Normalization integrals of orthogonal Heun functions

From MaRDI portal



Abstract: A formula for evaluating the quadratic normalization integrals of orthogonal Heun functions over the real interval 0 <= x <= 1 is derived using a simple limiting procedure based upon the associated differential equation. The resulting expression gives the value of the normalization integral explicitly in terms of the local power-series solutions about x=0 and x=1 and their derivatives. This provides an extremely efficient alternative to numerical integration for the development of an orthonormal basis using Heun functions, because all of the required information is available as a by-product of the search for the eigenvalues of the differential equation. 02.30.Gp; 02.30.Hq; 02.70.-c; 02.60.Jh


Any second-order Fuchsian equation with four singular points is equivalent to Heun's equation \[ {d^2y\over dx^2}+ \Biggl({\gamma\over x}+ {\delta\over x-1}+{\varepsilon\over x-a}\Biggr) {dy\over dx}+ {\alpha\beta x-\lambda\over x(x-1)(x- a)} y=0,\tag{1} \] represented by the Riemann \(P\)-symbol \[ P\left\{\begin{matrix} 0 & 1 & a & \infty\\ 0 & 0 & 0 & \alpha & x\\ 1-\gamma & 1-\delta & 1-\varepsilon & \beta\end{matrix}\right\}, \] \(\lambda\) is a so-called accessory parameter. Suppose \(a\not\in [0,1]\). We consider four classes of boundary value problems on \([0,1]\), \(\lambda\) being an eigenvalue. \hskip17mm I:\quad (1), \(y\) is smooth at \(0\), \(y\) is smooth at \(1\). \hskip17mm \(\cdots\) \hskip17mm IV:\quad (1), \(x^{\gamma- 1}y\) is smooth at \(0\), \((1-x)^{\delta-1}y\) is smooth at \(1\). Let us consider, for example, class I. Let \(y_0(\lambda, x)\) be the solution of (1) satisfying \(y_0(\lambda, 0)= 1\), and \(y_1(\lambda, x)\) be the solution of (1) satisfying \(y_1(\lambda, 1)= 1\). Problem. For an eigenvalue \(\lambda_n\) of the BVP I, estimate the norm \[ N_n:= \int^1_0\omega(x)[H_n(x)]^2 dx, \] where \(H_n(x):= y_0(\lambda_n, x)\), \(\omega= x^{\gamma- 1}(x- 1)^{\delta-1}(x- a)^{\varepsilon- 1}\). Theorem. \[ N_n= -p(x) {\partial W\over\partial\lambda} (\lambda_n, x) {y_0(\lambda_n, x)\over y_1(\lambda_n, x)}, \] where \(p(x)= x^\gamma(x- 1)^\delta(x- a)^\varepsilon\), and \(W\) is the Wronskian of \(y_0\) and \(y_1\). Note that \(p(\partial w/\partial\lambda)\) and \(y_0/y_1\) are independent of \(x\). These quantities are already used during the evaluation-algorithm of the eigenvalue \(\lambda_n\). Thus the theorem tells that the evaluation of the norm \(N_n\) can be obtained as a by-product of the search for the eigenvalues, and so that this formula greatly improves the efficiency of numerical procedures involving Heun functions.











This page was built for publication: Normalization integrals of orthogonal Heun functions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4352188)