简单序约束条件下参烽估计的EM方法

Estimation of Simple Constrained Parameters with EM Method

在有些数学模型中,参数之间是有约束的.已经有文章给出最简单的情况——对数据均值的简单序贯下的估计方法,这里进一步探讨对一般条件下的线性模型(y=Xθ+e,θ=(1θ,2θ,…,kθ),θ1cθ2≤…≤θk),在简单序约束下的参数的估计.通过一些变形,构造出所谓"潜在变量",并用EM方法,求出参数的估计.

There are many constrained parameter problems in mathematic model. Since there were papers focusing on the simplest occasion,which estimated the mean with simple ordering of parameters（for example, θ=（θ1,θ2,…, θk ） θ1 ≤ θ2 ≤…≤ θk ） ）, we intend to resolve more common occasion-linear model （y = Xθ＋ e） ,to estimate the parameter O. With transformation,the ＂potential variable＂ is constructed and the estimation of parameters is abtained by using EM method.