Installation

Installation of Linux System Configuration

Debian-based distributions (also Windows 11 with WSL)

You can also use the same commands on a pure Debian-based Linux system or Windows systems with WSL. If you do not know how to install Linux on Windows 11 with WSL, you can view this video. On the WSL system, you can use either Debian or Ubuntu. We recommend Ubuntu due to the support provided by Microsoft. First, install the required system files:

$ sudo apt update && sudo apt upgrade -y
$ sudo apt install -y python3-venv python3-pip unzip python-is-python3 \
                 python3-dev libopenblas-dev libxc-dev libscalapack-mpi-dev \
                 libfftw3-dev libkim-api-dev openkim-models libkim-api2 pkg-config \
                 task-spooler

Fedora-based distributions

First, install the required system files:

$ sudo dnf update
$ sudo dnf install python3-devel openblas-devel libxc-devel scalapack-openmpi-devel fftw-devel pkgconf

You also must install kim-api, kim-api-devel, and openkim-models. At the time of writing these instructions, packages for Fedora 43 cannot be installed remotely. Therefore, we must download them, then install them with dnf locally. The order is important:

$ wget https://download.copr.fedorainfracloud.org/results/lecris/cmake-ninja/fedora-rawhide-x86_64/08840866-kim-api/kim-api-2.2.1-11.fc43.x86_64.rpm
$ wget https://download.copr.fedorainfracloud.org/results/lecris/cmake-ninja/fedora-rawhide-x86_64/08840866-kim-api/kim-api-devel-2.2.1-11.fc43.x86_64.rpm
$ wget https://download.copr.fedorainfracloud.org/results/lecris/cmake-ninja/fedora-rawhide-x86_64/08841484-openkim-models/openkim-models-2021.01.28-12.fc43.src.rpm
$ sudo dnf install kim-api-2.2.1-11.fc43.x86_64.rpm
$ sudo dnf install kim-api-devel-2.2.1-11.fc43.x86_64.rpm
$ sudo dnf install openkim-models-2021.01.28-12.fc43.src.rpm

Python Virtual Environment Installation

Then, if you do not have a Python environment, create one and activate it:

$ python -m venv ~/.venv_nw
$ source ~/.venv_nw/bin/activate

Python Libraries Installation

For dftsolve, mdsolve, and mlsolve commands, we need to install many Python libraries.

ASE and GPAW for dftsolve

After preparing your Linux system and environment, you must have the ase and gpaw codes on your computer. You can find more information about installing ASE and GPAW from their related sites.

(.venv_nw) $ pip3 install --upgrade ase
(.venv_nw) $ pip3 install setuptools_scm
(.venv_nw) $ pip3 install spglib docutils elastic requests phonopy pyrapl pymongo pandas

Creating a siteconfig.py file is important. You can use any text editor. Here, we are creating a file with the cat command, writing necessary information inside it, then closing it with the Ctrl-D command (^D).

(.venv_nw) $ mkdir -p ~/.gpaw
(.venv_nw) $ cat > ~/.gpaw/siteconfig.py
fftw = True
scalapack = True
libraries = ['xc', 'blas', 'fftw3', 'scalapack-openmpi']
^D

If you have problems with libraries fftw, scalapack, you can remove them from siteconfig.py file. They are simply optional. Then continue to install gpaw

(.venv_nw) $ pip3 install --upgrade gpaw

Use gpaw info to see installation information. However, PAW datasets are not installed yet. To install them, first create a directory under ~/.gpaw and then install PAW datasets.

(.venv_nw) $ mkdir ~/.gpaw/gpaw-setups
(.venv_nw) $ gpaw install-data --gpaw ~/.gpaw/gpaw-setups/

Installation of ASAP and KIM for mdsolve

For quick optimization, we need simple interatomic modeling. For this, we need ASAP3 for ASE, then we must use KIM with OpenKIM models and kimpy libraries.

(.venv_nw) $ pip3 install --upgrade asap3
(.venv_nw) $ pip3 install --upgrade kimpy

Installation of Machine Learning libraries for mlsolve

For machine learning features, we must install the libraries related with PyTorch, MACE (Multi-Atomic Cluster Expansion), CHGNet (Charge-Informed Graph Neural Network), SevenNet (Scalable Equivariance Enabled Neural Network):

(.venv_nw) $ pip3 install torch mace-torch chgnet sevenn

Installation of Nanoworks

Then, lastly, install nanoworks:

(.venv_nw) $ pip3 install nanoworks