Mh_jscale changed with new draws

Dear All,
I was wondering about the following question: I am trying to ad additional draws to a successfully completed RWMH run, using the load_mh_file option. However, by mistake I changed the mh_jscale parameter to quite a different value in the estimation command. I didn’t notice this because the acceptance ratio actually settled down very close to the average I obtained for the original chains I had completed. So my question is: When using load_mhfile, does dynare simply ignore the mh_jscale setting, and continues to use the original value? If so I am wondering where it is stored. I looked at the mh_history file but couldn’t find anything that looked like the mh_jscale value.

Many thanks
Ansgar

The record.ProposalScaleVec vector in _mh_history_*.mat should provide the value of the mh_jscale. Could you please report back what you find.

Hi Johannes,
ah I see. I checked the mh_history_1.mat file. This was the file produced when I did the additional draws. The value there equals the setting of the mh_jscale parameter in my first run (actually it’s vector with 67 times the same value). I also checked the mh_history_0.mat file (the file from the first run) and found the same value again.

So I take it that means that when using mh_recover dynare ignores any manual mh_jscale setting by the user, and simply relies on the most recent mh_history_* file?

Thank you for your help, and apologies I didn’t see this by myself…
Best,
Ansgar