Exact constants in the inequalities between norms of functions and their derivatives (Q1109259)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 4069503
Language Label Description Also known as
default for all languages
No label defined
    English
    Exact constants in the inequalities between norms of functions and their derivatives
    scientific article; zbMATH DE number 4069503

      Statements

      Exact constants in the inequalities between norms of functions and their derivatives (English)
      0 references
      0 references
      1987
      0 references
      In connection with the study of the best approximation of the differentiation operator, \textit{V. N. Gabushin} [(*) ibid. 6, No.5, 573- 582 (1969; Zbl 0216.136)] has indicated a method of a construction of the best constant \(Q_{nk}\) for a Kolmogorov type inequality on the half line [0,\(\infty)\) with the mixed norms: \[ (1)\quad \| f^{(k)}\|_ c\leq Q_{nk}\| f\|_{L_ 2}^{1- (2k+1)/(2n)}\| f^{(n)}\|_{L_ 2}^{(2k+1)/(2n)}, \] where \(n=1,...,\infty\), \(k=0,...,n-1\), \(\| f\|_ C=\| f\|_{C[0,\infty)}\), \(\| f\|_{L_ 2}=\| f\|_{L_ 2[0,\infty)}\). A natural space of functions in which inequality (1) is considered is the space \(W_ 2^ n\) of functions f(t) absolutely continuous on [0,\(\infty)\) together with their derivatives up to the (n- 1)-th degree, and such that \(f^{(n)}\in L_ 2\). Changing slightly the notation of (*), we can describe the Gabushin method in the following way. At first it will be established that the following function is extremal for (1): \(g(t)=\sum^{n}_{v=1}a_ ve^{z_ vt}\), where \(z_{\nu}=ie^{i\pi (\nu -)/n}\), and the values \(a_ 1,a_ 2,...,a_ n\) are defined by the system of equations: \(\sum^{n}_{v=1}a_ vz_ v^ j=:\delta_{2n-k-1,j}\) \((j=1,...,2n-1)\). Next, the values of the quadratic forms \(A^ 2_ k=\| g\|_{L^ 2_ 2}\), \(B^ 2_ k=\| g^{(n)}\|_{L^ 2_ 2}\) are computed for the variables \(a_ 1,a_ 2,...,a_ n\), and \(Q_{nk}\) is calculated: \[ Q_{nk}=(A^ 2_ k+B^ 2_ k)/A_ k^{1-(2k+1)(2n)}B_ k^{(2k+1)/(2n)}. \] It is possible to calculate \(a_ v\) with the help of the Vieta formulas. In such a construction of \(Q_{nk}\) it is difficult to obtain any information on the behaviour of the value \(Q_{nk}\) depending on n and k. In the present note we develop another construction of \(Q_{nk}\), which allows us to show the property \(Q_{nk}=Q_{n,n-k-1}\) in an arbitrarily simple way, and to establish the asymptotics of \(Q_{nk}\) for a fixed k. For small values of k and an arbitrary n one can give explicit expressions for \(Q_{nk}\).
      0 references
      best constant
      0 references
      Kolmogorov type inequality
      0 references
      Gabushin method
      0 references

      Identifiers