CSX combined optimization

This repository uses the simsopt framework and the combined approach technique described by R. Jorge et.al. to optimize tentative designs of the Columbia Stellarator eXperiment (CSX), which will eventually replace the existing Columbia Nonneutral Torus (CNT)

Dependencies

• simsopt, windowpane branch. commit ec271a44ca557efa9428dad39e994f883338e8b6
• pystellplot. commit a37cbcd70cd37605f7c0c4d553337a7c09f40f72
• everything in requirements.txt

Installation

First, install simsopt and VMEC2000. You might also need to create a runs directory manually, by running

mkdir runs

Running the code

The main script is combined_csx_optimization.py. It can be run locally by creating an input file in the ./inputs directory - just copy-paste the inputs/standard_input.py file into a new file, and set each input to the value you desire. Then, run the code locally (not recommended) with:

python combined_csx_optimization_with_windowpane_coils.py --input /path/to/input

To run the code on the Ginsburg cluster, simply use the sbatch script run_combined_approach.sb and do

sbatch run_combined_approach.sb

Analysing the output

Outputs can be analyzed interactively using the plot_combined_approach_results.ipynb jupyter notebook, or all relevant figures can be generated by running

python run_analysis.py /path/to/output/directory