0
$\begingroup$

Intuitively, if I want to update two parameters in one step, I have to come up with a proposal that are good for both parameters. Assuming that the parameters are independent, is it correct to demonstrate this intuition by saying the following?

if the probability of accepting the proposal for one parameter is X, then the probability of accepting the proposal for two parameters is lower at $X^2$

Based on my experience running MCMC I have a hunch that this statement is not mathematically correct (i.e. the probability of accepting for two parameters can't be as low as $X^2$), but I cannot figure out why mathematically.

$\endgroup$
2
  • 1
    $\begingroup$ The question does not make much sense to me: what is the meaning of independence of two parameters in this MH context? how can you marginalise the MH algorithm to accept only one parameter? Please produce an illustration. $\endgroup$
    – Xi'an
    Apr 2, 2018 at 8:27
  • $\begingroup$ My apology -- I indeed asked this in haste and it's a really low effort question. I'll think about this more carefully and come up with a concrete example. $\endgroup$
    – Heisenberg
    Apr 2, 2018 at 16:12

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.