Geno P. Nikolov

A discrete weighted Markov--Bernstein inequality for polynomials and sequences

For parameters $\,c\in(0,1)\,$ and $\,β>0$, let $\,\ell_{2}(c,β)\,$ be the Hilbert space of real functions defined on $\,\mathbb{N}\,$ (i.e., real sequences), for which $\| f \|_{c,β}^2 := \sum_{k=0}^{\infty}\frac{(β)_k}{k!}\,c^k\,[f(k)]^2<\infty\,.$ We study the best (i.e., the smallest possible) constant $\,γ_n(c,β)\,$ in the discrete Markov-Bernstein inequality $\|ΔP\|_{c,β}\leq γ_n(c,β)\,\|P\|_{c,β}\,,\quad P\in\mathcal{P}_n\,,$ where $\,\mathcal{P}_n\,$ is the set of real algebraic polynomials of degree at most $\,n\,$ and $\,Δf(x):=f(x+1)-f(x)\,$. We prove that: (i) $\displaystyle γ_n(c,1)\leq 1+\frac{1}{\sqrt{c}}\,$ for every $\,n\in \mathbb{N}\,$ and $\displaystyle \lim_{n\to\infty}γ_n(c,1)= 1+\frac{1}{\sqrt{c}}\,$. (ii) For every fixed $\,c\in (0,1)\,$, $\,γ_n(c,β)\,$ is a monotonically decreasing function of $\,β\,$ in $\,(0,\infty)\,$. (iii) For every fixed $\,c\in (0,1)\,$ and $\,β>0\,$, the best Markov-Bernstein constants $\,γ_n(c,β)\,$ are bounded uniformly with respect to $\,n$. A similar Markov-Bernstein unequality is proved for sequences in $\,\ell_{2}(c,β)\,$. We also establish a relation between the best Markov-Bernstein constants $\,γ_n(c,β)\,$ and the smallest eigenvalues of certain explicitly given Jacobi matrices.