Computing deltaF (with error bar) after reweighting

223 views
Skip to first unread message

casalini...@gmail.com

unread,
Feb 22, 2024, 10:35:20 AM2/22/24
to PLUMED users
Dear all,
I run three independent well-tempered metadynamics simulations and I would like to obtain a free energy surface and the deltaF values between two known states. 
In order to do this, I am trying to use the Tiwary/Parrinello reweighting scheme. 
This is what I have done so far:

1) I have combined the COLVAR files removing the first half of the simulation (so that I am postprocessing the data where c(t) is somehow linearly increasing).
2) I computed histograms using appropriate weights.
3) Using the script provided in the Lugano tutorial, I computed the size of the average error bar as a function of block length and I identified a suitable block length. 
4) I computed the FES with the error bars using the same script. 

My question is: how can I compute a meaningful value of the dF between two known states along with an estimation of the error?

I thank you for your help and I apologize if you have already answered this question (I had a look but I could not find some suggestions about this).

Tommaso

Michele Invernizzi

unread,
Mar 4, 2024, 4:42:12 AM3/4/24
to plumed...@googlegroups.com
Hi Tommaso,

To compute the dF you should integrate the FES over the respective basins, see for example eq. S12 in the SI of https://pubs.acs.org/doi/10.1021/acs.jpclett.0c00497
In the GitHub repo linked to that paper you'll find a python script that computes it, option --deltaFat  https://github.com/invemichele/opes/blob/master/postprocessing/FES_from_Reweighting.py#L46 (there is also an option --reverse, that I find useful to see the effect of different truncation choices on the dF estimate)

The uncertainty estimation is a little delicate, when possible I personally prefer to run multiple independent simulations rather than relying on block average or bootstrapping, see e.g. https://github.com/invemichele/OPES-explore/blob/main/mueller/mueller.ipynb

Hope this helps,
Michele

--
You received this message because you are subscribed to the Google Groups "PLUMED users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to plumed-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/plumed-users/5df9385a-f577-461b-b04e-06132743377an%40googlegroups.com.

Tommaso Casalini

unread,
Mar 5, 2024, 8:00:43 AM3/5/24
to plumed...@googlegroups.com
Dear Michele,
thanks a lot for your suggestions! They helped me to achieve clearer overview.

Best,
Tommaso

Reply all
Reply to author
Forward
0 new messages