Maximal and minimal triangular matrices (Q722468)

Let \(\mathcal{A}\) be a matrix algebra and \(a\) a non-scalar matrix. We denote by \(\mathcal{C}(a)\) the centralizer of \(a \in \mathcal{A}\). Now we can define a preorder \(\preceq\) on \(\mathcal{A}\) as follows: \[ a \preceq b \,\, \text{ if and only if } \,\, \mathcal{C}(a) \subseteq \mathcal{C}(b). \] If \(a \in \mathcal{A}\) is a maximal (or minimal, respectively) element in \((\mathcal{A}, \preceq )\), then we call \(a\) a maximal (or minimal, respectively) matrix. The main aim of this article is to study maximal and minimal elements of \(\mathcal{T}_{k}(\mathbb{F})\) -- the algebra of all \(k \times k\) upper-triangular matrices over a field \(\mathbb{F}\). In order to formulate the main result, let us recall that \(e_{k}\) denotes the \(k \times k\) identity matrix, and \(e_{n,m}\) is the matrix with \(1\) in the position \((n,m)\) and \(0\) otherwise. {Theorem 1}. Let \(\mathbb{F}\) be a field and \(k\geq 3\). Then, \(a \in \mathcal{T}_{k}(\mathbb{F})\) is maximal iff \(a\) is of the form \(\alpha e_{k} + \beta b\) with either {\parindent=6mm \begin{itemize}\item[i)] \(b\) is similar to \(e_{n,m}\) for some \(n < m\), or \item[ii)] \(b\) is idempotent different from \(0\) and \(e_{k}\). \end{itemize}} An analogous result also holds for \(\mathcal{T}_{\infty}(\mathbb{F})\) -- the algebra of \(\mathbb{N} \times \mathbb{N}\) upper-triangular matrices over \(\mathbb{F}\). In the case of minimal elements, one can show the following characterization. By \(\sigma( \cdot )\) we denote the spectrum of a matrix. {Theorem 2}. Let \(\mathbb{F}\) be a field and let \(k\) be a natural number. The matrix \(a \in \mathcal{T}_{k}(\mathbb{F})\) is minimal iff \(a\) is \(T\)-similar (see Definition 1) to a matrix that is a generalized direct sum (see Definition 2) of matrices \(a_{1} \in \mathcal{T}_{k_{1}}(\mathbb{F}),\dots, a_{n} \in \mathcal{T}_{k_{n}}(\mathbb{F})\) that satisfy the following conditions: {\parindent=6mm \begin{itemize}\item[i)] each of the matrices \(a_{i}\) is minimal in \(\mathcal{T}_{k_{i}}(\mathbb{F})\); \item[ii)] we have \(|\sigma(a)| = \min \{n, |F|\}\). \end{itemize}} As in the case of maximal elements, one can conclude an analogous result for minimal elements in \(\mathcal{T}_{\infty}(\mathbb{F})\).