Installation

MFTT supports Python 3.10 or newer. The standard installation includes the library and everything needed to run the tutorials and test suite.

Create an isolated environment

We recommend setting up a virtual environment specifically for MFTT.

On Linux or macOS:

python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip

On Windows PowerShell:

py -m venv .venv
.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip

Run the installation command below after activating the environment. To leave it later, run deactivate.

Install MFTT

Clone the repository and enter its root directory:

git clone https://github.com/sandialabs/multifidelity-triangular-transport.git
cd multifidelity-triangular-transport

Then run the standard installation:

python -m pip install .

This installs MFTT and everything needed to run the tutorials and test suite; there are no separate tutorial or test profiles to install.

Dependency

Version

NumPy

>=1.21

SciPy

>=1.8

Jupyter

>=1.0

Matplotlib

>=3.6

nbconvert

>=7.0

pytest

>=7.0

Verify the installation

From the root of a repository clone, run:

python -m pytest

A successful run confirms that MFTT and its dependencies work in the active environment.

Run the tutorials

The notebooks are in docs/tutorials/. Launch Jupyter from the repository root:

jupyter lab docs/tutorials

The hosted documentation renders the notebooks and their saved outputs for reading. Running them locally uses the same standard installation above.