Merging windows in eabf

LAMMPS
1

Dear all,
While using ABF, InputPrefix is used to merge windows and get a free energy profile. Is the same suggested for eABF?
I used InputPrefix to merge windows and get a free energy profile, but as mentioned in the documentation there are visual blips in it. I tried the recommended solutions, taking values for the extreme points in a window from .grad file and/or interpolate the end points in .zgrad file. But the free energy profiles I got from these two and the one I get if I do not do any changes are slightly different. Different in the sense that now free energy difference of bounded and unbounded state has increased by 0.05-0.2 kcal/mol (Order is like this: No changes < Interpolation < taking values from .grad file). Therefore I am confused which method should I use here to get the free energy profile?
Or shall I use slightly overlapping windows ?

Thanks,
Megha

2

@giacomo.fiorin
Can you please suggest me a solution for this?

Thanks,
Megha

3

Hi @Megha thanks for pointing me to it.

Yes, a slight overlap is generally desirable, even if just for a consistency check. Due to how diffusion works, you do not lose much at all in sampling quality by making the windows only slightly wider.

As for the extended-system / CZAR part of the problem I’ll point Jérôme to your message so that he can reply to you directly.

Giacomo

4

Dear @giacomo.fiorin
Thank you so much your reply and letting Jérôme also know about this!

Thanks,
Megha