Two classes of symmetric sign patterns that require unique inertia. (Q1413711)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Two classes of symmetric sign patterns that require unique inertia. |
scientific article |
Statements
Two classes of symmetric sign patterns that require unique inertia. (English)
0 references
17 November 2003
0 references
Sign pattern matrices (sign patterns) \(A\) of order \(n\), whose entries are from the set \(\{+, -, 0\}\) are studied. A class of real matrices, called the qualitative class of \(A\) and denoted by \(Q(A)\) is associated with each \(A\) from the set of all \(n \times n\) sign patterns. A sign pattern \(A \in Q(A)\) is a double star (respectively cycle) sign pattern if the undirected graph of \(A\) is a double star (respectively cycle) graph, possibly with loops. The inertia of the real symmetric matrix \(A\), written as \(i(A)\) is the set \(i(A)= \{i(B)=(i_+(B), i_-(B), i_0(B)) \mid B=B^T \in Q(A)\}\), where \(i_+(B)\) (respectively \(i_-(B)\), \(i_0(B)\)) denotes the number of positive (respectively negative, zero) eigenvalues of a matrix \(B\) counted with their algebraic multiplicities. In a special case, when \(i(B_1)=i(B_2)\) for all real symmetric matrices \(B_1, B_2 \in Q(A)\), it is said that the symmetric sign pattern \(A\) requires unique inertia. \textit{F. J. Hall}, \textit{Zhongshan Li } and \textit{Di Wang} [Linear Algebra Appl. 338, 153-169 (2001; Zbl 0994.15028)] have given sufficient and necessary conditions when a symmetric sign pattern requires unique inertia. These conditions are difficult to fulfill and in this work the special cases which characterize the double star and cycle sign patterns requiring unique inertia are presented and respective unique inertia are found.
0 references
sign pattern
0 references
double star sign pattern
0 references
cycle sign pattern
0 references
inertia
0 references
unique inertia
0 references
undirected graph
0 references
symmetric matrix
0 references
eigenvalues
0 references