Decomposing total PMF into vdW and coul parts - regd

LAMMPS LAMMPS Mailing List Mirror
1

Dear all,

I have used USER-COLVAR module in LAMMPS to generate PMF profile. Now I want to decompose the total PMF profile into vdW and coulombic contributions. I have tried the same thing by using “rerun” on full interactions PMF trajectory just by changing pair_style buck (which was buck/coul/long in original run) and have used the following fix colvars (which is same in both runs) “fix 1 all colvars abf.colvars output distance_cv”
But couldn’t succeeded. I am getting the error “colvars: Error: trying to allocate twice the collective variable module”.
Here my doubt is does rerun support fix colvars, with modified pair_style and I am wondering whether is there any better way to do this in LAMMPS.

Thank you in advance.

2

Dear all,

          I have used USER-COLVAR module in LAMMPS to generate PMF profile.
Now I want to decompose the total PMF profile into vdW and coulombic
contributions. I have tried the same thing by using "rerun" on full
interactions PMF trajectory just by changing pair_style buck (which was
buck/coul/long in original run) and have used the following fix colvars
(which is same in both runs) "fix 1 all colvars abf.colvars output
distance_cv"
But couldn't succeeded. I am getting the error "colvars: Error: trying to
allocate twice the collective variable module".
Here my doubt is does rerun support fix colvars, with modified pair_style
and I am wondering whether is there any better way to do this in LAMMPS.

yes, fix colvars should be compatible with rerun. there must be an
error in your input or you may have a very outdated LAMMPS version. if
you want somebody to take a closer look, you have to provide more
information.

axel.

3

Dear Axel,

Thank you very much for your reply. I have updated my LAMMPS version and now I am able to perform rerun, but in my “out.pmf” file all PMF values are zero.
Here I am wondering whether can one do PMF calculations by using rerun, as these calculation might require WHAM (Weighted Histogram Analysis) kind of averaging, this may require MD integrator ( I am not sure ! ).
Anyhow here I am sending my input files can you please have a look

units real
dimension 3
boundary p p p
atom_style full
read_restart NPT_eq_300k.restart
include parm_LAMMPS_XL
neighbor 2.0 bin
neigh_modify every 1 delay 5 check yes
fix 1 all colvars abf.colvars output distance_cv
rerun PMF_run.dcd first 0 last 15000000000 every 0 dump x y z box yes scaled yes format molfile dcd .:/vmd/vmd-1.9.1/plugins/LINUXAMD64/molfile

COLVARS file

colvarsTrajFrequency 100
colvarsRestartFrequency 1000
colvar {
name Translocation
width 0.1
lowerboundary 2.0
upperboundary 9.0
lowerwallconstant 100.0
upperwallconstant 100.0
distance {
group1 {
atomNumbers 5650
}
group2 {
atomNumbers 5657
}
}
}

abf {
colvars Translocation
fullSamples 200
}

Thank you in advance,

Best,

Ramesh

4

I think you should first make clear to yourself and to the mail list what method you are using for calculating a PMF.

For example, you mention WHAM but have ABF defined in the input, and these are totally unrelated methodologies.

5

Dear Giacomo Fiorin,

Thank you very much for your reply. Actually I am using ABF for generating PMF profile. Yeah I know that ABF & WHAM are entirely different methodologies. Here my intention to mention WHAM is to say that a protocol similar to WHAM (not exactly WHAM, something to get meaningful exp<-H/kT>) might be used in ABF to generate PMF. This may require a MD run rather than rerun (post processing might not possible).

Thank you,

Best,

Ramesh

6

Again, you’re explaining what you’re looking for.

The contribution of the Coulomb potential to the PMF is the Coulomb potential itself. You can recalculate it yourself with a rerun, with or without colvars.

If you are looking for anything more than that, you haven’t made that clear yet.

Giacomo

7

Dear Giacomo