Installation

These instructions install PySTRA 2.0.0, the version these pages describe. Python 3.12 or later is required for the package; Python 3.13 is used for the documentation tools.

Create an environment

With conda, create and activate an environment:

conda create -n pystra2.0 python=3.13
conda activate pystra2.0

Alternatively, create a Python virtual environment with python -m venv .venv and activate it using source .venv/bin/activate on Linux/macOS or .venv\Scripts\Activate.ps1 in Windows PowerShell.

Install PySTRA

Install the release from PyPI into the active environment:

python -m pip install pystra
python -c "import pystra; print(pystra.__version__)"

The version should be 2.0.0. The core installation includes NumPy, SciPy, Matplotlib and pandas. You can now run A first reliability analysis.

Optional active-learning dependencies

The Kriging and PC-Kriging examples additionally require scikit-learn, which the al extra installs:

python -m pip install "pystra[al]"

PCE and the classical reliability algorithms use the core installation. Each example states its dependency requirements. Notebook downloads also need Jupyter; install jupyterlab to run them interactively.

For building all documentation and contributing, follow Contributing to PySTRA. It covers the documentation extras, Pandoc and validation commands.

Install from source

To use the latest development version, clone the repository and install it in editable mode, with the extras you need:

git clone --branch v2.0 https://github.com/pystra/pystra.git
cd pystra
python -m pip install -e ".[al]"

Using PySTRA 1.x

To keep the 1.x API, install a 1.x release and follow the 1.x documentation:

python -m pip install "pystra<2"

The two versions use different APIs. Migrating from 1.x to 2.0 explains how to update existing code.

Troubleshooting installation

If a notebook reports an unexpected version or missing package, check the interpreter used by its kernel:

import sys
import pystra
print(sys.executable)
print(pystra.__version__)

Select the kernel for the environment where PySTRA is installed. For further help, report the interpreter, package version and error message in the issue tracker.