icarus-fusion 0.1.1

A machine learning tool for fusion simulation validation.

Icarus

Machine learning for fusion simulation validation. Named after the Icarus of greek mythology - because to reach for the stars, you risk a little sunburn.

The purpose of this package is to provide a set of machine learning tools that engineers can use to assess the agreement between an experiment and simulation; that is, to validate the simulation with experimental data. When the experiment does not agree with the simulation the tools should provide the engineer with a probable reason for the mismatch to allow further investigation and diagnosis.

Installation

Standard Installation (PyPI)

You can install icarus from PyPi as follows:

pip install icarus

Developer Installation

Clone icarus to your local system and cd to the root directory of icarus. Ensure that your virtual environment is activated and run from the icarus root directory:

pip install -e .

PyTorch

Icarus requires the latest stable version of PyTorch. The installation process varies depending on your hardware and operating system. Please follow the appropriate instructions below:

CPU Installation:

If you do not have access to NVIDIA GPUs, install the CPU version of PyTorch. Use the following commands based on your operating system:

• Windows/macOS:

pip3 install torch

• Linux:

pip3 install torch --index-url https://download.pytorch.org/whl/cpu

GPU Installation (NVIDIA CUDA):

If you have access to NVIDIA GPUs and want to leverage CUDA for faster computation, use these commands (note: CUDA is not available on MacOS):

• Linux:

pip3 install torch

• Windows:

pip3 install torch --index-url https://download.pytorch.org/whl/cu121

Note: The CUDA version (cu121 in this example) may change. Always check the official PyTorch website for the most up-to-date installation instructions and CUDA version compatibility.

Verifying Installation:

After installation, you can verify that PyTorch is installed correctly by running:

import torch

print(torch.__version__)
print(torch.cuda.is_available())  # returns True if CUDA available and properly installed

Getting Started

The examples folder includes a sequence of examples using icarus : to generate the dataset and train an ml model from the suite available on the generated data.

Contributors

• Arjav Poudel, UK Atomic Energy Authority, (arjavp-ukaea)
• Baris Cavusoglu, UK Atomic Energy Authority, (barisc-ukaea)
• Luke Humphrey, UK Atomic Energy Authority, (lukethehuman)