Spectral density of Jacobi matrices with small deviations (Q1935026): Difference between revisions
From MaRDI portal
Latest revision as of 03:21, 6 July 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Spectral density of Jacobi matrices with small deviations |
scientific article |
Statements
Spectral density of Jacobi matrices with small deviations (English)
0 references
30 January 2013
0 references
Let \(\mathbf{a}_{k} = (a_{1,k}, \dots, a_{k,k}) \in \mathbb{R}^{k}\) and \(\mathbf{b}_{k} = (b_{1,k}, \dots, b_{k - 1,k}) \in \mathbb{R}^{k - 1},\) and let \(J (\mathbf{a}_{k}, \mathbf{b}_{k})\) denote the Jacobi matrix \[ (\mathbf{a}_{k}, \mathbf{b}_{k}) = \left( \begin{matrix} a_{1,k} & b_{1,k} & 0 & \cdots & 0 \\ b_{1, k} & a_{2,k} & b_{2,k} & \ddots & \vdots \\ 0 & b_{2,k} & a_{3,k} & \ddots & 0 \\ \vdots & \ddots & \ddots & \ddots & b_{k - 1,k} \\ 0 & \cdots & 0 & b_{k - 1,k} & a_{k,k} \end{matrix} \right). \] Under some additional assumptions on the coefficients \(a_{j,k}\) and \(b_{j,k}\), called ``small-deviation conditions'', asymptotic trace formulas are proved for the family of considered Jacobi matrices \(J (\mathbf{a}_{k}, \mathbf{b}_{k})\). It is said that a sequence \((\mathbf{a}_{k})\) of elements \(\mathbf{a}_{k} = (a_{1,k}, \dots, a_{k,k}) \in \mathbb{R}^{k}\) belongs to the class \(S\) (or, also, satisfies small-deviation conditions) if \[ \sum_{i = 1}^{k - 1} | a_{i + 1,k} - a_{i,k} | = o (k) \;\;\text{as }k \to \infty, \tag{i} \] \[ 0 \leq a_{i,k} \leq 1 \text{ for all }\;\;i = 1,\dots, k.\tag{i} \] The first formula is concerned with the moments of Jacobi matrices \(J (\mathbf{a}_{k}, \mathbf{b}_{k})\), namely, it is proved that, if the sequences \((a_{k})\), \((b_{k})\) satisfy the small-deviation conditions (i) and (ii), then \[ \operatorname{Trace } [ J^{k} (\mathbf{a}_{k}, \mathbf{b}_{k}) ] = \sum_{j = 0}^{[ n/2 ]} \;\frac{n!}{(j!)^{2} (n - 2 j)!} \;\sum_{j = 2}^{k} \;(a_{i,k})^{n - 2j} \;( b_{i -1,k})^{2j} + o (k). \] This trace formula remains true if, instead of \((\mathbf{a}_{k}), (\mathbf{b}_{k}) \in S\), it is required that \((\mathbf{a}_{k}), (\mathbf{b}_{k}) \in S',\) where the class \(S'\) is determined by the all sequences \((\mathbf{a}_{k})\) such that, for any \(\delta \in (0, 1)\), there hold (also called small-deviation conditions) \[ \sum_{i = 1}^{k - 1} | a_{i + 1,k} - a_{i,k} | = O (k^{1 - \delta}) \text{ as }k \to \infty,\tag{i'} \] \[ \max_{1 \leq i \leq k} \;| a_{i,k} | \;= \;O (\log k) \text{ as } k \to \infty.\tag{ii'} \] The next asymptotic trace formulas concern Jacobi matrices whose defining sequences \( (\mathbf{a}_{k}), (\mathbf{b}_{k}) \) are distributed according to some probability measure. It is said that a sequence \(( (\mathbf{a}_{k}, \mathbf{b}_{k}) )\) with \(a_{j,k}\) and \(b_{j,k}\) in \([0, 1]\) is \(\mu\)-distributed if, for any continuous function \(\psi\) on \(I^{2} = [0, 1] \times [0, 1]\), there holds \[ \lim_{k \to \infty} \frac{1}{k} \sum_{j = 2}^{k } \psi (a_{j,k}, b_{j - 1,k}) = \int_{I^{2}} \psi (x, y) \,d \mu (x, y). \] It is found that, if \((\mathbf{a}_{k}), (\mathbf{b}_{k}) \in S\) and the sequence \(( (\mathbf{a}_{k}, \mathbf{b}_{k}) )\) is \(\mu\)-distributed for a probability measure \(\mu\) on \(I^{2}\), then \[ \lim_{n \to \infty} \frac{1}{k} \operatorname{Trace} [ \varphi (J (\mathbf{a}_{k}, \mathbf{b}_{k}) ) ] = \frac{1}{\pi} \int_{I^{2}} \int_{0}^{\pi} \varphi (x + 2 y \cos t) \,d t\, d \mu (x, y) \] for any \(\varphi \in C [- 2, 3]\). This asymptotic formula is extended to the case of unbounded sequences. Several existence results are also proved. According to those, for any given sequences \((\mathbf{a}_{k}), (\mathbf{b}_{k})\) satisfying the small-deviation conditions, one can always find subsequences \((\mathbf{a}_{k_{j}}), (\mathbf{b}_{k_{j}})\) and a probability measure \(\mu\) for which similar trace asymptotic formulas are valid.
0 references
Jacobi matrices
0 references
spectrum
0 references
asymptotic density
0 references
0 references
0 references