Volterra method for the radial Schrödinger equation (Q1877177)

This paper is concerned with the numerical solution of the radial Schrö\-din\-ger equation: \[ [ d^2/dr^2 + l(l+1)/ r^2 + V(r)] R(r) = k^2 R(r), \;r \in (0, \infty) \] with the boundary conditions \( R(0)=0\), \(\lim_{r \to \infty} (R(r) - \sin (kr -l \pi/2)- w \exp ( ikr -i l \pi/2))=0.\) Here \(l,k\) and \(w\) are given constants with a physical significance in the context of Quantum Mechanics and \( V(r)\) the given potential function. For the numerical solution of the above boundary value problem the author starts considering an equivalent Volterra integral equation formulation in some interval \( [0,T]\) with \(T\) sufficiently large. Further, to derive the discrete approximation this interval is partitioned into smaller subintervals \( [0,b_1],[b_1,b_2], \ldots ,[b_{m-1},b_m=T]\) and in each sub interval a composite Clenshaw-Curtis quadrature rule is used to approximate the integrals at Chebyshev nodes. The results of some numerical experiments are presented testing the behaviour of the new method with the standard finite difference method based on the fourth order Numerov algorithm [cf. \textit{J. Stoer} and \textit{R. Bulirsch}, Introduction to numerical analysis (1993; Zbl 0771.65002)] and also with a variable step size finite difference method of order six due to \textit{A. D. Reptis} and \textit{J. R. Cash} [Comput. Phys. Commun. 36, 113--119 (1985; Zbl 0578.65086)]. The results of these experiments show that the proposed method is able to achieve higher accuracy than Reptis-Cash and Numerov's methods.