Uniqueness of the Gaussian quadrature for a ball (Q1567421): Difference between revisions

In a previous paper the authors [Numer. Math. 80, No. 1, 39-59 (1998; Zbl 0911.65015)] investigated quadrature formulae for the unit disk \(D=\{(x,y); x^2+y^2\leq 1\}\) which are based on integrals over \(n\) chords. Among all variety of quadrature formulae that use \(n\) chords, the following result was established: ``The Gaussian quadrature formula \[ \iint_D f(x,y)dx dy \approx \sum^n_{k=1} A_k\int^{+\sqrt {1-\eta_k^2}}_{-\sqrt{1- \eta_k^2}}f (\eta_k,y)dy \tag{*} \] where \(A_k= {\pi\over n+1}\sin {k\pi\over n+1}\) \((k=1,\dots,n)\) has a highest degree of precision with respect to the set of algebraic polynomials of two variables of total degree \(2n-1\). Here \(\eta_k\), \(k=1,\dots,n\), stand for the well-known zeros of the Chebychev polynomial of second kind of degree \(n\)''. In the present paper they prove the uniqueness (up to rotation) of the quadrature formula (*) with this extremal property. The authors give also a general multivariate analogue of (*) for numerical weighted integration of functions over the unit ball \(B^d\) in \(\mathbb{R}^d\) using integrals over the intersection of \(B^d\) with \(n\) hyperplanes \((n\) Radon projections of these functions likely as used in computer tomography). The uniqueness of this formula is proved when taking the weight equal to 1.