Inverting linear combinations of identity and generalized Catalan matrices (Q5962297): Difference between revisions

The Catalan numbers are defined by \( C_n=\frac{1}{n+1}{2n\choose n}. \) The authors with \textit{M.~Miladinović} and \textit{A.~Ilić} [Appl. Math. Comput.~215, 796--805 (2009; Zbl 1175.05018)] defined the Catalan matrix of order~\(n\) as a lower triangular matrix \({\mathcal C}_n[x]=(c_{ij}[x])\) with lower triangular entries \( c_{ij}[x]=x^{i-j}C_{i-j}. \) The Catalan numbers can be generalized in many ways. One of them is \( C_n(a,b)=\frac{a}{a+bn}{a+bn\choose n}, \) where \(a>0\) and \(b\geq 0\) are integers. The authors define the generalized Catalan matrix of order~\(n\) as a lower triangular matrix \({\mathcal C}_n[a,b;x]=(c_{ij}[a,b;x])\) with lower triangular entries \( c_{ij}[a,b;x]=x^{i-j}C_{i-j}(a,b). \) Assuming that \(|\lambda|<1/\|{\mathcal C}_n[a,b;x]\|\), where \(\|.\|\) is a (submultiplicative) matrix norm, they find an explicit formula for \((I-\lambda{\mathcal C}_n[a,b;x])^{-1}\). Under this assumption, they also construct a lower triangular matrix~\(\Delta_n\) such that \((I-\lambda{\mathcal C}_n[a,b;x])^{-1}={\mathcal C}_n[a,b;x]\circ\Delta_n\), where \(\circ\) denotes the Hadamard product. The authors' motivation arises from similar results concerning Pascal matrices, due to \textit{R.~Aggarwala} and \textit{M.~P.~Lamoureux} [Am. Math. Monthly~109, 371--377 (2002; Zbl 1027.15012)] and \textit{S.-L.~Yang} and \textit{Z.-K.~Liu} [Int. J. Math. Math. Sci. 2006, Art.~ID~90901, 7~pp. (2006; Zbl 1131.15006)]. Reviewer's remarks: In Theorem~3.2, the condition \(|\lambda|<\|{\mathcal C}_n[a,b;x]\|\) should read \(|\lambda|<1/\|{\mathcal C}_n[a,b;x]\|\) as in Theorem~3.1. But in both of these theorems this condition simplifies into \(|\lambda|<1\). Note that the spectral radius is equal to the greatest lower bound of all (submultiplicative) matrix norms.