两类分布函数的投影寻踪随机经验分布的极限分布

THE LIMIT DISTRIBUTIONS OF EMPIRICAL DISTRIBUTIONS OF RANDOM PROJECTION PURSUIT OF TWO KINDS OF DISTRIBUTIVE FUNCTIONS

  • 摘要: Xm维随机向量,X1,X2,…,Xn是来自母体X的子样,ZNm(0,Im),Bm>0为实数列,经验分布 \hatF_n^Z / B m(x)=\frac1n \#\left\i: Z^\prime X_4 / B_iX~BK:P1,P2,…,Pm),Pi≥0,d_m=\sum_i=1^m P_i^z。若m→∞时,Bm-2·K→σ2,Kdm→0 那么\hatF_n^8 / \ln (x) \xrightarrowP N\left(0, \sigma^2\right) 当n→∞,m→∞(2)V>0,XNmu,V),若M→∞时, B_m^-2 T_r(V) \rightarrow \sigma^2, B_m^-2\|u\|^2 \rightarrow 0, B_m^-2\left(T_r\left(V^2\right)+2 u^\prime \nabla u\right) \rightarrow 0,那么 \hatF_n^\mathrmz / \mathrmBm(x) \xrightarrowP N\left(0, \sigma^2\right),当n→∞,m→∞。

     

    Abstract: Let X=(x1, x2,…, xm), be a m-dimensional random vector, X1, X2,…, X subsamples from the populatton X, the m-dimensional random vector Z have the Nm(0, Im) distribution and Bm>0 be a sequence of real numbers. Let \hatF_n^\ell / B m(x)=\frac1n \#\left\i: Z^\prime X_4 / B_i X be multinomial distribution\mathscrB\left(k_;, p_1, p_2, \cdots, p_m\right), with each pi≥0, m=\sum_i=1^m p_i^2, there be a sequence of real numbers Bm>0, k/Bm2→σ, k*dm→0, as m→∞ then \hatF_n^Z / B_m(x) \xrightarrowP N\left(0, \sigma^2\right) as n→∞, m→∞. (2) Let X~Nmu,V),V>0, there be a sequence of real numbers B_m^-2 T_r(V) \rightarrow \sigma^2, B_m^-2\|u\|^2 \rightarrow 0, \left(T_r(V)+2 u^\prime V u\right) / B_m^4 \rightarrow 0,as m→∞,then \hatF_n^\mathrmZ / B m(x) \xrightarrowP N\left(0, \sigma^2\right) as n→∞, m→∞.

     

/

返回文章
返回