On the spectral radius of positive operators on Banach sequence spaces (Q972799)

From MaRDI portal





scientific article; zbMATH DE number 5710788
Language Label Description Also known as
default for all languages
No label defined
    English
    On the spectral radius of positive operators on Banach sequence spaces
    scientific article; zbMATH DE number 5710788

      Statements

      On the spectral radius of positive operators on Banach sequence spaces (English)
      0 references
      0 references
      0 references
      21 May 2010
      0 references
      The authors study under what conditions on a Banach lattice, every positive weak Dunford-Pettis operator is weakly compact. They mainly prove the following: Let \(K_1,\dots,K_n\) be non-negative matrices that define operators on \(L\in{\mathcal L}\), where \(\mathcal{L}\) denotes the collection of all Banach sequence spaces \(L\) satisfying the property that \(e_n=\chi_{\{n\}}\in L\) and \(\| e_{n}\| _L=1\) for all \(n\in\mathbb R\). Given a function \(f:[0,\infty)\times\dots\times[0,\infty)\rightarrow[0,\infty)\) of \(n\) variables, the authors define the matrix \(\widehat{f}(K_1,\dots ,K_n)=[m(i,j)]_{i,j\in\mathbb R}\) by \[ m(i,j)= \begin{cases} f(k_{1}(i,j),\dots ,k_{n}(i,j)) &\text{if }i\neq j,\\ (k_1(i,i)+\dots+k_n(i,i))/n &\text{if }i=j. \end{cases} \] Hence the diagonal part of \(\widehat{f}(K_1,\dots ,K_n)\) is equal to the diagonal part of \((K_1+\cdots+K_n)/n\). By consideration of diagonal matrices, the authors conclude that this is the largest possible diagonal part for the inequality \[ r\big(\widehat{f}(K_1,\dots ,K_n)\big)\leq \tfrac1n \big(r(K_{1})+\cdots +r(K_n)\big)\tag{1} \] to hold for all \(K_1,\dots ,K_n\). The authors also remark that, if \(g(x_1,\dots ,x_n)=\root{n}\of{x_1\dots x_n}\), then \(\widehat{g}(K_1,\dots ,K_n)\) \(=C(K_1,\dots ,K_n)\). Let \(F_n\) be the collection of all functions \(f:[0,\infty)\times\dots\times [0,\infty)\rightarrow[0,\infty)\) of \(n\) variables such that (1) holds for all \(L\) in \({\mathcal L}\) and for all non-negative matrices \(K_1,\dots ,K_n\) defining operators on \(L\), and such that \(f(x_1 ,x_2,\dots ,x_n)=f(x_2,x_3,\dots ,x_n,x_1)\) for all nonnegative numbers \(x_1,x_2,\dots ,x_n\). A natural partial order can be introduced in the collection \(F_n\): for \(f_1,f_2\in{\mathcal F}\), the inequality \(f_1\leq f_2\) means that the function \(f_1-f_2\) is nonnegative. According to this partial order, the authors prove that the function \(g\) defined by \(g(x_1,\dots ,x_n)=\root{n}\of{x_1\dots x_n}\) is the greatest element of the partially ordered set \(F_n\).
      0 references
      spectral radius
      0 references
      inequalities
      0 references
      infinite non-negative matrices
      0 references
      positive operators
      0 references
      Hadamard product
      0 references
      Schur product
      0 references
      weighted geometric mean
      0 references
      Banach sequence spaces
      0 references
      Banach function spaces
      0 references

      Identifiers