Circular Hessenberg pairs

From MaRDI portal



Abstract: A square matrix is called Hessenberg whenever each entry below the subdiagonal is zero and each entry on the subdiagonal is nonzero. Let M denote a Hessenberg matrix. Then M is called circular whenever the upper-right corner entry of M is nonzero and every other entry above the superdiagonal is zero. A circular Hessenberg pair consists of two diagonalizable linear maps on a nonzero finite-dimensional vector space, that each act on an eigenbasis of the other one in a circular Hessenberg fashion. Let A,A∗ denote a circular Hessenberg pair. We investigate six bases for the underlying vector space that we find attractive. We display the transition matrices between certain pairs of bases among the six. We also display the matrices that represent A and A∗ with respect to the six bases. We introduce a special type of circular Hessenberg pair, said to be recurrent. We show that a circular Hessenberg pair A,A∗ is recurrent if and only if A,A∗ satisfy the tridiagonal relations. For a circular Hessenberg pair, there is a related object called a circular Hessenberg system. We classify up to isomorphism the recurrent circular Hessenberg systems. To this end, we construct four families of recurrent circular Hessenberg systems. We show that every recurrent circular Hessenberg system is isomorphic to a member of one of the four families.


In [\textit{A. Godjali}, Linear Algebra Appl. 432, No. 12, 3231--3249 (2010; Zbl 1209.15007)], the concept of Hessenberg pair is introduced as an extension of the Leonard pairs which are defined in the framework of irreducible tridiagonal matrices (see [\textit{P. Terwilliger}, Linear Algebra Appl. 330, No. 1--3, 149--203 (2001; Zbl 0980.05054)]). In the present paper, the author deals with a particular family of Hessenberg pairs which are said to be circular Hessenberg (CH) pairs. A (upper) Hessenberg matrix \(M\) of size \((d+1)\times (d+1) \) with entries in a field \(\mathbb{F}\) is said to be circular if the \((0,d)\) entry is nonzero and every other entry above the superdiagonal is zero. Given a \((d+1)\)-dimensional linear space \(V\) on a field of scalars \(\mathbb{F},\) a pair of elements \((A, A^{*})\) in End\((V)\) is said to be a CH pair if \begin{itemize} \item there exists a basis of \(V\) such that the matrix representing \(A\) is diagonal and the matrix representing \(A^{*}\) is circular Hessenberg; \item there exists a basis of \(V\) such that the matrix representing \(A^{*}\) is diagonal and the matrix representing \(A\) is circular Hessenberg. \end{itemize} Notice that, as the author points out, \(A^{*}\) is not the conjugate transpose of \(A\) as usual, but an arbitrary element in End\((V)\). On the other hand, a circular Hessenberg (CH) system on \(V\) is defined as a family \( \Phi= (A; \{E_{i} \}_{i=0}^{d}; A^{*}; \{E^{*}_{i} \}_{i=0}^{d})\) such that each of \(A, A^{*} \) is diagonalizable, their corresponding eigenvalues \(\theta_{k}, \theta^{*}_{k}, k= 0,1, \dots , d\) are distinct and ordered, and for the matrices \[E_{i}=\prod _{0\leq j\leq d, j\neq i} \frac{A- \theta_{j}I}{\theta_{i}- \theta_{j}}, \qquad E^{*}_{i}=\prod _{0\leq j\leq d, j\neq i} \frac{A^{*}- \theta^{*}_{j}I}{\theta^{*}_{i}- \theta^{*}_{j}}\] one has that \( E_{i} A^{*} E_{j}, E^{*}_{i} AE^{*}_{j} \) are particular cases of circular Hessenberg matrices with every entry above the diagonal (different of the entry \((0,d)\)) is zero. Given \(\beta \in \mathbb{F}\) a finite set \(\{\theta_{k}\}_{k=0}^{d}\) of elements of \(\mathbb{F}\) is said to be \(\beta\)-recurrent if \(\theta_{k+1} = (\beta+1) \theta_{k} -(\beta+1) \theta_{k-1}+ \theta_{k-2}, 2\leq k\leq d-1.\) The description of \(\beta\)-recurrent sets is given in [\textit{P. Terwilliger}, Linear Algebra Appl. 330, No. 1--3, 149--203 (2001; Zbl 0980.05054)]. Consider a lower bidiagonal matrix \(M= (m_{i,j})_{i,j=0}^{d},\) with diagonal entries \(m_{k,k}= \theta_{d-k}\) and \(1\) in the subdiagonal and \(M^{*}= (a^{*}_{i,j})_{i,j=0}^{d}\) an upper bidiagonal matrix with diagonal entries \(m^{*}_{k,k}= \theta^{*}_{k}\) and \(m^{*}_{k, k+1}= \phi_{k}\) in the superdiagonal. Assuming \(\theta_{k}, \theta^{*}_{k}, k= 0,1, \dots , d,\) and \(\vartheta_{k}, k= 0,1, \dots , d+1,\) where \(\vartheta_{k}= \phi_{k}- (\theta^{*}_{k}- \theta^{*}_{0}) (\theta_{d-k+1}-\theta_{0}), 1\leq k\leq d, \vartheta_{0}= \vartheta_{d+1}=0, \vartheta_{1} \neq \vartheta_{d}\) are \(\beta\)-recurrent, then \[ \Phi= (M; \{E_{i} \}_{i=0}^{d}; M^{*}; \{E^{*}_{i} \}_{i=0}^{d})\] is a CH system. Notice that there is a unique set \(\{\phi_{k}\}_{k=0}^{d}\) of nonzero scalars such that if \[ \Phi= (A; \{E_{i} \}_{i=0}^{d}; A^{*}; \{E^{*}_{i} \}_{i=0}^{d})\] is a CH system, then there exists a basis of \(V\) such that the matrices representing \(A, A^{*}\) in such a basis are \(M. M^{*},\) defined as above, respectively. \(\Phi\) is said to be recurrent if there exists \(\beta\in\mathbb{F}\) such that \(\theta_{k}, \theta^{*}_{k}, k= 0,1, \dots , d,\) and \(\vartheta_{k}, k= 0,1, \dots ,s, d+1,\) are \(\beta\)-recurrent. Four basic families of recurrent CH systems are described and the author proves that every recurrent CH system is isomorphic to one of the above four families. Six bases of \(V\) are deduced in such a way the corresponding transition matrices as well as the matrices representing \(A\) and \(A^{*}\) with respect to such bases are given. The author conjectures that every CH system in the linear space \(V\) is recurrent.











This page was built for publication: Circular Hessenberg pairs

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2085426)