Dear mathematicians and theoretical physicists,

I am a theoretical physicist and I am bothering to you since I need to know some asymptotic and analytical properties of Jacobi polynomials $P^{(A,B)}_n(z)$ (where the discrete index $n$ is related to the angular momentum and, in "physical variables", $z=\cos\Theta$) in the complex plane as well as for large $n$.

Before writing down my two questions ($Q_1$ and $Q_2$), let me explain a simpler case in which the answer is known for the Legendre polynomial $P_n(z)$. Namely, in a simpler (rather standard) problem it would enter the Legendre function of order $n$: $P_n(z)$ where, in Physical coordinates, $z=\cos(\Theta)$ and $n$ is the "orbital angular momentum". Two properties which are of great help in the standard case are bounds on the Legendre function when it is analytically continued to the complex $z$-plane and when $n$ is very large. The properties which are very useful are:

  • The first property has to do with a bound on the Legendre function as order $n \to \infty$): $$| P_n(\cos \Theta) | < n^{-1/2}e^{ n | \text{Im }\Theta | }$$ where $| x |$ is the absolute value of $x$. So when $n$ is very large, one gets an exponential bound in terms of the order $n$ of the Jacobi polynomial and the (absolute value of) the imaginary part of $\Theta$ (as well as the square root of 1/$n$).

  • The second has to do with the behavior of $P_n(z)$ as $|z|\to\infty$: $P_n(z) \sim c_n z^n$

I would really need similar properties for $P^{(-q-p,-q+p)}_n(z)$ where $p$ is an integer and $q$ can be both integer and half-integer (the most important situation for me being $q$=1/2 ). Namely:

  • Is there a bound, when $n \to \infty$ , on $P^{(-q-p,-q+p)}_n(z)$ such as $$| P^{(-q-p,-q+p)}_n(\cos \Theta) | < n^{-1/2}e^{n|\text{Im }\Theta|} ?$$
  • Is there a bound on $P^{(-q-p,-q+p)}_n(z)$ as $|z|\to\infty$ such as $P^{(-q-p,-q+p)}_n(z) \sim c^{(p,q)}_n z^n$ ? If yes, how does $c^{(p,q)}_n$ behave asymptotically?

I would really appreciate if you could help me even with some references (which a physicist is able to understand).

  • $\begingroup$ Fabrizio: I've edited your question submission so that it displays well on this site. In doing so, I wrote your notation of Legendre polynomials and Jacobi polynomials to match references at Wikipedia. Please examine the changes and let me know if I've erred somewhere. $\endgroup$ – Semiclassical Jul 24 '14 at 2:25
  • $\begingroup$ The English version of Wikipedia has some material on the Jacobi polynomials, in particular a section on their asymptotics. $\endgroup$ – Semiclassical Jul 24 '14 at 2:46
  • $\begingroup$ See also the page for the Mehler-Heine formula which discusses asymptotic bounds on classical orthogonal polynomials including Legendere and Jacobi. $\endgroup$ – Semiclassical Jul 24 '14 at 2:56

For your second question, by the second formula under "Definitions" on the wikipedia page, the term with the largest power $n$ in $P_n^{(\alpha,\beta)}(z)$ is

$$ \frac{\Gamma(\alpha + \beta + 2n + 1)}{n! \Gamma(\alpha + \beta + n + 1)} \left(\frac{z-1}{2}\right)^n \sim \frac{\Gamma(\alpha + \beta + 2n + 1)}{2^n n! \Gamma(\alpha + \beta + n + 1)} z^n $$

Taking $\alpha = -q-p$ and $\beta = -q+p$ as in your question, we obtain

$$ P_n^{(-q-p,-q+p)}(z) \sim \frac{\Gamma(-2q + 2n + 1)}{2^n n! \Gamma(-2q + n + 1)} z^n $$

as $|z| \to \infty$. Stirling's formula tells us that the coefficients behave asymptotically like

$$ \frac{\Gamma(-2q + 2n + 1)}{2^n n!\Gamma(-2q + n + 1)} \sim 2^{n-2q} \sqrt{\frac{1}{\pi n}} $$ as $n \to \infty$.


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.