Dear All,
I have been running CosmoMC with five chains for the same dataset for 4 days. I noticed that two of five chains last modification date was just on the first day while the other three get updated occasionally. However, one of the chains seems to progress must faster than the rest. Just to make sure you understand what I am talking about I will give you the number of steps for each chain:
chain 0 - 182
chain 1 - 207
chain 2 - 46786
chain 3 - 29340
chain 4 - 20480
Runtime 4 days 1 hour approx.
My questions are:
1) This is kind of chain progression normal?
2) If this is not normal, should I stop and rerun the chain again, possibly with some snapshot? How do I fix this in the ini file.
Best,
COSMOMC - some chains do not progress at all.
-
- Posts: 12
- Joined: May 03 2021
- Affiliation: Naresuan University
-
- Posts: 1973
- Joined: September 23 2004
- Affiliation: University of Sussex
- Contact:
Re: COSMOMC - some chains do not progress at all.
It is not normal, you can usually tell if chains are running OK within a couple of hours (and stop if not). You can certainly generate a covmat from that run and re-start. Check your propose and start ranges for each parameter.
-
- Posts: 12
- Joined: May 03 2021
- Affiliation: Naresuan University
Re: COSMOMC - some chains do not progress at all.
Thank Anthony,
I have checked it. There is nothing wrong with the parameter range and the initialization file. It is more to do with the server itself - probably with the MPI setting on the server. I run the same initialization file on another machine and it work fine.
I have checked it. There is nothing wrong with the parameter range and the initialization file. It is more to do with the server itself - probably with the MPI setting on the server. I run the same initialization file on another machine and it work fine.