Bounds for Levinger's function of nonnegative almost skew-symmetric matrices (Q2496646)

Let \(A\) be an \(n\times n\) real matrix. Then \(A\) can be written as \(A=S(A)+K(A)\), where \(S(A)\) and \(K(A)\) are the real symmetric part and the real skew-symmetric part of \(A\), respectively. The authors study the spectrum and the spectral radius of an entry-wise nonnegative matrix whose symmetric part has rank one. They refer to such a matrix as ``non-negative almost skew-symmetric''. For any \(n\times n\) matrix \(A\), Levinger's transformation and Levinger's function are defined by \(L(A,\alpha)= (1-\alpha)A+\alpha A^t\), \(\alpha\in [0,1]\), and \(\phi(A,\alpha)=\rho(L(A,\alpha))\), \(\alpha\in [0,1]\), where \(\rho(L(A,\alpha))\) is the spectral radius of \(L({A},\alpha)\). Levinger's transformation is a basic tool in this paper. The authors extend their results [ibid. 360, 43--57 (2003; Zbl 1019.15005)] and in a joint paper with \textit{J. J. McDonald} [Rocky Mt. J. Math. 34, No. 1, 269--288 (2004; Zbl 1058.15029)] to obtain new and sharper bounds for the Perron root of Levinger's transformation and its derivative. They present an illustrative example and apply their results to tournament matrices to obtain a comparison result for spectral radii. They also show that the spectral radius of the Brualdi-Li matrix is maximum among \(n\times n\) matrices whose ``score variance'' exceeds a certain function of \(n\).