Create bulk material from a single molecule data file

[wswgG]

Dear Axel,
thanks for your reply! Greetings from Germany, Lower Saxony

I apologe in advance for my potentially silly discussion. I’ve been using LAMMPS for CG simulations of KG polymer models, and only recently started working with all-atom models with OPLS force-field. So, currently, I’m “rediscovering” LAMMPS so to say

@akohlmey

It is up to you, how large the box is. Just make it smaller.

AFAIU, replicate command doesn’t allow for choosing the simulation box size, does it? Thus, I believe, fix deform is the way to go, following your further statement :

@akohlmey

Then don’t use fix npt but fix nvt and fix deform.

Regarding create_atoms and molecule commands that you recommended :

@akohlmey

An alternative to using a data file would be to create a molecule file instead and then using create_atoms with the molecule.

I tried this way, but haven’t found it easy. I happened to find these two discussions :

how to prepare molecule file for lammps - LAMMPS / LAMMPS Mailing List Mirror - Materials Science Community Discourse

Molecule auto special bond generation overflow - LAMMPS / LAMMPS Beginners - Materials Science Community Discourse

and also carefully looked through this tutorial :

https://lammpstutorials.github.io/sphinx/build/html/tutorials/level2/polymer-in-water.html

and understood that molecule requires as an input not a LAMMPS data file, but a molecule template file. While preparing the latter one using the existing LAMMPS data file [ generated by LigParGen ] , I faced the problem : pair interactions are required for, e.g. write_data, delete_atoms, etc., but there is no section Pair Coeffs in LAMMPS molecule template file, and thus pair coefficients need to be defined in a LAMMPS input script. For my particular problem [ 57 different atom types ] , this means at least another 57 lines in the LAMMPS input script, and personally I don’t consider this convenient, or separate LAMMPS file force-field parameters and include command.

Alternatively, I’m looking into Moltemplate [ don’t find it easy to start with too though ] and Packmol.

Generally speaking, it’s hard to believe that my task that doesn’t sound complicated at first sight, i.e. generate a bulk material model out of the initial SMILES sequence or coordinates file that is already parametrised using MoSDeF or LigParGen, causes so much trouble.

Kind regards,
Igor.