Sharp interpolation inequalities for discrete operators (Q492739)

Let \(l^{2}({\mathbb Z}^{d})\) (\(d\geq1\)) be the standard sequence space on \({\mathbb Z}^{d}\) with the norm \(\| \cdot \|\). For \( u\in l^{2}({\mathbb Z}^{d})\), denote \(\| \nabla u \|^{2} = \sum_{i=1}^{d} \| D_{i}u\|^{2}\), where \((D_{i}u)(n)=u(n+e_{i})-u(n)\), \(n\in {\mathbb Z}^{d}\), \(e_{i}\in {\mathbb Z}^{d}\) is the vector with the coordinates \(\delta_{ij}\), \(j=1,\dots, d\). Interpolation inequalities of the form \( \sup_{n\in {\mathbb Z}^{d}}u(n)^{2} \leq K_{d}(\theta)\cdot \| u \|^{2\theta}\cdot \| \nabla u \|^{2(1-\theta)} \), \(0 \leq \theta\leq 1\), are considered. A general method for finding sharp constants \(K_{d}(\theta)\), extremal elements and possible correction terms for these inequalities is developed. Some applications to the spectral theory of discrete Schrödinger operators are given.