Software

CosmoFuse

A highly parallelized, GPU-accelerated Python framework for measuring the integrated 3-point correlation function directly from simulated weak lensing maps on HEALPix grids.

  • CUDA-based backend for efficient GPU computation
  • Operates directly on HEALPix shear and density maps
  • Used in the SBi3PCF pipeline to process ~105 mock observations

i3PCF

Analytical predictions for the integrated 3-point correlation function of projected cosmic density fields. I contributed implementations of the area prefactor, pixel window function, some usability improvements, as well as extensive testing and validation against simulations.

affinax

An affine-invariant ensemble MCMC sampler implemented in JAX, with vectorised walker updates for GPU-accelerated posterior sampling.

emax

A JAX-based implementation of a simple neural emulator architecture for fast predictions of expensive model evaluations.

cosm_parameter_mcmc

MCMC reconstruction of the posterior for the alpha-DGP cosmological model fitted to supernovae observations. Written in C using GSL.

SBI Tutorial

A tutorial for machine learning methods and simulation-based inference.