Multivariate Markov-type and Nikolskii-type inequalities for polynomials associated with downward closed multi-index sets (Q476136): Difference between revisions

Let \(\Lambda \subset \mathbb{N}^d_0\) be a finite multi-index set. This \(\Lambda\) is said to be downward closed if (\(\nu \in \Lambda\) and \(\mu \leq \nu\))\(\Rightarrow \mu \in \Lambda\). Let \(p(n):=\sum_{l=0}^\eta \alpha_l n^l\), where \(n\in \mathbb{N}_0\) and \(\alpha_l\in\mathbb{R}\). The following result is proved: { Theorem 1.} For any downward closed multi-index set \(\Lambda\) and for any \(\eta \in \mathbb{N}_0\), if the \(\eta +1\) coefficients of the polynomial \(p\) satisfy \(\alpha_l \leq \binom{\eta+1}{l}\) for any \(l=0,\ldots,\eta\), then \(K_p(\Lambda)\leq (\# \Lambda)^{\eta +1}\), where \[ K_p(\Lambda):=\sum_{\nu \in \Lambda}\prod_{q=1}^d p(\nu_q). \] Multivariate Markov-type inequalities are derived from Theorem 1. One of these results is the following. {Theorem 2.} Let \(\Lambda\) be downward closed and let \(D=[-1,1]^d\). For any \(d\)-variate polynomial \(u \in \mathbb{P}_\Lambda(D)\) it holds that \[ \left\|\frac{\partial^d}{\partial y_1 \cdots \partial y_d}u\right\|_{L^2(D)}\leq \left(\frac{3}{5}\right)^{d/2}(\#\Lambda)^{5/2}\|u\|_{L^2(D)}. \] The remaining results correspond to \(L^2_\rho\) spaces, where \(\rho\) is a weight function associated with various tensorized orthonormal polynomials, namely Legendre, Chebyshev, Gegenbauer and Jacobi polynomials. Furthermore, they are derived multivariate Nikolskii-type \(L^\infty -L_\rho^2\) inequalities over multivariate polynomial spaces associated with arbitrary downward closed multi-index sets in any dimension.