t**k 发帖数: 260 | 1 Given a joint distribution P(A,B,C), we can compute various marginal
distributions. Now suppose:
P1(A,B,C) = P(A) P(B) P(C)
P2(A,B,C) = P(A,B) P(C)
P3(A,B,C) = P(A,B,C)
Is it true that d(P1,P3)>=d(P2,P3) where d is the total variation distance?
In other words, is it provable that P(A,B)P(C) is a better approximation of
P(A,B,C) than P(A)P(B)P(C) in terms of the total variation distance? | K*****2 发帖数: 9308 | 2 我假定你的问题是用P(A,B)P(C)的形式,或者P(A)P(B)P(C)的形式,去近似给定的一个
P(A,B,C)。
那估计你算distance的时候,是要求inf的吧,比如inf d(P1,P3), inf对P1求。
那既然P(A)P(B)永远都是P(A,B)的一个特例,取inf肯定是P(A,B)P(C)产生的distance
更小啊。 | f**********d 发帖数: 4960 | 3 应该没有。
total variation distance基于sigma field的任意取最大差的事件。直觉上感觉是很
难保证。
of
【在 t**k 的大作中提到】 : Given a joint distribution P(A,B,C), we can compute various marginal : distributions. Now suppose: : P1(A,B,C) = P(A) P(B) P(C) : P2(A,B,C) = P(A,B) P(C) : P3(A,B,C) = P(A,B,C) : Is it true that d(P1,P3)>=d(P2,P3) where d is the total variation distance? : In other words, is it provable that P(A,B)P(C) is a better approximation of : P(A,B,C) than P(A)P(B)P(C) in terms of the total variation distance?
| t**k 发帖数: 260 | 4 Thanks for the replies. I just find a counter-example, so it is not true.
http://stats.stackexchange.com/a/65975/28478
of
【在 t**k 的大作中提到】 : Given a joint distribution P(A,B,C), we can compute various marginal : distributions. Now suppose: : P1(A,B,C) = P(A) P(B) P(C) : P2(A,B,C) = P(A,B) P(C) : P3(A,B,C) = P(A,B,C) : Is it true that d(P1,P3)>=d(P2,P3) where d is the total variation distance? : In other words, is it provable that P(A,B)P(C) is a better approximation of : P(A,B,C) than P(A)P(B)P(C) in terms of the total variation distance?
|
|