ppap4lmp

PostProcess and Analysis Program for LAMMPS

Documentation (generated by Doxygen)

• GitHub Pages

Documenting

1. Checkout to gh-pages branch.
2. Remove docs/ from .gitignore.
3. Run ./documentation.py.
4. To publish the documentation, push docs/ directory as origin/gh-pages.

   git subtree push --prefix docs origin gh-pages

Note that local and remote gh-pages branch are different.

Features

• Implemented by C++, Used from Python
• Data stored in Json structure
• Extensible without editing existing codes

Installation

via Conda

conda install -c irista56 ppap4lmp

Development version

Just clone this repository and pip install. Note that Eigen is required.

git clone --recursive https://github.com/irisTa56/ppap4lmp.git
cd ppap4lmp
pip install .
# (or) pip install -e .
# (or) pip install --user .
# (or) pip install --user -e .

Test your installation.

cd tests
tar xzvf archive.dumps_bead.tar.gz
tar xzvf archive.dumps_atom.tar.gz
python test.py

Tests can be conducted separately.

python test.py starter # filter, adder, processor

Development version in Mac

If you prefer Homebrew

1. install dependencies

# install compiler and build tool
brew install gcc cmake
# install dependent libraries
brew install eigen pybind11
# install runtime dependency
brew install python; pip install numpy

2. get & build source

git clone --recursive https://github.com/irisTa56/ppap4lmp.git
cd ppap4lmp
export CXX=g++-8
pip install .
# (or) pip install -e .
# (or) pip install --user .
# (or) pip install --user -e .

3. run tests!

If you prefer Conda

1. install dependencies

conda install -c conda-forge \
  clangxx openmp cmake eigen pybind11 nlohmann_json numpy

2. get & build source; you do not need submodules

git clone https://github.com/irisTa56/ppap4lmp.git
cd ppap4lmp
export CXX=clang++  # Conda's clang++
export CXXFLAGS="-I/path/to/conda/include -L/path/to/conda/lib"
# (example) export CXXFLAGS="-I$HOME/miniconda3/include -L$HOME/miniconda3/lib"
pip install .
# (or) pip install -e .
# (or) pip install --user .
# (or) pip install --user -e .

3. run tests!

Profiling

Install a profiling tools library.

conda install gperftools  # need conda-forge channel
pip install yep  # libprofile should be found

Append the below line to CMakeLists.txt.

target_link_libraries(_ppap4lmp PRIVATE profiler)

Build & install.

Profiling tests

Run the below command instead of python test.py.

python -m yep -- test.py

Profile interactively.

# If you used Conda's Python
pprof $CONDA_PYTHON_EXE test.py.prof

Acknowledgement

This project would not be possible without the following great open-source projects.

• pybind/cmake_example
• pybind11
• Eigen
• nlohmann/json

Known Bugs

• Error in 'python': double free or corruption (out): occurs when exiting from Jupyter.