hdnnpy2phono3py.py is script to interfacing phono3py and hdnnpy
(high-dimensional neural network potential code, see
https://github.com/masayoshi-ogura/HDNNP (archived) or
https://github.com/eminamitani/hdnnpy-update (mostly same as the above but including minor update))
For prepare the input files to predict the force by hdnnpy
python hdnnpy2phono3py.py prep
In this preperation process, the job-script predictionRun.sh for hdnnpy calculation
is generated based on the information written in config file.
Then, submit hdnnpy job using jobscheduler.
For example
qsub predictionRun.sh
After that, run phono3py calculation by
python hdnnpy2phono3py.py run
In this script, two situations are considered.
- 2nd and 3rd order force constant is evaluated in the same size of supercell.
- different supercell size is used for 2nd and 3rd order force constant
These two cases can be specified by setting --dimfc2 in the config file as shown in below.
- config file
The name of config file forprepandrunisphono3py_config.yaml. The meaning of the tags in config file is
dim: [2,2,2] #supercell size for phono3py calculation
symfc: True #use Acoustic sum rule or not
prefix: crystal #prefix used in hdnnpy calculation
poscar: POSCAR #file name of POSCAR used in phono3py
dimfc2: [3,3,3] #if different supercell size for 2nd order FC case
mesh: [11,11,11] #mesh in phono3py-BTE calculation
pipfile: /Users/emi/PycharmProjects/hdnnpy+phono3py/Pipfile
#path for Pipfile of virtualenv to use hdnnpy and phono3py
nnpoutput: ../output # the path for the directory contain master_nnp.npz (hdnnpy output)
fc3output: ../fc3-output # the path for working directory for 3rd order FC calculation
fc2output: ../fc2-output # the path for working directory for 2nd order FC calculation
mpicommand: mpirun -np 24 #command fo mpi
strpa: [0,1/2,1/2,1/2,0,1/2,1/2,1/2,0] #translation matrix to conventional to primitive cell, optional. used in phono3py
#the header part of job script in your system
jobscriptheader: |
#!/bin/csh
#$ -cwd
#$ -V -S /bin/bash
#$ -N prediction
#$ -o stdout
#$ -e stderr
#$ -q all.q
#$ -pe smp 24
export OMP_NUM_THREADS=1
root=$PWD
MPIROOT=/usr/local/openmpi-1.8.8/
export PATH=$MPIROOT/bin:$PATH
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$MPIROOT/lib
export MANPATH=$MANPATH:$MPIROOT/share/man
If dimfc2 is not empty, the calculation for 2nd order FC is prepared.
in mac OS + zsh
git clone https://github.com/pyenv/pyenv.git ~/.pyenv
echo 'export PYENV_ROOT="$HOME/.pyenv"' >> ~/.zshrc
echo 'export PATH="$PYENV_ROOT/bin:$PATH"' >> ~/.zshrc
echo 'eval "$(pyenv init -)"' >> ~/.zshrc
In my case, install 3.7.6 by
pyenv install 3.7.6
Minor update for newer version of several python package is inplemented in initialize branch. This, after clone, checkout to that branch is recommended.
git clone [email protected]:eminamitani/hdnnpy-update.git
git fetch
git checkout -b initialize origin/initialize
This package include Pipfile. (The Pipfile.lock is also included, but some of library is old. This, it may be better to remove Pipfile.lock.)
pipenv --python 3.7
pipenv install --dev
Activate virtual environment by pipenv
pipenv shell
Then, install phonopy
pip install phonopy
Clone phono3py source package. The development branch version requires phonopy version >2.7. But pip install phonopy install the stable version of 2.6. Thus, here I checkout to the master branch after clone.
git clone https://github.com/phonopy/phono3py.git
git fetch
git checkout -b master origin/master
new version of gcc is installed by Macport
sudo port install gcc6
sudo port select --set gcc mp-gcc6
sudo port install OpenBLAS +gcc6Then, make a minor change in setup.py if necessary.
In gcc on Mac installed by Macport,
I encounter compile error caused by openmp related option.
So, I removed before compile
extra_compile_args = [ ]
if cc == 'gcc' or cc is None:
lib_omp = ''
python setup.py build
pip install -e .
In Linux, it is better to make setup_mkl.py
and use intel MKL to avoid compile error.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent