This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: Week 9 Lecture 16 Quadratic functional estimation Model: Observe the sequence model: y i = & i + n & 1 = 2 z i where z i i:i:d: & N (0 ; 1) . The model comes from the white noise model (or many other models): dy ( t ) = f ( t ) dt + n & 1 = 2 dB ( t ) , t 2 [0 ; 1] . Let f i ( t ) ; i = 1 ; 2 ;::: g be an orthonormal basis of L 2 ([0 ; 1]) . The white noise problem is then equivalent to the sequence model with y i = h i ;y i ; & i = h i ;f i ; and z i = h i ;B i i:i:d: & N (0 ; 1) . Assumption : Let & = ( & 1 ;& 2 ;::: ) . Assume & &;M & = ( & : X i =1 i 2 & & 2 i M ) . If the orthonormal basis is the Fourier basis, this assumption corresponds to the periodic Sobolev ball with smoothness . Now we may write the sequence model as P n = f P n; ; & 2 & &;M g where P n; is the joint distribution of independent y i & N ( & i ; 1 =n ) . Problem : Estimate Q = P i =1 & 2 i ( or R f 2 with f = P i =1 & i i ), and determine the optimal minimax rate n & satisfying lim n...
View Full Document
This note was uploaded on 11/06/2009 for the course STAT 680 at Yale.