On trigonometric n-widths and their generalization
暂无分享,去创建一个
is called the n-width of Q in X (in the sense of Kolmogorov). The left infimum in (1) is taken over all n-dimensional linear subspaces r,, c X. One can obtain various modifications of this definition by taking the inlimum over special classes of i;, . For instance, in X = L,[O, 27r] one may consider subspaces r,, spanned by any tz of the functions {exp(ik . )}, k E Z. If the infimum in (1) is taken over all such subspaces, the corresponding n-width, introduced by Ismagilov [I], is called the trigonometric n-width, d,T(R, L,). It is obvious that d,T > d,, but if the class 0 is translation-invariant, one would expect that dc = d,. And indeed, in all cases for which dz has been estimated, d,T d,(n + 03). In a more general setting, let G be a compact Abelian group with the invariant measure ,u, p(G) = 1. In X = L, = L,(G, p) we consider the subspaces r, of the form
[1] Klaus Höllig,et al. Approximationszahlen von Sobolev-Einbettungen , 1979 .
[2] R. S. Ismagilov,et al. DIAMETERS OF SETS IN NORMED LINEAR SPACES AND THE APPROXIMATION OF FUNCTIONS BY TRIGONOMETRIC POLYNOMIALS , 1974 .
[3] V. E. Maiorov,et al. ON LINEAR WIDTHS OF SOBOLEV CLASSES AND CHAINS OF EXTREMAL SUBSPACES , 1982 .
[4] H. Rosenthal. On the subspaces ofLp(p>2) spanned by sequences of independent random variables , 1970 .