Uniqueness of Markov-extremal polynomials on symmetric convex bodies (Q5945739)

If \(K\) is a compact subset of \(\mathbb R^d\), the n\(th\) Markov constant of \(K\) is defined by \[ M_n(K):=\sup\{|||\text{grad} p|||_{C(K)}/\|p\|_{C(K)}:\;p\in\pi_n^d\}, \] where \(\pi_n^d\) denotes the set of all polynomials in \(d\) variables of total degree \(\leq n\). It is known by \textit{Y. Sarantopoulos} [Math. Proc. Camb. Philos. Soc. 110, 307-312 (1991; Zbl 0761.46035)], \textit{M. Baran} [J. Approximation Theory 79, No. 2, 190-198 (1994; Zbl 0819.41013)] that if \(K\) is a symmetric convex body, then \(M_n(K)=n^2/r(K)\), where \(r(K):=\inf\{|y|:\;y\in\partial K\}=\sup\{r:\;B(0,r)\subset K\}\). If \(a\) is a boundary point of \(K\) such that \(|a|=r(K)\) then one can easily check that the polynomial \(g_{n,a}(x):=T_n(<x,a>/r^2(K))\), where \(T_n\) is the usual Chebyshev polynomial, is extremal for the Markov constant \(M_n(K)\). On the other hand, if \(K\) is a symmetric convex body in \(\mathbb R^2\), \textit{A. Kroó} [J. Math. Anal. Appl. 253, No. 1, 322-333 (2001; Zbl 0971.41018)] has proved that in order that every extremal polynomial for the Markov constant \(M_n(K)\) coincide with \(cT_n(\langle x,a\rangle/r^2(K))\) for some \(a\in K^0:=B(0,r(K))\cap\partial K\) and \(c\in\mathbb R\) it is necessary and sufficient that each \(a\in K^0\) be flat (\textit{i.e.} no disk touching the tangential line \(L_a\) of \(K\) at \(a\) contains \(K\)). In this paper, the author extends Kroó's result to arbitrary dimension. To do this, he finds the corresponding flatness condition for the boundary points of \(K\) called the ``parabolic flatness''.