Hi all,
I am using bayesmh to estimate a model and would like to use a "continuous adaptation regime" as mentioned in two places of stata manual (screenshots attached at the end). The continuous adaptation regime will allow the adaptation of proposal covariance matrix beyond the burn-in stage and into the simulation stage.
I have searched the manual and did not find a way to invoke such a regime, but remember finding it somewhere online that this can be achieved by setting the tuning parameter gamma in adaptation(gamma( )) at a positive number. I will also need to override the maximum adaptation allowed in adaptation(maxiter( )) to be (# of burn-in + # of simulation)/every( ).
Could someone please confirm or disconfirm this approach, or provide other ways to allow adaptation of proposal covariance matrix into the simulation stage? Many thanks.
Kind regards,
Sherry
I am using bayesmh to estimate a model and would like to use a "continuous adaptation regime" as mentioned in two places of stata manual (screenshots attached at the end). The continuous adaptation regime will allow the adaptation of proposal covariance matrix beyond the burn-in stage and into the simulation stage.
I have searched the manual and did not find a way to invoke such a regime, but remember finding it somewhere online that this can be achieved by setting the tuning parameter gamma in adaptation(gamma( )) at a positive number. I will also need to override the maximum adaptation allowed in adaptation(maxiter( )) to be (# of burn-in + # of simulation)/every( ).
Could someone please confirm or disconfirm this approach, or provide other ways to allow adaptation of proposal covariance matrix into the simulation stage? Many thanks.
Kind regards,
Sherry
Attached Files
