# Temperature and RDF in NVE ensemble

Dear lammps user,

My lammps version is 7 August 2019.

I am trying to simulate the radial distribution function (rdf) for an NVE ensemble considering Lennard-Jones potential. I already have a verified result for the rdf of that system, to which I compare my rdf. The point is that the verified rdf was plotted close to the triple point of the Lennard-Jones system which is at reduced temperature T*=0.48 and at reduced density rho*=0.76 where the researchers used NVE simulation.

How to plot my rdf at temperature T*=0.48 for NVE ensemble?

I already know that the NVE ensemble represents replica of a system of constant number of particles N, constant volume V, and constant energy E. Also, I know that in the NVE ensemble, the system temperature changes over the simulation.

I already wrote the following lammps script. Then I plotted the rdf for the last timestep in my simulation whic is 6000 timestep. The figure shows the comparison between my rdf and the verified rdf. I have been told that there is still much hope to improve my rdf plot by plotting it at the right temperature.

I am forced to choose the timestep and the L-J cutoff distance as they appear in the script. My rdf plot was plotted by averaging over 20 runs of the script, each only differing in the random number in the velocity command. In each run, I did not consider computing time-average rdf as you can see in the fix ave/time command.

Could you please tell me how to plot my rdf at temperature T*=0.48 for NVE ensemble?
Also, I would be grateful if you can advise me how to further improve my rdf apart from chossing certain temperature to plot my rdf at ?

Correction:

sorry, but in the lammps script in the previous email, please consider this modification for the fix ave/time command

fix 2 all ave/time 1 1 6000 c_1[*] file tmp20.rdf mode vector

Regards

You are asking the wrong question here. It is not a matter of how to tweak the settings so that you get an identical plot, but you have to ask yourself, what is causing the differences? Only after understanding those you can think about what to change. Making arbitrary changes to somehow match the reference result could lead to using bogus simulation settings. From looking at the plot it appears that the two simulations (i presume those were done with different simulation software) may have used a different definition of sigma in the lennard-jones potential.
Axel.

Also, you haven’t resolved the issue that your simulation does not do any equilibration.
And the use of atom style sphere makes no sense with a lennard-jones potential.

axel.

Dear Axel,

I already understood your past advice about L-J potential. I completely realized that when using (atom style sphere) it will not make sense in terms of reducing the overlap between particles. I assert that for my simulation I need spherical particle interacting through L-J potential. I could reduce the overlap until it became acceptable for my work.

Regarding the equilibration, all I know about it is that it is the set up that leads the system towards equilibrium. I am considering it as well. The point is that every week my supervisors inform me to do new things, and there are no much time for every thing to be done at once.

I do not play with numbers inside LAMMPS to get the desired plot. As far as I know, I could consider all the relevant quantities that determine the verified rdf except for the Temperature. For sigma, I will try to read the paper I am simulating again and hopefully I could see new information about sigma.

However, I have become more curious about how to plot any plot in general at certain temperature in NVE ensemble ?. I do not know if my question is physically acceptable or not. On the other hand, I attached the original a picture of the verified rdf mentioned in the paper I am simulating. I aim at producing the rdf plot for 2-d L-J fluid.

Mohammed

Dear Axel,

I already understood your past advice about L-J potential. I completely realized that when using (atom style sphere) it will not make sense in terms of reducing the overlap between particles. I assert that for my simulation I need spherical particle interacting through L-J potential. I could reduce the overlap until it became acceptable for my work.

that makes no sense. pair style lj/cut will IGNORE the radius set for atom style sphere. you should get the EXACT SAME result with atom style atomic.

Regarding the equilibration, all I know about it is that it is the set up that leads the system towards equilibrium. I am considering it as well.

no, you don’t. you collect data for the ENTIRE simulation. you do NOT wait until the system is equilibrated.

The point is that every week my supervisors inform me to do new things, and there are no much time for every thing to be done at once.

this is a topic for discussion with your supervisor. if you don’t understand how to do things correctly, you have to ask our supervisor for additional instructions and advice. it is the job of your supervisor, NOT the job of this mailing list to teach you how to do research and MD simulations.

I do not play with numbers inside LAMMPS to get the desired plot. As far as I know, I could consider all the relevant quantities that determine the verified rdf except for the Temperature. For sigma, I will try to read the paper I am simulating again and hopefully I could see new information about sigma.

However, I have become more curious about how to plot any plot in general at certain temperature in NVE ensemble ?. I do not know if my question is physically acceptable or not. On the other hand, I attached the original a picture of the verified rdf mentioned in the paper I am simulating. I aim at producing the rdf plot for 2-d L-J fluid.

i understand that that is what you are doing, but please also understand that I have talked to MANY beginners and know many mistakes and misconceptions that happen to beginners (in part also through first hand experience). you are definitely confusing the process with the science.

you lack of fully understanding of the process of equilibration and how and when to collect data (and how much and how long) is a major problem here. at the moment and based on the input you provide, your data cannot be trusted. it is not at all obvious whether your system will be in equilibrium after 6000 MD steps. please also note, that systems near the critical point are subject to large fluctuations and thus large systems and long trajectories are usually required.

axel.