DPFlow - A workflow to generate deep potential force field

---

DPFlow is a code to:
(1) Optimize neural network force field
(2) Analyze CP2K and LAMMPS trajectory

The code is mainly written in Python and Fortran.

Email: lujunbo15@gmail.com

Installation for DPFlow

• Prerequisites

  • Python 3.5 or higher
  • Numpy 1.8.0 or higher

  Suggestion: as we will use deepmd-kit software, we recommend users could
  install deepmd-kit at firsti: https://github.com/deepmodeling/deepmd-kit.
  Then add the environmental variable of deepmd-kit, it will include python
  and numpy.

• Compile core module

  git clone https://github.com/JunboLu/DPFlow.git
  !Caution: When downloading DPFlow through zip version, please change the
  name "DPFlow-main" to "DPFlow" after you unzip it.
  cd DPFlow_directory/lib
  change directory of f2py in Makefile
  !Caution: If your gcc version is low, f2py cannot compile core code please
  update your gcc up to 6.3.
  make

• Environmental variable

  !Caution: You may need to add environments of gcc
  export PYTHONPATH=../DPFlow_directory:$PYTHONPATH
  change python_exe and DPFlow directory in DPFlow_directory/bin/DPFlow file
  export PATH=DPFlow_directory/bin:$PATH

How to use

• DPFlow is an user-friendly code.

  The input files are in example directory.
  Users just need to run:
  DPFlow input.inp