Jackson-type theorems on some transcendental curves in \(\mathbb R^n\) (Q705255): Difference between revisions

Let \(\gamma(t)\), \(t \in [a,b]\) be a continuous curve of finite length in \({\mathbb R}^d\), \(d \geq 2\), and denote by \(C_\gamma\) the space of continuous functions on \(\gamma\), with the supremum norm. A function \(f \in C_\gamma\) is called a contraction if \(| f(\gamma(u))- f(\gamma(v))| \leq | u-v| \) for all \(u,v \in [a,b]\). The set of all contractions is denoted by \(K_\gamma\), \(P_n^d\) is the space of real algebraic polynomials of \(d\) variables and degree \(\leq n\) in each variable, and \[ E_n(\gamma):=\sup_{f \in K_\gamma}\inf_{p \in P_n^d} \| f -p\| . \] It follows from a general result of Lorentz that \(E_n(\gamma) \geq c/r_n(\gamma)\) for some \(c>0\), where \(r_n(\gamma)\) is the dimension of the restriction of \(P_n^d\) to \(\gamma\). The curve \(\gamma\) is called efficient if this lower bound can be matched by an upper bound of the same order. D. J. Newman and L. Raymon proved in 1969 that algebraic curves are efficient, but claimed that the exponential curve \(\gamma:=\{t, e^t\}\), \(0 \leq t \leq 1\), is not. In the present paper, the authors show that the latter claim is incorrect, that is, the exponential curve (as well as its generalizations to higher dimensions) is indeed efficient, and so are some other curves erroneously claimed to be not.