A bivariate Markov inequality for Chebyshev polynomials of the second kind (Q655447): Difference between revisions

The main result is a Markov-type inequality for polynomials in two variables where the Chebyshev polynomials of the second kind \(U_m\) in either one of the variables are extremal. Let \(\mathcal N_k\) be the set of ordered pairs \((h_n,h_q),\quad 0\leq n,q\leq m,\) where \(n\) and \(q\) are both even or both odd for even \(k\) and of different parity for odd \(k,\) and \[ h_n=\cos\frac{(n+1)\pi}{m+2},\quad 0\leq n\leq m. \] If \(|p(x)|\leq 1\) whenever \(x\in\mathcal N_k,\) then \[ |\hat D^kp(r,r)(1,-1)|\leq U_m^{(k)}(r) \] for \(r\geq r_k,\) where \[ r_k=\cos\frac{\pi}{(l+1)(m+2)},l\equiv k\mod 2, l\in\{0,1\}. \] Here equality holds when \(p(s,t)=U_m(s),\) and \[ \hat D^kp(x)y=\left.\frac{d^k}{dt^k}p(x+ty)\right|_{t=0}. \] To prove the result the author constructs Lagrange polynomials for the class of nodes considered here and gives a corresponding Christoffel-Darboux formula.