I understand that traditionally the value of 0.23 is suggested for the acceptance rate when monitoring a Metropolis-Hastings algorithm, why is this and what would the effect of having a higher acceptance rate be?
Asked
Active
Viewed 55 times
1
-
It still works you just don't converge as fast. I think it has been fine tuned, i'm sure you can find the research articles about it online. – Daniel Adams Jan 22 '21 at 13:21