0

I have a dummy question on the calculation of the likelihood ratio in MCMC. In some codes it is written: `

$\alpha = exp(a-b)$ with $a = -\sum(obs - y_{candidate})^2$ and $b = -\sum(obs-y_{previous})^2$. I don't understand where the $exp$ comes from and why we don't calculate directly $\alpha = a/b$ ?

Thanks a lot for your help, Best, D2B

D2B
  • 1
  • Can you provide more context and source? – gunes Jun 02 '20 at 08:14
  • https://stats.stackexchange.com/questions/73629/how-to-sample-using-mcmc-from-a-posterior-distribution-in-general/469797?noredirect=1#comment867582_469797 – D2B Jun 02 '20 at 13:01

0 Answers0