LAMMPS codes for DFTB+

Reference: Gang Seob Jung, Stephan Irle, and Bobby Sumpter. "Dynamic aspects of graphene deformation and fracture from approximate density functional theory." Carbon (2022)

Please check DFTB+ code https://github.com/dftbplus/dftbplus

This package provides a fix dftbp command which is a wrapper on the DFTB+ DFTB code, so that molecular dynamics can be run with LAMMPS using density-functional tight-binding calculations by DFTB+.

GS JUNG@ORNL made this package based on fix_latte.cpp and fix_latte.h (developed by LANL)

Please report any bug/commetns to [email protected] or [email protected]

The necessary functions for DFTB+ API have been implemented (update is allowed by dftb+).

Until that, you can check the details of changes in https://github.com/gsjung0419/dftbplus

0. The code is only tested with following environments.

-. Intel compiler (19 and 20)

-. MPICH 3.2.1 version compiled by intel compiler

-. CMAKE 3.12 or 3.19

-. Matching the versions of programs are recommended (The author did not check other possible combinations).

1. DFTB+ options (tested) (See your confg.cmake in dftb+)

-. WITH_OMP: TRUE or FALSE

-. WITH_DFTD3: TRUE or FALSE

-. WITH_API: TRUE (Mandatory for static library, e.g., library.a)

-. BUILD_SHARED_LIBS: TURE (Mandatory for shared library, e.g., library.so, mandatory for python-based lammps users)

-. You may have compile issues if you try with other environments.

2.1. Installation of DFTB+ (recommended for the most fundamental version)

-. Compile the dftb+ (WITH_OMP, WITH_API: TRUE)

-. Check libdftbplus.a is generated.

-. Locate libdftbplus.a in preffered folder (ex., ~/applic/lib/dftbplus)

2.2. Installation of LAMMPS (recommended for the most fundamental version)

-. Use LAMMPS version 29OCT20, the newest version did not work for "fix" based modification.

-. Copy lib/dftbp to LAMMPS_sourcecode/lib

-. Edit Makefile.lammps.mpi in lib/dftbp, "location of library" you set in DFTB+ installation (ex., ~/applic/lib/dftbplus).

-. Move src as lammps/src/DFTBP

!! Make sure your dftbplus.h is the same as provided from dftbplus (there might be updates for API functions).

-. Edit Makefile in lammps/src folder;include "dftbp" in PACKAGE.

-. Check "make package-status" whether you have DFTBP package

-. Make available with "make yes-dftbp"

-. Compile lammps with "make mpi" in src folder (compile with intel compiler + mpich3.2.1)

3 Usage and Example

!! Before you are running the simulation. Check OMP_NUM_THREADS= 1 or specify a number

!! Without specified the OMP number, DFTB+ tries to use all available cores, making it slow.

3.1 Graphene md (example C-C.skf is MIO)

-. data file for lammps and dftb+ should be prepared separately.

-. The order of input atoms should be matched.

-. Check dftb_in.hsd for options for dftb+. Most options are turned off except force calculations