[lammps-users] question about verification process

Hi all,

I was curious if there was any sort of procedure for validation and/or verification of the LAMMPS code aside from the benchmark suite.



David E. Farrell

Graduate Student

Mechanical Engineering

Northwestern University

email: [email protected]…469…

Hi David,

It really depends on what type of systems you’re studying. For instance, if you were using eam materials, then you could try to calculate linear thermal expansion of an infinite crystal and compare to other simulations/experiments. In case of surfaces, you could try calculating phonon modes and compare with literature values. This serves two purposes, it validates the eam potential files and also the ability of lammps to calculate such things. I have found that lammps correctly models these things. Or a simpler case would be just to calculate density vs. temperature for the LJ fluid.

I hope this helps.


to verify the correctness of LAMMPS results for simple model systems, this
website might be interesting:


  -- Lutz

    Basically, some literature mentioned that Nose-Hoover doesn't apply to
small or stiff system, where it is not ergodic any more.
    There was an article by Martyna,
     J Chem Phy 97(4), 2635
    which presented "Nose Hoover Chain" algorithm to address this problem.
    I have done a "fix nvt" experiment for small polymer, using bead spring
model, and found it deviate significantly from equilibrium state at low
temperatures. Could this be the reason? Does Lammps NVT implement only exact
Nose-Hoover algorithm?


Nose/Hoover in LAMMPS is the original, not NH chains.

Someone could implement it though.



    In my simulation of small polymer chains, I found it ends up spinning
quite fast, so that I have to use command:
    fix 2 all momentum 100 angular
    It is actually the reason why small chain expand even at very low
temperature ( I posted the question to mail list before). This fix angular
momentum command address the problem.
    I wonder what "sideeffect" this command has. Also, why does the system
pick up angular momentum in an apparently random process?


A system with zero initial angular momentum (see an option on
the velocity command for this), shouldn't acquire it, and it shouldn't
increase. So I don't know why it is spinning, unless you have
some odd boundary condition that is adding angular momentum.


Thanks for your help. I understand now. It seemed I didn't zero the linear
and angular momentum when velocity is created. My boundary condition is "s"
( shrink ) in every direction.