广义线性模型中极大拟似然估计的弱相合性

Weak Consistency of Quasi-Maximum Likelihood Estimate in Generalized Linear Models

在λn=O(λn)及其它正则条件满足时,用压缩映射原理证明一般联系函数多维广义线性模型的极大拟似然估计(MQLE)的弱相合性,将误差为独立序列的情形推广到相依序列,其中λn(λn)为矩阵Sn=∑in=1XiXiT的最大(最小)特征根.

Under the assumption of λ-n=O（λX-n） or some other regular conditions,the weak consistency of quasi-likelihood equation was proven in a generalized linear model with q×1 responses,bounded and fixed p×q regressors Xi and general link function by means of the contractive mappings,and the error was extended to the dependent sequence from the independent sequence.Thereinto,λ-n（λX-n）denotes the largest（smallest） eigenvalue of the matrix Sn=∑ni=1XiXTi.