半参数回归模型参数估计的收敛速度

The convergence rate of estimation for parameter in a semi-parammetric model

  • 摘要: 没有半参数回归模型Y=X’β+gT)+e,其中(X,T)为取值于Rp×0,1上的随机向量,βp维未知参数向量,g是定义在0,1上的未知函.e为随机误差,Ee=0,Ee22>0,且(X,T)与σ独立.参数β和σ2的估计量\hat\beta_n和\hat\sigma_n^2通常可利用非参数的权函数估计法与参数的最小二乘方法的结合得到.本文对核函数的情形得到了\hat\beta_n和\hat\sigma_n^2的精确的收敛速度——重对数律.所施条件则与证明\hat\beta_n和\hat\sigma_n^2的渐近正态性时施加的条件一致.又本文的证明方法对一般的权函数也适用。

     

    Abstract: We consider semiparametric modelY=X’β+gT)+e,where (X,T)is random vector onRp×0,1, β is a p-dimensional unknown parameter, g(·) is an unknown function on0, 1, e is random error with mean 0 and variance σ2.And(X,T) and ε are independent each other. We estimate β, σ2 by nonparam-etric weight function and least squared method. In this paper, we prove that these estimations of β, σ2 have law of the iterated logarithm.

     

/

返回文章
返回