Estimates of the Hermite and the Freud polynomials (Q753068): Difference between revisions

Using a new approach to the derivation of certain estimates for the Hermite polynomials, the authors extend these results to the case of the Freud polynomials \(p_ n(x)\) (orthonormal on the real axis with respect to the weight function \(\exp (-x^ m),\) where m is a positive even integer; the leading coefficient \(\gamma_ n\) is taken to be positive). After deriving an approximate second order differential equation for these polynomials, the authors give their main result: the existence of positive constants C, D, E with \[ p^ 2_ n(x)\exp (-x^ m)\leq C/\sqrt{(2\beta n^{1/m})^ 2-x^ 2}\quad when\quad | x| \leq 2\beta n^{1/m}, \] \[ Dn^{1/3-1/m}\leq \max_{x\in {\mathbb{R}}}p^ 2_ n(x)\exp (-x^ m)\leq En^{1/3-1/m}. \] Here \(\beta\) is Freud's constant \(\{\pi^{1/2}\Gamma (m/2)/\Gamma (m+1/2)\}^{1/m}\); the proof uses the first term of a deep result on the asymptotic behaviour of \(\gamma_{n- 1}/\gamma_ n\) given by Máté, Nevai and Zaslavsky: \[ \gamma_{n- 1}/\gamma_ n=\beta n^{1/m}+\epsilon_ n,\quad | \epsilon_ n| \leq Hn^{-1+1/m}\quad (n=1,2,...;\quad H>0). \] Quite an interesting paper.