This is a repo for training a Graph Convolutional Generative Adversarial Network to learn and predict galaxy alignments in a hydrosimulation such as IllustrisTNG.
The general idea is that we have list of features (orange box) that are relevant for capturing dependence inside a halo (dashed red box) and the tidal fields that are relevant for capturing the dependence outside of halos (dashed purple). Then, these inputs are fed into the GANGenerator(crimson box) where it tries to learn the desired output labels (yellow box). At the end the output from the GAN-Generator, together with the input, are fed into the GAN-Discriminator (blue box) to determine the performance of the GAN-Generator.
To run the example notebook, the following Python packages are required other than standard ones like pandas, numpy, scipy, matplotlib:
For questions/issues regarding this repo please use the Issues feature by clicking on "New Issues"