[ase-users] LAMMPS in ASE

[Efrem Braun]

I *think* I've actually gotten it figured out. The energy and forces I get
out of ASE are the same as what I'm getting out of the direct LAMMPS
simulation. I don't have the dynamics right yet because I don't have the
masses set correctly in ASE, but I should be able to figure this out on
Sunday. I'll shoot an update with some documentation then.

I did need to make one small update to lammpslib.py to get this working
(since I'm using create_atoms=False, which I guess hasn't been used much
before), so I sent a pull request in for that.

Efrem Braun

On Fri, Mar 16, 2018 at 1:57 PM, Noam Bernstein <noam.bernstein at nrl.navy.mil
> wrote:

> On Mar 16, 2018, at 4:23 PM, Efrem Braun via ase-users <
> ase-users at listserv.fysik.dtu.dk> wrote:
>
> Very happy to hear that LAMMPS is being supported! I switched over to the
> development branch from the Gitlab repo, but unfortunately I'm still having
> the same issue. Running "ase test" gives me "OK (skipped=71)", "ase test -c
> lammpsrun" gives me  "OK (skipped=68)", and "ase test -c lammpslib" gives
> me the same "MPI_ABORT was invoked on rank 0 communicator
> MPI_COMM_WORLD..." error message that I was getting before. Like I
> mentioned, I do have LAMMPS working in Python, so I don't know what I could
> be doing wrong.
>
> Assuming I'm able to get this fixed, I'd like to ask my more specific
> question...The LAMMPS potential I'm trying to run has
> bonds/angles/dihedrals defined. The actual LAMMPS input file reads:
>
> atom_style      full
> boundary        p p p
> pair_style      lj/cut/coul/long 12.500
> bond_style      harmonic
> angle_style     hybrid fourier cosine/periodic
> dihedral_style  harmonic
> improper_style  fourier
> kspace_style    ewald 0.000001
> dielectric      1.0
> pair_modify     tail yes mix arithmetic
> special_bonds   lj/coul 0.0 0.0 1.0
> box tilt        large
> read_data       data.input
> ...(the actual running follows)
>
> The "data.input" file is quite similar to what's found in
> https://gitlab.com/ase/ase/blob/master/ase/test/lammpsdata/lammpsdata.py;
> the only difference is that there are more atom types and the velocities
> are not specified.
>
> I don't think that lammpsrun will work for this, because I see no way to
> feed a data file in. So lammpslib is probably the only way to go, which is
> why I've been trying to get it to work. Obviously I'd prefer to figure this
> out on my own before asking for help, but since I can't get lammpslib
> working, I can't yet do this. And if I'm told that there's no way to do
> this with lammpslib anyway, then it'd be nice to know that I shouldn't
> waste time trying.
>
>
> I haven’t been keeping up with what version of lammpslib is currently
> being included in ASE, but there is definitely a version some of us have
> been working on that can create bonded potentials from the ASE atoms
> structure (using some lammps syntax that has been added recently, namely
> “create_bonds single/bond” and the like). I’ll ask the people who (I think)
> were responsible for the recent improved capability lammpslib and see if
> there’s a move to add those features as well.  Impropers will need to be
> added for you, but otherwise it should probably work, and I’m pretty sure
> there’s a branch someplace that has impropers already, just need to merge
> it back in.
>
> Noam
>
> ____________
> |
> |
> |
> *U.S. NAVAL*
> |
> |
> _*RESEARCH*_
> |
> LABORATORY
>
> Noam Bernstein, Ph.D.
> Center for Materials Physics and Technology
> U.S. Naval Research Laboratory
> T +1 202 404 8628 <(202)%20404-8628>  F +1 202 404 7546 <(202)%20404-7546>
> https://www.nrl.navy.mil
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20180316/5ff228d5/attachment.html>