Quick start and run

First of all, make sure you have all libraries (mentioned in prerequisites note below) installed. Then you can install the binary of NNimaker with pip. Detailed version of installation explained in install section (How to install).

Note

Prerequisite:

It is recommended to use virtual environment to install this package and its prerequisites. The following python libraries are needed before using this package:

pip install -U numpy pandas json collections glob. You also may need argparse and ase.

Mandatory inputs

  1. QE output

User have to provide a Quantum ESPRESSO output (PWSCF) as a necessary ingredient for this package! The folder that includes this file, should be specify with argument --folder [directory] in runtime!

  1. Platform or supercomputer

Supercomputer should be identified with argument --platform [supercompure name] in runtime too!

Important

There is no other necessary inputs, but optional one!

We recomend to consider the Optional input if Identikeep plugin is available on the desired supercomputer.

How to USE

It is possible to use terminal commands (The console point is provided). Since it supports just Quantum ESPRESSO output (till this moment), the command is:

NNimake_QE  --folder [DIRECTORY] --outname [OUTPUT] --platform [SUPERCOMPUTER] --algname [ALGORITHM]

All of the arguments have default values. Default for directory is ./ , Default for output name is data.json, Default for platform is Leonardo-booster, and Default for algorithm name is davidson. For example, if your desired directory which contains output file of QE is ../work/ directory, the command line would be:

NNimake_QE --folder ../work/ --platform "Leonardo-booster"

You can import the package in your python code directly:

from NNimaker import create_json

create_json("../work", outname="data.json", platform="Leonardo-booster", algoname='davidson')

Then you can find the created JSON file (data.json) in the directory that you specified with flag outname.

As it is mentioned on the main page (index), you can help us to grow our ANN model with your data, if you willing to do so please share your results with us through this google form

Good luck!