cosmicjellyfish#

CosmicJellyfish: Fisher and Monte Carlo cosmological forecasts with Python and JAX CosmicJellyfish: Fisher and Monte Carlo cosmological forecasts with Python and JAX

Fisher-matrix forecasts for galaxy clustering, weak lensing and CMB surveys, with CAMB, CLASS, a built-in symbolic backend or optional JAXcosmo.

pip install cosmicjellyfish

Get started

API reference

Installation

Install from PyPI or from source, choose optional extras, and check the installation.

Installation
A verified first forecast

Run a compact photometric forecast end to end with the built-in symbolic backend.

A verified first forecast
User guide

Core concepts, derivative schemes, likelihood sampling with Nautilus, and backend background.

User guide
Observables and probes

Euclid photometric and spectroscopic clustering, weak lensing, CMB and intensity mapping.

Observables and survey settings
API reference

Modules, classes and functions generated from the package docstrings.

API reference
Contributing and citing

Contribution guidelines, changelog, validation reports and the scripts reference.

Development

Features#

  • Multiple backends. Use CAMB, CLASS, the built-in symbolic backend that needs no external Einstein–Boltzmann solver, or precomputed external tables.

  • Surveys and observables. Photometric (GCph, WL) and spectroscopic (GCsp) Euclid forecasts, CMB (CMB_T, CMB_E, CMB_B) and intensity mapping (IM).

  • Finite-difference derivatives. 3PT, STEM, POLY and 4PT_FWD schemes, selected with a single option.

  • Automatic differentiation. Optional JAXcosmo theory and JAX-native photometric and spectroscopic observables, with matched AD/FD comparisons (JAXcosmo and automatic differentiation).

  • Analysis and plotting. Marginalise, reshuffle and combine Fisher matrices, and draw triangle plots to compare surveys and settings.

  • Likelihood sampling. Run Nautilus posterior sampling with the same observables and compare chains with Fisher forecasts (Likelihood sampling with Nautilus).

  • Validated. CAMB-versus-CLASS Fisher comparisons and likelihood checks are documented and reproducible from the repository scripts.

Cite us#

Please cite the software if you use it. The repository’s CITATION.cff is the maintained source of citation metadata; a minimal BibTeX entry is:

@software{cosmicjellyfish,
  author = {Casas, Santiago and Martinelli, Matteo and Pamuk, Sefa and Sabarish, V. M.},
  title  = {cosmicjellyfish: Fisher-matrix forecasts for cosmological surveys},
  url    = {https://github.com/santiagocasas/cosmicjellyfish}
}