\(L_p\) Markov exponent of certain domains with cusps (Q6558233)

Definition 1.1. We say that a compact set \(\emptyset\neq E\subset\mathbb{R}^{m}\) satisfies \(L_{p}\) Markov type inequality (or: is a \(L_{p}\) Markov set) if there exist \(k\), \(C>0\) such that, for each polynomial \(P\in\wp(\mathbb{R}^{m})\) and each \(\alpha\in\mathbb{N}_0^{m}\),\N\[\N\Vert D^\alpha P\Vert_{L_{p}(E)}\leq(C(\deg P)^{k})^{|\alpha}|\Vert P\Vert_{L_{p}(E)},\tag{1}\N\]\Nwhere \(D^\alpha P=((\partial^{|\alpha|}P)/(\partial x^{\alpha_1}\dots \partial x^{\alpha_{m}}))\) and \(|\alpha|=\alpha_1+\cdots+\alpha_{m}\). In this paper, the author studies the following problem: For a given \(L_{p}\) Markov set \(E\), determine \(\mu_{p}(E):=\inf k:E\) satisfies (1). The quantity \(\mu_{p}(E)\) is called the \(L_{p}\) Markov exponent and was first considered by \textit{M. Baran} and \textit{W. Pleśniak} [Proc. Am. Math. Soc. 123, No. 9, 2785--2791 (1995; Zbl 0841.41011)] for \(p=\infty\). The origin and history of the development of this problem are given in [\textit{M. Baran} and \textit{W. Pleśniak}, Proc. Am. Math. Soc. 123, No. 9, 2785--2791 (1995; Zbl 0841.41011); \textit{P. Goetgheluck}, J. Approx. Theory 49, 303--310 (1987; Zbl 0643.41011); J. Approx. Theory 30, 149--154 (1980; Zbl 0457.41015); \textit{G. Mastroianni} and \textit{V. Totik}, Constr. Approx. 16, No. 1, 37--71 (2000; Zbl 0956.42001); \textit{A. Kroó} and \textit{J. Szabados}, J. Approx. Theory 102, No. 1, 72--95 (2000; Zbl 0953.41008); \textit{A. Kroó}, J. Approx. Theory 250, Article ID 105336, 6 p. (2020; Zbl 1431.41005); \textit{W. Pawłucki} and \textit{W. Pleśniak}, Math. Ann. 275, 467--480 (1986; Zbl 0579.32020); \textit{W. Pleśniak}, J. Approx. Theory 61, No. 1, 106--117 (1990; Zbl 0702.41023)]. In this paper, the main result is the establishment of the Markov exponent \(L_{p}\) for given ``special'' domains \(\Psi_{k}\), \(Y_{k}\) and \(\Lambda_{k}\), are defined as following:\N\[\N\Psi_{k}:=\{(x,y)\in\mathbb{R}^2:|x|\leq 1,0\leq y\leq x^{2k}\};\N\]\N\[\NY_{k}:=\{(x,y)\in\mathbb{R}^2:|x|\leq 1,0\leq y\cdot\operatorname{sgn} x\leq |x|^{2k}+1\};\N\]\N\[\N\Lambda_{k}:=\{(x,y)\in\mathbb{R}^2:0\leq x\leq 1,ax^{k}\leq y\leq bx^{k}\},k\in N,0<a<b.\N\]\NIt is proved that \(\mu_{p}(\Psi_{k})=\mu_{p}(\Lambda_{k})=2k\) and \(\mu_{p}(Y_{k})=2k+1\) for any \(k\in N\), \(1\leq p<\infty\).