EFTofLSS galaxy clustering

Full-shape galaxy clustering analysis using EFTofLSS with modern differentiable tooling (PyBird-JAX).

I work on full-shape galaxy clustering using the Effective Field Theory of Large-Scale Structure (EFTofLSS). Together with my collaborators Pierre Zhang and Henry Zheng, we built PyBird-JAX: a JAX-based version of PyBird that enables rapid computation of the 1-loop galaxy power spectrum under EFTofLSS modelling. The code leverages just-in-time compilation, automatic differentiation, and neural network emulation of slow internal loops to achieve significant speedups over traditional implementations.

This framework has enabled new analyses, including a study of volume projection effects in EFTofLSS inference, where we showed how projection effects can introduce systematic biases in parameter estimation and developed methods to correct for them using information geometry.

Volume projection effects in EFTofLSS

Measuring dark energy with DESI: Forecasted constraints on the dark energy equation of state parameters (w₀, wₐ) from DESI Year 1 full-shape galaxy clustering analysis using PyBird-JAX. The analysis accounts for volume projection effects that can bias cosmological parameter inference when fitting EFTofLSS models to survey data. From Reeves et al. (2025c).