Approximation of ultra-differentiable functions by polynomials and entire functions (Q1063832)

From MaRDI portal
Revision as of 10:21, 30 July 2024 by Openalex240730090724 (talk | contribs) (Set OpenAlex properties.)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
scientific article
Language Label Description Also known as
English
Approximation of ultra-differentiable functions by polynomials and entire functions
scientific article

    Statements

    Approximation of ultra-differentiable functions by polynomials and entire functions (English)
    0 references
    1984
    0 references
    The article deals with the germ of a \(C^{\infty}\)-function f on a compact set \(K\subset {\mathbb{R}}^ n\). On the one hand we consider the speed at which \[ d_{n,p,v}(f):=\inf \{| f-p|_{L^ p(V)};\quad p\text{ a polynomial of degree }\leq n\} \] tends to zero. Hereby \(p\in [1,\infty]\) and V is a neighbourhood of K. On the other hand we measure the increase of the derivatives of f. The gauge is given in terms of an increasing weight function \(\omega\) :\((0,\infty)\to {\mathbb{R}}\) which grows faster than log and slower than t and which satisfies a mild regularity condition. With \(\omega\) we associate \({\tilde \omega}\)(t):\(=\sup_{s>0}\{t \log (s)-\omega (s)\}\). E.g. we consider the following conditions: \[ (\alpha_{\rho,p})\quad \sup_{n\in {\mathbb{N}}}d_{n,p,V}(f)e^{\omega (\rho n)}<\infty \quad for\quad some\quad V. \] \[ (\beta_{\rho})\quad \sup_{\alpha \in {\mathbb{N}}^ N_ 0}| f^{(\alpha)}|_{L^ 2(V)}e^{-\rho {\tilde \omega}(\rho | \alpha |)}<\infty \text{ for some } V. \] One of the results states \((\alpha_{1,p})\Rightarrow (\beta_{\rho})\) for every \(0<\rho <1\) and \((\beta_ 1)\Rightarrow (\alpha_{\rho,p})\) for every \(p\in [1,\infty]\) and \(0<\rho <1\). Hence there is a certain loss which can be made arbitrarily small by choosing \(\rho\) very close to 1. Another method of measuring the increase of the derivatives of a function f is to determine the decrease of its Fourier transform. Since f is not defined on all of \({\mathbb{R}}^ N\) we use suitable cutting-off functions which must be chosen very carefully in order to include quasi-analytic and real analytic cases [see also \textit{L. Hörmander}; The analysis of linear partial differential operators, Vol. I (1983; Zbl 0521.35001), Sect. 8.4.]. The latter point of view combined with Markov's inequality is essential for the proof.
    0 references
    germ
    0 references
    weight function
    0 references
    measuring the increase of the derivatives of a function
    0 references
    decrease of its Fourier transform
    0 references
    Markov's inequality
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references