Bounds and inequalities for general orthogonal polynomials on finite intervals (Q1801566)

Quite often one finds upper bounds for orthogonal polynomials and integrals of polynomials. This paper gives some nontrivial lower bounds which are sharp, useful and beautiful. To get a taste, take a look at the following lower bound \[ {\gamma_{n-1} \over \gamma_ n} \sum_{k=1}^ n \lambda_{k,n} | p_{n-1}(\alpha,x_{k,n}) | \geq{(2-\delta) \sigma(\delta) \over 24}, \] where \(p_ n(\alpha,x)\) are orthogonal polynomials for a probability measure \(\alpha\) on \([-1,1]\) with leading coefficients \(\gamma_ n\), Christoffel numbers \(\lambda_{k,n}\) and zeros \(x_{k,n}\), and \(\sigma(\delta)\) measures the thickness of the measure \(\alpha\) by \[ \sigma(\delta)=\inf_{A \subset[-1,1],| A | = \delta} \alpha(A), \] the infimum taken over all Lebesgue measurable sets of Lebesgue measure \(\delta\). Such inequalities are also obtained for \(L_ p\) norms of orthogonal polynomials, e.g., lower bounds for \[ \int_{-1}^ 1| p_ n(\alpha,x)|^ p\;d \beta(x), \] where \(\beta\) is an arbitrary measure on \([-1,1]\). These inequalities are very useful to obtain information on the smoothness or absolute continuity of the orthogonality measure \(\alpha\) and to obtain relevant asymptotic properties of the Christoffel function \(\lambda_ n(\alpha,x)\), which in turn leads to important results on the convergence of orthogonal Fourier series. Some problems on Lagrange interpolation posed by Turán are taken as an example to show that the obtained inequalities come in very handy. Finally, the inequalities are also applied to Hermite-Fejér interpolation and to an \(L_ 2\) version of the principle of contamination. Quite a remarkable paper.