On the imbedding problem for three-state time homogeneous Markov chains with coinciding negative eigenvalues (Q662879)

The imbedding problem for finite Markov chains has been initiated by \textit{G. Elfving} [Acta Soc. Sci. Fennicae, N. Ser. A 2, No. 8, 1--17 (1937; Zbl 0017.31603)] and advanced by many researchers (M. Bladt, M. Søren, P. Carette, H. Frydmann, B. Fuglede, S. Johansen, W. J. McCausland, J. T. Runnenberg and others). \textit{J. F. C. Kingman} [Z. Wahrscheinlichkeitstheor. Verw. Geb. 1, 14--24 (1962; Zbl 0118.13501)], while D. G. Kendall proved that a \(2\times 2\) transition probability matrix is compatible with a continuous Markov process if and only if the sum of the two diagonal entries is larger than 1. The authors provide a necessary and sufficient conditin of the embedding problem, when the \(3\times 3\) indecomposable transition probability matrix of a homogeneous Markov chain has coinciding negative eigenvalues, by using an alternate parametrization of the intensity matrix.