This repository contains code and resources aimed at enhancing the particle resolution of dark matter-only simulations using a Wasserstein Generative Adversarial Network (WGAN). This work has applications in astrophysics and cosmology, particularly for improving simulation fidelity in large-scale structure studies.
The DMSR-WGAN uses adversarial training to super-resolve dark matter simulations. It generates high-resolution particle distributions from low-resolution inputs. This approach enables:
- Enhanced particle resolution.
- Improved substructure identification.
- Reduced computational cost compared to high-resolution simulations.
For more details, refer to: On the Use of WGANs for Super Resolution in Dark-Matter Simulations.
For scripts and steps to reproduce results from On the Use of WGANs for Super Resolution in Dark-Matter Simulations,
please switch to the on-wgan-super-resolution branch.