On the structure rank of a class of structure matrices (Q1375103)

A structure matrix \(S\) is a \((0,1)\) matrix \(S= (s_{ij})\) of order \(n\). An \(S\)-submatrix of a square matrix \(A\) of order \(n\) is a matrix \(A[K\mid L]\) of \(A\) such that \(s_{kl}=1\) for all \((k,l)\in K\times L\). The \(S\)-rank of \(A\) is the maximum rank of the \(S\)-matrices of \(A\). A structure matrix \(S\) of order \(n\) is called a transpose complementary structure matrix if (i) the size of every all-1's submatrix \(S[K\mid L]\) of \(S\) is equal to or less \(n\); (ii) whenever \(S[K\mid L]\) is any all-1's submatrix of \(S\) with size \(n\), then \(S[\overline{L}\mid \overline{K}]\) is also an all-1's submatrix of \(S\) with size \(n\), where \(\overline{L}= \{1,\dots,n\} \setminus L\), \(\overline{K}= \{1,\dots,n\} \setminus K\). A transpose complementary structure matrix \(S\) of order \(n\) is called an \(n\)-extendable transpose structure matrix if every all-1's submatrix \(S[K\mid L]\) of size less than \(n\) can be extended to an all-1's submatrix of size \(n\). The authors give a characterization of the \(n\)-extendable transpose complementary structure matrices with trace \(k=0,2\). They prove that a nonsingular matrix and its inverse have the same \(S\)-rank for this class of structure matrices \(S\).