The Markov brothers inequality in \(L_0\)-space on an interval (Q2577344)

From MaRDI portal
scientific article
Language Label Description Also known as
English
The Markov brothers inequality in \(L_0\)-space on an interval
scientific article

    Statements

    The Markov brothers inequality in \(L_0\)-space on an interval (English)
    0 references
    19 December 2005
    0 references
    Let \(\mathcal P_n=\{P\in \mathbb C[x]:\;\text{deg}\,P=n\}\). For \(P\in\mathcal P_n\), we set \[ \|P\|_q=\left(\frac12\int_{-1}^1|P(t)|^qdt\right)^{1/q},\;\;0<q<\infty,\quad\|P\|_{\infty}=\max_{-1\leq t\leq1}|P(t)|, \] \[ \|P\|_0=\lim_{q\to +0}\|P\|_q=\exp\left(\frac12\int_{-1}^1\ln|P(t)|dt\right). \] Denote by \(M_q(n,k)\) the best constant in the \(L_q\) version of the well-known Markov brothers inequality \[ \|P^{(k)}\|_q\leq M_q(n,k)\|P\|_q,\quad P\in\mathcal P_n,\;0\leq q\leq\infty. \] In the paper, the author recalls the crucial results concerning the asymptotic values of \(M_q(n,k)\). They show that \[ M_q(n,k)\asymp n^{2k},\quad n\to\infty, \] for fixed \(q\in(0,\infty]\) and \(1\leq k\leq n\). Then the author studies the limit case \(q=0\) and shows the following assertion: for any integer \(0\leq k\leq n\), the polynomials \(P\in\mathcal P_n\) satisfy the inequality \[ \|P^{(k)}\|_0\leq\frac{n!e^k}{(n-k)!}\|P\|_0, \] where the equality occurs if and only if \(P(t)=ct^n,\;c\in\mathbb C\). In particular, in contrast to the case \(q>0\), it follows that \(M_0(n,k)\) grows as \(n^k\), if \(n\to\infty\).
    0 references
    0 references
    0 references
    0 references
    0 references
    Markov inequality
    0 references
    algebraic polynomial
    0 references
    uniform norm
    0 references
    Chebyshev polynomial
    0 references
    \(L_{q}\)-metric
    0 references
    0 references