线性模型中误差分布的基于L＿1估计的相合核估计

ERROR DISTRIBUTIONS CONVERGENT KERNEL ESTIMATE WHICH BASED ON L_1-ESTIMATE IN LINEAR MODEL

线性模型ｙ＿ｉ＝ｘ＿ｉ＇β＋ｅ＿ｉ，ｉ＝１，…，ｎ，的误差序列有未知密度ｆ（ｘ）．本文在一定条件下证明了ｆ（ｘ）的核估计的弱相合性，逐点强相合性，一致强相合性，其中为Ｌ＿１估计的残差．

For a linear model y_i= x'iβ+e_i,i=,···,n,the error sequence 1 has anunknown density f(x). The authors prove the weak convergence,the strong convergence and thestrong uniform convergence of f(x)'s kernel estimation under some conditions wherem, e_i is the residuals of L_1-estimation.