Re: [分析] [分析]測度一題

看板Math作者 (日夕)時間12年前 (2012/04/19 18:10), 編輯推噓1(100)
留言1則, 1人參與, 最新討論串2/2 (看更多)
※ 引述《sitma (sitma)》之銘言: : m(A)>0,m(B)>0 => : 存在一個open interval (c,d)包含於R : 使得(c,d)包含於A+B Without losing generality , Assume that 0< m(A),m(B)<∞ 證明 : A-B 包含某個區間 . (A,B:Lebesgue measurable) 用下面的lemma大概就是最少預備知識. (這就不證了) ──────── Lemma : Let E be a measurable subset of |R with m(E)>0 , then (3.37) the set of differences E-E = {d:d=x-y | x,y屬於E } contains an interval centered at origin . R. L. Wheedem and A. Zygmund :"measure and integral," p.46 ──────── Choose closed connected intervals I1,I2 so that 0 < (2/3) * m(I1) < m(A1) (記 m(I1)=s , m(I2)=t ) 0 < (2/3) * m(I2) < m(B2) where A1 = A∩I1 , B2 = B∩I2 ( 這個選取是可以的 ,在上面Lemma證明中可以看出,當然也可以metric density來看) Suppose that I1=[a,b] , t≦s and [s/t]= n . ([]:Gauss's symbol) [a,b] = [a,a+t]∪[a+t,a+2t]∪..[a+nt,b] .....(1) Claim : There exists some x 屬於 |R such that m(A1 ∩ (B2+x))>0 ( B2+x ={ b+x | b 屬於B2 } ) Proof : if not , ∵(1) ,then s = m(I1) ≧ m(E1) + [s/t]* m(E2) ( n*t ≦ s ﹤(n+1)*t ) > (2/3)s + n*(2/3)t > (2/3)s + (1/3)s = s →← -------------------- So for some x , we have m(A ∩ (B+x)) ≧ m(A1 ∩ (B2+x))>0 Let T = A ∩ (B+x) , then A-B = ( (A-(B+x)) + x ) 包含 (T-T)+x  ̄ ̄ ̄ ̄ ̄ ̄ ̄ ̄# ─────────── W. Rudin :"Real And Complex Analysis," (3d ed.) Chapter 7, Exercise 5 有提示2種方法(第一種基本上就是類似上面) 第二種是用convolution去做 (很簡潔) -- ※ 發信站: 批踢踢實業坊(ptt.cc) ◆ From: 59.112.238.160

04/20 10:35, , 1F
萬分感謝....真神
04/20 10:35, 1F
文章代碼(AID): #1FZ-HyIC (Math)
討論串 (同標題文章)
文章代碼(AID): #1FZ-HyIC (Math)