一类矩阵椭球等高分布在数据不完全的极大似然估计

THE MLE OF PARAMETERS FOR A CLASS OF MATRIX ELLIPTICALLY CONTOU RED DISTRIBUTIONS IN INCOMPLETE DATA

摘要: 对于一类矩阵椭球等高分布族在数据不完全情况下的参数估计,本文提出一种迭代算法,建立了三个引理,证明迭代过程概率为1地收敛,而且该极限点即为参数在数据不完全情况下致联合密度达极大的估计(本文借用MLE记号).

Abstract: The estimation of parameters for a class of matrix elliptically contoured distributions in incomplete data is discussed. We present an iterative algorithm, establish three lemmas, and prove that the algorithm converges with probability 1, and its limit point is the maximum likelihood estimators （MLE） of parameters in incomplete data.