正态模型参数线性不等式约束估计的优化算法

The Optimization Algorithm of the Restricted Estimation of Normal Distribution under Linear Inequality

研究含有不完全数据的多元正态模型参数在一般线性不等式约束下的极大似然估计问题;利用约束EM算法求得多元正态模型参数的迭代解,同时提出M-步的优化算法;并证明了此解是一般线性不等式约束下的最优解.

This paper studies maximum likelihood estimation of multivariate normal distribution modes with incomplete data under linear inequality restrictions on the parameters. The research gets the iterative solution of the parameters of multivariate normal distribution modes and propose an optimization algorithm of M-step. It is the proof that this solution is optimum solution under linear inequality restrictions.