[lammps-users] Simulation results is same output on changing LJ parameters

ABHISHEK KUMAR <[email protected]>

Attachments

visibility_off
Sun, Jun 13, 6:54 PM (23 hours ago)


to lammps-users

Dear LAMMPS users,
I am using the 7Aug19 version of LAMMPS. The molecular geometry for cubic tip4p/ew water box with a flat silicon surface was generated using Moltemplate.

While running the simulation I am not getting any error but when I changed the Lennard-jones parameter the simulation result is not changing at all. On looking into the log file I suspect the step(given below) to be the error.
]3 neighbor lists, perpetual/occasional/extra = 3 0 0 (1) pair lj/cut/tip4p/long, perpetual, skip from (3) attributes: half, newton on pair build: skip stencil: none bin: none (2) pair lj/cut, perpetual, skip from (3) attributes: half, newton on pair build: skip stencil: none bin: none (3) neighbor class addition, perpetual attributes: half, newton on pair build: half/bin/newton stencil: half/bin/3d/newton bin: standard.

I am unable to grasp how to remove this error. I am attaching the run file(run1.in.nvt), log file(slurm-41196.out), and the relevant data files as zip.
I would really appreciate any help from you regarding the error.

Abhishek Kumar Agarwal

Yours Sincerely
Abhishek Kumar Agarwal
Fourth Year Undergraduate
Department Of Chemical Engineering
Indian Institute of Technology Kharagpur

Contact: +91 - 8317865765

[email protected]
[email protected]
IIT Kharagpur

input.rar (87.8 KB)

ABHISHEK KUMAR <[email protected]>

visibility_off
Sun, Jun 13, 6:54 PM (23 hours ago)

to lammps-users

Dear LAMMPS users,
I am using the 7Aug19 version of LAMMPS. The molecular geometry for cubic tip4p/ew water box with a flat silicon surface was generated using Moltemplate.

While running the simulation I am not getting any error but when I changed the Lennard-jones parameter the simulation result is not changing at all. On looking into the log file I suspect the step(given below) to be the error.
]3 neighbor lists, perpetual/occasional/extra = 3 0 0 (1) pair lj/cut/tip4p/long, perpetual, skip from (3) attributes: half, newton on pair build: skip stencil: none bin: none (2) pair lj/cut, perpetual, skip from (3) attributes: half, newton on pair build: skip stencil: none bin: none (3) neighbor class addition, perpetual attributes: half, newton on pair build: half/bin/newton stencil: half/bin/3d/newton bin: standard.

this section is just reporting the setup for neighbor list construction and there is nothing inherently wrong with it.

what is not clear from the information you provide is a) where you made changes and which changes, and b) how you determined that those changes had no impact.

axel.

also, there is not data file contained and your system.in.settings file contains the same settings 8000 times.
this strongly suggests, that you need rather basic help with setting up your simulation and should talk to your adviser about this.
perhaps, it would be better if you first practice with setting up some smaller and simpler simulations (like liquid argon) and build your skills in using LAMMPS and your experience with how the different parts of the input work.

axel.

To LAMMPS users,
Thank you for your response and time. I will look forward to improving upon your advice.
Coming to the changes, In the run1.in.nvt file run section: line 28
pair_coeff 1 3 lj/cut 3 2.95 9.0
I am changing the interaction parameter(shown in bold) between 1 (oxygen) and 3 (silicon) and running the simulation. The values used for different simulations is 2 kcal/mol and 3 kcal/mol. I visualized the dump file( run1.in.nvt line 54: dump 1 all custom 5000 full03_new.lammpstrj id mol type q x y z) using VMD and got the same simulation. I was expecting higher wetting for a higher value of the parameter but that was not the case. I am attaching a screenshot( left -2 kcal/mol, right - 3kcal/mol ) of the final configuration for reference.
It would be really helpful if you could provide your insight on it.

Yours Sincerely
Abhishek Kumar Agarwal
Fourth Year Undergraduate
Department Of Chemical Engineering
Indian Institute of Technology Kharagpur

To LAMMPS users,
Thank you for your response and time. I will look forward to improving upon your advice.

you should have asked for this advice from your adviser/supervisor. this is really more an issue of basic skills of how to conduct a scientific study rather than anything LAMMPS specific.

I don’t have the time to fill in for your adviser/supervisor, so I will respond this one time. but in the future you have to argue your case more convincingly

Coming to the changes, In the run1.in.nvt file run section: line 28
pair_coeff 1 3 lj/cut 3 2.95 9.0
I am changing the interaction parameter(shown in bold) between 1 (oxygen) and 3 (silicon) and running the simulation. The values used for different simulations is 2 kcal/mol and 3 kcal/mol. I visualized the dump file( run1.in.nvt line 54: dump 1 all custom 5000 full03_new.lammpstrj id mol type q x y z) using VMD and got the same simulation. I was expecting higher wetting for a higher value of the parameter but that was not the case. I am attaching a screenshot( left -2 kcal/mol, right - 3kcal/mol ) of the final configuration for reference.
It would be really helpful if you could provide your insight on it.

a visualization is a very qualitative measure and totally unsuitable to support your claim. while those images look similar they may still be representing accurate results. that are certainly not identical like you are claiming.

since you are doing a computer simulation it is actually very simple to determine whether changing the parameter has an impact on your simulation: there should be no change in the potential energy also the two trajectory files themselves would have to have identical coordinates if there would be no change from the change of the parameter, but that is obviously not the case. even whether there is change in wetting behavior can be easily quantified by computing a density profile for each simulation and carefully comparing the results.

now, whether your specific system under the specific conditions you are simulating should show a more visible change in behavior is a question of your research and not a LAMMPS issue, which makes it definitely a topic for discussion with your adviser/supervisor.

axel.

1) I agree with Axel's suggestion to look at the potential energy in
the log.lammps files for the two simulations. If there is a change
in the interaction strength between the water and the silicon, you
should see a change in the non-bonded energy ("epair") or the
potential energy ("pe"). It seems likely to me that there is a
change, but it does not manifest itself visually.
2) I am also surprised that you are using such a large epsilon value
for the interaction between Si and O (from the water).
3) Is it realistic to have water directly touching a surface of pure
silicon? I am not a chemist, but I remember seeing several examples of
water on an SiO2 surface.
4) Regarding the repetition in the "system.in.settings" file, it looks
like you put these parameters in a
write("In Settings") { ... } statement, instead of a
write_once("In Settings") { ... } statement.
That's why you have all the "_coeff" commands repeated thousands of times.
Moltemplate used to flag this as an error, but I changed it so it no
longer complains anymore.
(Perhaps I should change it back.) Either way, it doesn't do any
harm (other than causing LAMMPS to spend a few extra seconds reading
your file at the beginning of the simulation.

Cheers

Andrew

To Andrew
Thank you for your response.
Regarding your second point, I wanted a super hydrophilic surface here but I came to know this is a bit too high epsilon.
Regarding your fourth point, I realized it after Axel’s suggestions and I have rectified it.
The fundamental question in your third point needs discussion with the advisor.
The problem I had before has receded after such changes. I hope to learn and improve upon it.
Thank you for your time.