On the spectrum of pairwise comparison matrices (Q1827515)

\(A=[a_{ij}]\in C^{n\times n}\) is called symmetrically reciprocal (SR) if \(a_{ij}a_{ji}=1\) for \(i\neq j\) and \(a_{ii}=1\). \(B=[b_{ij}]\) with \(b_{ij}>0\) is called transitive if \(b_{ij}b_{jk}=b_{ik}\) for all \(i,j,k\). A transitive matrix \(B\) is necessarily SR and \(\text{rank}B=1\). The authors introduce the so-called pseudo-circulant matrices. Let me describe them for \(n=3\), the generalization for \(n>3\) will be obvious. \[ \Omega=\begin{pmatrix} 0&1&0\\0&0&1\\\exp\varphi&0&0\end{pmatrix} \] is a primitive pseudo-circulant matrix (for \(\varphi=0\) it is the well-known primitive circulant matrix) and \(\sum_{\nu=0}^2c_{\nu}\Omega^{\nu}\) is a pseudo-circulant matrix. Then using pseudo-circulant matrices the spectral decomposition of specific (SR) perturbations of transitive matrices is obtained. The results can be applied in the analytic hierarchy process in mathematical psychology and also in vehicle system dynamics. Three application examples are included in the paper.