作者b218h (Gordon Mercer)
站内Statistics
标题Re: [问题] Stochastic Dominance
时间Wed Feb 18 06:26:47 2009
很抱歉. 当这个积分的 support 不是 finite 时 分部积分会挂掉
我尝试了一下证出来了
Sketch of the proof :
1. It suffices for us to prove that Y comletely dominates X implies
E(X) <= E(Y), provided the expectations exist.
+ - + -
2. Let X=X -X , Y=Y -Y
Find out the complete dominance relationship between
+ + - -
X , Y and X , Y
3. Use the following formula:
for a nonnegative random variable (assumed continuous) X.
∞
E(X)=∫ [1-F (t)]dt
0 X
※ 引述《Rubyfish (过去的已不再重要)》之铭言:
: 你先把 E[m(X)] 用积分的形式写出来
: 然後再表示成分部积分的样子
: 相同的做法对於 E[m(Y)] 也是
: 然後 E[m(X)] - E[m(Y)]
: 你会发现分部积分的前面那项 两者是相同的!!
: 於是就比较後面那项
: 你就会得到 积分 m'(t) (Fy - Fx) dt
: 因为 P(X≦c) ≧ P(Y≦c)
: 所以 (Fy - Fx) <= 0
: 所以 积分 m'(t) (Fy - Fx) dt <= 0
: E[m(X)] - E[m(y)] <= 0
: ※ 引述《b218h (Gordon Mercer)》之铭言:
: : Let X,Y be two random variables with the following properties:
: : For all c in |R,
: : P(X≦c) ≧ P(Y≦c)
: : Let m(‧) be a monotone increasing function. Show that
: : E[m(X)]≦E[m(Y)]
--
My Blog
http://www.b218h.blogspot.com
--
※ 发信站: 批踢踢实业坊(ptt.cc)
◆ From: 128.189.228.33
※ 编辑: b218h 来自: 154.20.158.117 (02/18 10:04)