foregrounds-diffusion
A denoising diffusion probabilistic model (DDPM) pipeline for generating correlated CIB and tSZ extragalactic CMB foreground maps.
Built on denoising-diffusion-pytorch and developed as MPhil Project 7 (Data Intensive Science, Cambridge).
Getting started
API reference
Tutorials
- 01 — Halo Catalogue
- 02 — Masking
- 03 — Patch Extraction and Normalisation
- 04 — Model Architecture and Training
- 05 — Sampling and Post-Processing
- 06 — Power Spectra Comparison
- 07 — Higher-Order Statistics (Bispectrum and Trispectrum)
- 08 — Pixel Histograms and Minkowski Functionals
- 09 — tSZ Cluster Stacking
- 10 — Peak and Minima Counts
- 11 — Scattering Transforms
- 12 — Minkowski Tensors
- 13 — Profiling and Benchmarking Baseline
- Paper figures