Hi all!
I’m trying to calculate the pressure of a system composed by 60 atoms (20 molecules of CO2). I fixed T=300K. From NIST database, knowing the density of the system from the simulation, I know that I should get from LAMMPS a pressure of 3 atm. But during my fix NVE pressure fluctuates from 9 to 2 atm.
Does anyone know how to reduce fluctuations?
Please find below my input and data files.
INPUT FILE:
units real
dimension 3
boundary p p p
atom_style full
read_data data_CO2atomsforP.txt
Hi all!
I’m trying to calculate the pressure of a system composed by 60 atoms (20 molecules of CO2). I fixed T=300K. From NIST database, knowing the density of the system from the simulation, I know that I should get from LAMMPS a pressure of 3 atm. But during my fix NVE pressure fluctuates from 9 to 2 atm.
Does anyone know how to reduce fluctuations?
if you search through the lammps-users mailing list archives, you will see this explained many times, or you can read up on the underlying theory in a statistical mechanics text book. your available options are to average or increase the system size.
axel.
1 Like
oh, and never EVER use fix temp/rescale in a production simulation. again, this has been explained many times on this list.
axel.
one more post-scriptum. it’ll be the last… i promise…
when you are using fix temp/rescale, you are, of course, not in an NVE ensemble anymore. in fact, you are not in any defined ensemble. axel.
Thank you very much for your reply and warning. Unfortunately without temp / rescale the temperature becomes too high and not equal to 300K as I would like. Can you suggest an alternative? Thanks again.
Kind regards
Beatrice
a) you first have to equilibrate your system with a thermostat, say fix nvt or fix langevin + fix nve
b) if you want to run in NVE you have to remove the thermostat with fix nve only. if the temperature still increases (too) much, you need to rethink your simulation settings (e.g. time step, cutoff etc.) as your simulation is not conserving energy and you cannot have meaningful results with a system not conserving energy either.
using fix temp/rescale to suppress the undesired change in kinetic energy is effectively cheating.
you probably want to talk about how to run your simulations with somebody that has some more experience.
axel.
1 Like
Thanks a lot for your suggestions. I tried doing what you suggested but actually with NVE+fix langevin (run 1million - timestep 1fs) and then NVE without thermostat I still see pressure fluctuations from positive to negative values, even after averaging. And pressure value is far from the expected value of 3 atm, it fluctuates from -30 atm to +30 atm. Can anyone suggest me a solution?
Please find below my new input script.
Thanks again for your help. I wish I could ask my supervisor, but he doesn’t know LAMMPS well enough.
Kind regards,
Beatrice
units real
dimension 3
boundary p p p
atom_style full
read_data Data_CO2atomsforP.txt
Thanks a lot for your suggestions. I tried doing what you suggested but actually with NVE+fix langevin (run 1million - timestep 1fs) and then NVE without thermostat I still see pressure fluctuations from positive to negative values, even after averaging. And pressure value is far from the expected value of 3 atm, it fluctuates from -30 atm to +30 atm. Can anyone suggest me a solution?
Please find below my new input script.
Thanks again for your help. I wish I could ask my supervisor, but he doesn’t know LAMMPS well enough.
this is not a LAMMPS problem, but an “understanding MD simulations” (or more precisely “understanding statistical mechanics”) issue. if your supervisor cannot help you, you need to request tutoring from somebody that can. it is the responsibility of your supervisor to provide you with sufficient guidance and advice. not the job of this mailing list.
I already provided suggestions for how to reduce the fluctuations of your output. they still apply. if you don’t get a converged value, either the averaging is not sufficient and/or the system size is too small.
axel.
I want to suggest you to use nvt and npt equilibrations at first. I recommend at least 10 to 100 ps equilibrations. Your system needs equilibration. Please, consider your time constant during the equilibrium depending on your length of equilibration. Then, you should try your desired productions.
-Mijiddorj