Constant stress ensemble

Hi Dr. Axel,

Thank you for the reply :slight_smile:

I have tried every npt option which allows the shape change, but couldn’t see the relevant results as such as NST from Dlpoly. I have been working on this problem constantly and am worried a lot for not getting similar results.

I apologize for troubling you, Dr. Axel. But, thanks for the info. I am trying out my best to solve this issue.

I would greatly appreciate if you could let me know if which fix npt( that allows six component dimensions will be equal to)

fix npt 123 123 100 iso 1.0 1.0 1000

I mean which value of x,y,z,xy,yz,xz(if I want to use all six compoenet of fix npt tri x,y,z values) in fix npt will give the same as if we use iso 1.0 1.0 1000.

I studied literature as you suggested but still couldn’t figure it out.

Could you please let me know about it?

Please don’t be angry, Dr. Axel.:slight_smile:

Would appreciate atleast if you could give me some hint or documentation link.Thanks

if you get different results with LAMMPS and DL_POLY, that can have many reasons, not just the choice of integrator settings. and even if they are equivalent, you may not always get comparable results, if you have incomplete sampling and/or were starting from a metastable configuration, or for many other reasons.

empirically trying all permutations of a command, especially one with so many options as the nose-hoover integrator is not an effective way to resolve a discrepancy.
you have to approach this from a more fundamental level and first make certain, that all other settings are consistent and lead to equivalent results and are not dependent on specifics of the initial configuration (and you have to make certain, that those are really the same), this is all standard MD best practice stuff, and something that you should be discussion with your adviser or senior, more experienced colleagues. a mailing list is not a viable place to learn this. communication via e-mail is too cumbersome (starting by you still not being able to describe your issues and what you did and what system you tested it with). i am most certainly tired of having to second guess what you really mean and what important information you didn’t tell.
at this point is is probably also a approach to take a step back and try to set up and simulation a much simpler system and do this both in DL_POLY and LAMMPS side by side and learn where there differences in conventions or how you have to translate input settings from one code to the other. for as long as you are really doing equivalent things, you should get equivalent results (in the statistical mechanical sense).

in general, if you cannot figure this out for yourself, you need more training, but you won’t get it from e-mails.


Dear Dr. Axel,

My apologies for not being clear. I am actually performing constant stress ensemble on a monoclinic crystal structure and heating the 6x4x6 supercell system from 123K to the melting point. With the XRD experiment( to observe the both size and shape change), I observed that there is a constant increase in the volume from 123K to higher temperature, but while I am simulating the crystal starting from 123K, 173K, 203K … till melting point. The main issue is the volume from 123K started increasing with NST simulation but from 203K it decreased and then from 273K it started increasing which obviously is not replicating the experimental change. ​Please check the volume change a/c to temperature. as below. This is the problem I have been working with which I am unable to resolve. Any kind of help would be much appreciated Dr. Axel :slight_smile: Thank you :slight_smile:

Kindly fine the lammps input file in the attached.

Temperature simulated Volume (A3)

123 63.892​0681
173 64.9183587
203 64.4960577

223 63.2773359

253 63.1985739

273 63.5742503

input_nst.MDRun (1.96 KB)

sorry, but this is something where you need to learn a lot about how to model the process that you are looking at. that includes understanding the limitations of the specific force field parameters and model that you are using and also the restrictions of doing a simulation of a (very!) finite size system, especially on studying behavior near phase transitions from MD simulations in comparison to experiments. expecting that you will easily reproduce your experimental results is neglecting a lot of detailed and careful research into this topic from a lot of good people.

this has now reached far beyond the scope of this mailing list. i will not respond to this any further. as i mentioned before, this is a discussion you need to have with your adviser and experienced colleagues. there are no (secret?) flags to fix npt that will make your concerns (magically?) go away and provide you with the expected results.