin{smallmatrix} m+1/2, & m+1/2, & m-n, &m+n+1 \\ m+1 & m+1 & 2m+1 \end{smallmatrix} | 1\bigr) $$

I derived it from the answer given in Math Overflow 291481, which expresses the product of the associated Legendre polynomials as a single sum with powers of $\sin^2{\theta}$ within the summand. The integration is then easy. I then simplified the formula to that above.

It is doubtful that the generalized hypergeometric ${}_4F_3$ simplifies to a ratio of gamma functions. I have a reference that has some ${}_4F_3$ evaluated at 1, but the 'numerator' parameters start off as $a, 1+a/2...$ and that's not the form of the answer. Furthermore, I calculated it a few for small $m$ and $n,$ and if a ratio of gammas was in fact true, I wouldn't expect to get large primes in my answer.

| cite | improve this answer | |
$\endgroup$

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.