A note on the zero distribution of orthogonal polynomials (Q2640824)

A sequence of orthonormal polynomials \(p_ n(x)\) satisfies a three-term recurrence relation \(xp_ n(x)=a_{n+1}p_{n+1}(x)+b_ np_ n(x)+a_ np_{n-1}(x).\) If another sequence of orthonormal polynomials \(q_ n(x)\) is given with recurrence coefficients \(a^ 0_ n\) and \(b^ 0_ n\) such that the Cesàro means of \(| a_ n-a^ 0_ n|\) and \(| b_ n-b^ 0_ n|\) tend to zero (it is assumed that all the recurrence coefficients are bounded), then it is shown that the zeros of \(p_ n(x)\) and \(q_ n(x)\) are equally distributed. This result was already given by \textit{J. S. Geronimo}, \textit{E. M. Harrell II} and the reviewer [Constructive Approximation 4, 403-417 (1988; Zbl 0646.15004)] but the present proof is in terms of estimating the mean of the entries of \(f(J_ 1)-f(J_ 2)\) along subdiagonals, where \(J_ 1\) and \(J_ 2\) are the infinite Jacobi matrices associated with \(p_ n(x)\) and \(q_ n(x)\), respectively, and f is a continuous function on the union of the supports of the two orthogonality measures.