Norms and lower bounds of operators on the Lorentz sequence space \(d(w,1)\) (Q1282984)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Norms and lower bounds of operators on the Lorentz sequence space \(d(w,1)\)
scientific article

    Statements

    Norms and lower bounds of operators on the Lorentz sequence space \(d(w,1)\) (English)
    0 references
    0 references
    10 August 1999
    0 references
    Let \(E\) be a Banach sequence space and \(A\) a positive operator of \(E\) to itself. Let \(\Delta_1(E)\) be the set of decreasing, non-negative sequences in \(E\) with norm 1. Conditions are given ensuring that \(\| A\|= \sup\{\| Ax\|\); \(x\in \delta_1(E)\}\). The ``lower bound'' of \(A\) is defined to be \(m(A)= \inf\{\| Ax\|: x\in \delta_1(E)\}\). Now let \(d(w,1)\) be the Lorentz sequence space determined by a weighting sequence \(w=(w_n)\). Let \(A\) be given by the matrix \((a_{i,j})\), and let \(v_j= \sum_{i=1}^\infty a_{i,j}w_i\). Write \(W_n= w_1+\cdots+ w_n\) (and \(V_n\) similarly). Under the conditions mentioned, \(\|A\|\) and \(m(A)\) are the supremum and infimum of \(V_n/W_n\): this means that both are determined by elements of the form \((1,\dots,1,0,\dots)\). The exact evaluation of these quantities is then considered for some classical matrix operators, with the weighting sequences \(w_{(1)}, w_{(2)}\) defined respectively by \(w_n= 1/n^\alpha\) and \(W_n= n^{1-\alpha}\): denote the corresponding norms and lower bounds by \(\|\;\|_{(j)}\) and \(m_{(j)}(\;)\). Some of the resulting problems are of interest purely as questions on inequalities. For the averaging (Cesàro) operator \(A\), we find that \(\| A\|_{(1)}= \zeta(1+\alpha)\) and \(m_{(1)}(A)= 1/\alpha\), while \(\| A\|_{(2)}= 1/\alpha\) and \(m_{(2)}(A)= \sum_{j=1}^\infty c_j\), where \(c_j= 1/[j^\alpha (j+1)]\): The proof is by showing that \(n^\alpha= \sum_{j\geq n}c_j\) is increasing if \(0< \alpha\leq 1\) (and decreasing if \(\alpha>1\)). For the ``Copson'' operator \(C\) (the transpose of \(A\)), both choices of \(w\) give \(\| C\|= 1/(1-\alpha)\) and \(m(C)=1\). Let \(H_0,H_1\) be the Hilbert operators with matrices \(1/(i+j)\) and \(1/(i+j-1)\) respectively. Then \(\| H_1\|_{(1)}= \| H_0\|_{(2)}= \pi/\sin \alpha\pi\) and \(m_{(1)}(H_1)= m_{(2)}(H_0)= \sum_{j=1}^\infty c_j\), with \(c_j\) as above. The case \((H_1,w_{(2)})\) is also solved, but no simple formula seems likely for \((H_0,w_{(1)})\).
    0 references
    0 references
    0 references
    0 references
    0 references
    Banach sequence space
    0 references
    positive operator
    0 references
    lower bound
    0 references
    Lorentz sequence space
    0 references
    Copson operator
    0 references