software

My goal is to deliver the next generation of precision cosmology results with the incoming Stage IV surveys. Over the last decade I have been involved in the development, verification and validation of the ESA Euclid mission, and along the way I have built new software to go from a cosmological model to predictions for the observables, from predictions to likelihoods, and from Euclid data products to parameter constraints we can trust.

I have always kept in mind that the best cosmological constraints we will ever achieve will come from combining several surveys and data sets, not from any single one. That is why I make sure the software I develop is transferable and can be extended and applied to other experiments. Everything I do is open source, and most of it is built together with colleagues in the community (I am proudly a team player). If you use it, break it or want to improve it, I'd love to hear from you.

gcanasherrera cloe-org euclidlib

cloe-org: the Euclid likelihood ecosystem

CLOE (Cosmology Likelihood for Observables in Euclid) is how Euclid goes from data to cosmological parameters. At the end of 2023 it was scientifically trusted but difficult to work with: years of technical debt meant every change felt risky, and turning scientific ideas into working projects had become exhausting. To do excellent science with Euclid's data, we needed to rethink how we built our tools. In a field already full of inference codes, we needed to do things better. That is how cloe-org was born.

Today cloe-org is building a fully differentiable inference framework for large-scale structure cosmology and beyond: open source, modular, written in modern Python and adopted by the Euclid Consortium. It has more than 80 contributors, around 2,000 downloads every week, and papers on arXiv. Beyond cloelib and cloelike, the ecosystem includes tools for running chains at scale, blinding, figures of merit and easy installation.

cloe-org is my little joy. It grew out of countless late nights, long discussions and difficult conversations, and above all out of the support of Euclid's younger community, who never let me down, and of the senior mentors who pushed this vision with us. What started as a difficult technical problem has become a community I am genuinely proud of, and this is just the beginning.

pip install cloelib

cloelike creator & lead developer

The likelihoods, built on cloelib. It covers all Euclid primary probes: photometric 3×2pt in harmonic and real space, BNT C(ℓ), COSEBIs, and spectroscopic full-shape and BAO, separately or combined. Installing it gives you the whole stack.

pip install cloelike

code PyPI paper

CLOE core developer

The original Euclid Consortium likelihood code, which we used for the official Euclid forecasts and validation. cloelib and cloelike grew out of it.

code

Euclid data and statistics

euclidlib core developer

A community Python library for reading Euclid's cosmology-ready data products, such as redshift distributions, two-point statistics and covariances, so you can spend your time on the science instead of the file formats.

pip install euclidlib

code PyPI

darkflipper creator

Figures of Merit, Figures of Bias, p-values and other diagnostics I kept re-implementing for systematics studies, now in one place with clear definitions.

pip install darkflipper

code PyPI

Teaching and reproducibility