Preservers of matrix pairs with bounded distance (Q875005)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Preservers of matrix pairs with bounded distance |
scientific article |
Statements
Preservers of matrix pairs with bounded distance (English)
0 references
10 April 2007
0 references
A classical result by \textit{L.-K.\ Hua} [A theorem on matrices over a field and its applications. Acta Math. Sinica 1, 109--163 (1951)] characterizes strong (=in both directions) adjacency preserving bijections on rectangular matrices. This result has been extended in various directions by different authors. We expose here a paper by \textit{H. Havlicek} and \textit{P. Å emrl} [Stud. Math. 174, No. 1, 99--109 (2006; Zbl 1107.47023)] where the authors classify bijections on rectangular matrices which strongly preserve the maximal rank distance (i.e., \(\text{rk}(A-B)\) is maximal precisely when same holds for \(\text{rk}(\Phi(A)-\Phi(B))\)). The article under review is an attempt to close the missing `rank gap'. The authors consider a tensor product space \(U\otimes V\) of vector spaces over a field with at least three elements. Given any integer \(k=2,\dots,\min\{\dim U,\dim V\}-1\) they classify surjections \(\Phi:U\otimes V\to U\otimes V\), which satisfy \[ \text{rk}(A-B)\leq k\Longleftrightarrow \text{rk}(\Phi(A)-\Phi(B))\leq k.\tag{1} \] This is readily applicable to rectangular matrices, and to pairs of maps that intertwine with the \(k\)-th compound. The proof is a nice localization. It builds on an idea of a second `commutant' \(\{A,B\}^{\bot\bot}\), with \(\Omega^\bot\) defined by \[ \Omega^\bot:=\{B\in U\otimes V; \text{ rk}(B-A)\leq k\;\forall A\in\Omega\}. \] With the similar techniques the authors also classify surjections on Hermitian matrices over a field with involution, and with at least three elements, that satisfy (1).
0 references
adjacency preserving mapping
0 references
rank
0 references
tensor spaces
0 references
Hermitian matrices
0 references