uv (and direnv) on DelftBlue: testing with ASE¶

These instructions install uv and direnv in your home directory.

uv is a very fast and powerful Python installer, package manager and virtual environment manager written in Rust.
direnv is a shell extensions that automatically loads or unloads environment variables when you enter or leave a directory. This allows you to activate or deactivate a Python virtual environment when entering or leaving a directory.

In this tutorial, you’ll use uv to create a virtual environment with the Python version from the 2025 software stack, install the ASE package, and run an ASE example script on DelftBlue using sbatch.

Install uv¶

Install uv in your $HOME/.local/:

curl -LsSf https://astral.sh/uv/install.sh | sh

Ensure that $HOME/.local/bin is included in your PATH (e.g. .bashrc).

Install direnv¶

Install direnv in your $HOME/.local/:

curl -sfL https://direnv.net/install.sh | bash

Add the following line to your shell startup file (e.g. ~/.bashrc):

eval "$(direnv hook bash)"

Reload your shell afterward, or manually execute this line in your current shell.

Load DelftBlue Python module¶

In this example we use the Python version of the current 2025 stack

module load 2025 python

Check wheter uv finds this version and note the correct Python version number for later (3.11.9):

uv python list

Example output (truncated):

cpython-3.15.0a1-linux-x86_64-gnu                 <download available>
cpython-3.11.13-linux-x86_64-gnu                  /usr/bin/python3.11
cpython-3.11.9-linux-x86_64-gnu                   /apps/arch/2025/software/linux-rhel8-cascadelake/gcc-13.3.0/python-3.11.9-y7fulttvp6obrlg7fu55qwdhcvbhlcrd/bin/python3.11
cpython-3.11.9-linux-x86_64-gnu                   /apps/arch/2025/software/linux-rhel8-cascadelake/gcc-13.3.0/python-3.11.9-y7fulttvp6obrlg7fu55qwdhcvbhlcrd/bin/python3 -> python3.11
cpython-3.11.9-linux-x86_64-gnu                   /apps/arch/2025/software/linux-rhel8-cascadelake/gcc-13.3.0/python-3.11.9-y7fulttvp6obrlg7fu55qwdhcvbhlcrd/bin/python -> /apps/arch/2025/software/linux-rhel8-cascadelake/gcc-13.3.0/python-3.11.9-y7fulttvp6obrlg7fu55qwdhcvbhlcrd/bin/python3
graalpy-3.8.5-linux-x86_64-gnu                    <download available>

Initialise ASE project¶

Create a new project directory, initialize it with uv, and set up a virtual environment using the detected Python version:

cd ~
uv init ase_project
cd ase_project
uv venv --python 3.11.9

Next, enable automatic activation of this virtual environment using direnv:

echo "source .venv/bin/activate" > .envrc
direnv allow

From now on, the virtual environment will activate automagically whenever you enter this directory (or its subdirectories).

Add the ASE package¶

Install the ASE package into your virtual environment:

uv add ase

Copy the ASE example from the Linux Command Line Course¶

From the DCSE Linux Command Line Course, locate the directory:

Exercises/ssh/delftblue_examples/03_gen_mod_folders

Copy this entire directory into your ase_project directory, then modify the sub_to_queue.sh file as follows:

#!/bin/bash
#
#SBATCH --job-name="gen_mol"
#SBATCH --time=00:05:00
#SBATCH --partition=compute
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --mem-per-cpu=1GB
#SBATCH --account=education-eemcs-courses-linuxcli

# Run Python script:
module load 2025
module load python

source $HOME/ase_project/.venv/bin/activate

srun python gen_mol_folders.py

Note: the python package modules have been removed as they are now installed as dependencies of ase and provided as such in the virtual environment. \ Also note the source line ‘activating’ this virtual environment.

Submit the job¶

You can now run sbatch and inspect the resulting molecules directory and .out file:

sbatch sub_to_queue.sh