Riemann--Hilbert analysis for Jacobi polynomials orthogonal on a single contour (Q557550): Difference between revisions

Aproximately 10 years ago an important breakthrough was made in the construction of tools and methods in the study of asymptotics for orthogonal polynomials with varying weights (cf. [\textit{P. Deift} and \textit{X. Zhou}, Ann. Math. 137, 295--368 (1993; Zbl 0771.35042)]. Reformulating matters in the form of a Riemann-Hilbert problem, several interesting results have been established lately (compare amongst others [\textit{A. B. J. Kuijlaars} and \textit{A. Martinez-Finkelshtein}, J. d'Analyse Math. 94, 195--234 (2004; Zbl 1126.33003)]). Using the new method the authors derive important results on th easymptotic behavior of Jacobi polynomials orthogonal on a contour in the complex plane. In the paper under review the authors consider Jacobi polynomials \[ P^{(\alpha_n,\beta_n)}(z) = 2^{-n}\, \sum_{k=0}^n \binom {n+\alpha_n} {n-k} \binom {n+\beta_n} {k}(z-1)^k(z+1)^{n-k}, \] with varying parameters \(\alpha_n,\,\beta_n\in\mathbf{R}\) satisfying \[ \lim_{n\rightarrow\infty} {\alpha_n\over n}=A,\;\lim_{n\rightarrow\infty} {\beta_n\over n}=B \] with \[ A<0<B,\quad A+B>-1,\quad A\not= -1.\tag \(*\) \] Without going into details concerning auxiliary functions and intermediate results (the interested reader should consult the paper), the main resluts are stated below. 1. If \((A,B)\) satisfies (\(\ast\)), the the monic Jacobi polynomials \(p_n={\widehat P}^{(An,Bn)}_n\) satisfy a) Locally uniformly on \textbf{C}: \[ p_n(z)=\left({G(z)\over \kappa}\right)^n N_{11}(z) \left( 1+{\mathcal O}\left({1\over n}\right)\right). \] b) Explicit formula for the asymptotic behavior away from the cases C2 (\(A<1\), \(A+B>-1\)) and C3 (\(-1<A<0\), \(B>0\)) are given. 2. For a sequence of Jacobi polynomials \((\alpha-n,\beta_n)\) with \((A,B)\) satisfying (\(\ast\)): \textbf{a)} If \((A,B)\) moreover satisfies C2, then the zeros accumulate on an arc \(\Gamma\) and the weak\(*{\ast}\)-limit of the normalized zero counting measures is just \[ d\mu(z)={A+B+2\over 2\pi i}\, {\sqrt{(z-\zeta_1)(z-\zeta_2)}\over 1-z^2}\,dz,\quad z\in\Sigma. \] b) For \((A,B)\) moreover satisfying C3, the statement of 2a holds with \(\Gamma\) and \(\mu\) replced by \(\Gamma_r\cup [\zeta_1,\zeta_2]\) and \(\mu_r\) with \[ d\mu_r(z)={A+B+2\over 2\pi i}\, {\sqrt{(z-\zeta_1)(z-\zeta_2)}\over 1-z^2}\,dz,\quad z\in\Gamma_r \cup (\zeta_1, \zeta_2). \] Here \[ \zeta_{1,2}={B^2-A^2\pm 4\sqrt{(A+1)(B+1)(A+B+1)}\over (A+B+2)^2}, \] and \(\Gamma,\Gamma_r\) arise as critical trajectories of quadratic differentials.