Weighted polynomial inequalities with doubling and \(A_\infty\) weights (Q1968777): Difference between revisions

In this landmark paper, the authors show that much of the approximation theory that was first developed in the unweighted case, then for Jacobi weights and then for generalized Jacobi weights, can actually be developed for doubling weights and/or weights satisfying Muckenhoupt's \(A_p\) condition. Recall that \(W: [-\pi,\pi]\) is said to be a doubling weight if there exists \(L>0\) such that for all \(a\in [-\pi,\pi]\) and \(\delta>0\), \[ \int^{a+ 2\delta}_{a-2\delta} W\leq L\int^{a+ 2\delta}_{a- 2\delta}W. \] They show that if \(W\) is a doubling weight, the weighted Bernstein inequality holds: let \(1\leq p<\infty\). There exists \(C>0\) such that for \(n\geq 1\), and all trigonometric polynomials \(T\) of degree \(\leq n\), \[ \int|T'|^p W\leq Cn\int|T|^p W. \] They also establish weighted analogues of (weak) Marcinkiewicz inequalities, Remez inequalities, restricted range inequalities, estimates for Christoffel functions, and other important inequalities. This paper will change the focus of weighted approximation on finite intervals, and is essential reading for anyone (even vaguely) interested in that topic.