|
BGonline.org Forums
Further reading: Control variates
Posted By: rambiz In Response To: Further reading: Control variates (Maik Stiebler)
Date: Thursday, 18 August 2016, at 9:13 a.m.
Thank you for the link to the wiki article "Control Variates". I certainly didn't know about the very key word "control variates" before. At least, I know what to google for now.
The wiki article minimizes the variance of the new random variable m*=m+c(t-E(t)), whereas you in your derivation tried to maximize
(Var(m) -Var(m*))/Var(m).
Since Var(m) is a constant, these two methods should lead to the same result.
In the wiki article, I can follow
Var(m*)=Var(m)+c^2Var(t)+2cCov(m,t)
but I don't understand how it is minimized. Any help, what is actually done there, please?
|
BGonline.org Forums is maintained by Stick with WebBBS 5.12.