This repository contains the scripts used to evaluate the submissions for the Hackathon accompanying the 2025 Spring School on Physics Informed Machine Learning for Medical Sciences.
To submit your solution for the hackathon task:
-
send the link to your GitHub repository that should be used for evaluation to the email address magnet4cardiac7t@uni-wuerzburg.de.
-
Include the name of your group in your submission and the names of all participants.
-
Make sure that your github repository is set to public so it can be cloned by the evaluation script (see below).
-
Both your submission email as well as your last commit should be made on Thursday, April 10th 6:30pm at the latest
To run the evaluation clone this repository, make the evaluation.sh executable by running
chmod +x evaluation.sh
After which you can run the evaluation by calling:
./evaluation.sh
URL_OF_YOUR_GITHUB_REPONAME_OF_YOUR_GROUP
e.g. ./evaluation.sh https://github.com/MAGNET4Cardiac7T/hackathon the_magnet_crushers
Remember to set your repository to public before the submission.
The script will clone your repository, create a new virtual environment with the packages specified in your requirements.txt file in your repository and then run your optimization algorithm for each of the cost functions specified in cost_list.txt and all simulations listed in data_list.txt (which at evaluation time will contain 5 unseen simulations).
The script calls the run method from the main.py script in your repository and expects a return value of type CoilConfig.
The solution found by your algorithm(s) will then be evaluated using the reference code in this repository and compared against the default configuration (all phases 0 and all amplitudes 1).
For each combination of cost function and simulation, the submissions will be ranked based on the cost achieved using their respective solutions. The overall rank of the submissions will be calculated based on the average rank over all dataset/cost function combinations.
This script only runs on Unix-based operating systems (probably). To make sure that your algorithm evaluates correctly you should test your github repository using the provided evaluation scripts