Asymptotic approximations of Jacobi functions and their zeros with error bounds (Q2508595)

This paper reports on an asymptotic approximation for the Jacobi function \(\varphi_\mu^{(\alpha,\beta)}(t)\) for \(\mu\to+\infty\), when \(\alpha\) and \(\beta\) are real (with \(\alpha>-1)\). The method adopted is the use a comparison differential equation, soluble in terms of Bessel functions, which possesses a similar structure to that satisfied by \(\varphi_\mu^{(\alpha,\beta)}(t)\), combined with an integral equation approach. In this way the authors establish the result \[ (\sinh t)^{\alpha +\frac 12}(\cosh t)^{\beta+\frac 12}\varphi_\mu^{ (\alpha,\beta)}(t)= \frac{c_0}{\mu^\alpha}\left(\frac{x}{x'}\right)^{1/2}J_\alpha(x)+I(t), \] where \(J_\alpha\) denotes the Bessel function, \(c_0\) is a given constant. The variable \(x=x(t)\) is given by an expansion of the form \[ x(t)=\mu t+\sum^{m-1}_{k=0}f_{2k+1}(t)\mu^{-2k-1} \] with \(f_{2k+1}(t)\) are computable coefficients. A simple explicit bound for the error \(I(t)\) for \(\mu\geq\mu_0\), where \(\mu_0\) is a specified value, is derived in the cases \(m=1\) and \(m=2\). The above result is then employed to obtain an approximation for the zeros \(t_{\mu, s}(s=1,2, \dots)\) of \(\varphi_\mu^{(\alpha,\beta)}(t)\) \((\alpha\geq 0)\) as \(\mu\to +\infty\) in the form \[ t_{\mu,s}=\frac{j_{\alpha,s}}{\mu}+ \frac{A} {\mu^2}+\frac {B}{\mu^4}+\delta_s. \] where \(j_{\alpha,s}\) is the \(s\)th zero of the Bessel function \(J_\alpha\), \(A\) and \(B\) are given coefficients and a simple bound is derived for \(|\delta_s|\). No numerical results are given to demonstrate the accuracy of these estimates.