Open amin-sagar opened 4 months ago
Thanks for the detailed report. I can tell that the warning message is coming from pymbar
. What version are you using of pymbar
in your environment? We have seen some instabilities with pymbar 4 and the jax
backend to do the estimates (see for example in https://github.com/choderalab/pymbar/issues/419#issuecomment-1955191199 ). Maybe using the robust solver or downgrading to pymbar 3 could do the trick, if that makes sense.
Thanks @ijpulidos I am using
pymbar 4.0.3 hff52083_1 conda-forge
pymbar-core 4.0.3 py310h1f7b6fc_1 conda-forge
I was looking at the tutorial and code and I am sorry I couldn't find how to specify the robust kwargs in the simulations script. Do I need to modif the [openmmtools/openmmtools/multistate file?
I have noticed that this seems to happen if I have too many replicas, for example n_replicas = 25 min_Temp = 300 max_Temp = 400
doesn't work but
n_replicas = 22 min_Temp = 300 max_Temp = 600
works.
I would be really grateful if you can help me in including the robust analysis protocol in the simulation script.
Hello. I am trying to run some Temperature replica exchange simulations for small molecules. The relevant part of my script is as follows.
The function to create temperatures is from cg_openmm repo
On running the simulations, I get many lines like the following
However, if I use a linear temperature ladder
I don't get any such errors and the simulation finishes as expected. Also, with linear scaling I get no errors with 10 temperature points in the same temperature range while with exponentially distributed temperatures, I see these errors with even 25 temperature points.
I am quite sure I am doing something wrong but I can't figure it out. I will be really grateful for any suggestions.
Best, Amin.