作者: Søren Ladegaard Buhl
DOI:
关键词: Combinatorics 、 Mathematics 、 Estimation theory 、 Maximum likelihood 、 Gaussian 、 Zero (complex analysis) 、 Likelihood function 、 Independence (probability theory) 、 Restricted maximum likelihood 、 Estimator 、 Statistics
摘要: In graphical Gaussian models, it is well known that the maximum likelihood estimator exists with probability one if number of replicates at least as large variates. this paper we deal case fewer replicates. For chordless p-cycle two replicates, prove strictly between zero and one. (In p = 4 independence, 2/3.)