Skip to content

Latest commit

 

History

History
20 lines (14 loc) · 2.61 KB

README.md

File metadata and controls

20 lines (14 loc) · 2.61 KB

Full-Shape Likelihoods

This repository contains the Montepython likelihoods needed to perform full-shape analysis of current and future survey data. The likelihood can be used to analyze any of the following datasets:

The likelihoods make extensive use of the CLASS-PT code (Chudaykin+20), which is an extension of the Boltzmann code CLASS. Although the one-loop bispectrum templates have been computed (see OneLoopBispectrum, Philcox+22), these are not currently included in the likelihoods since they add little constraining power.

We can analyze both (conventional) Gaussian initial conditions, as well as modifications for primordial non-Gaussianity. In particular, we include a full implementation of the equilateral and orthogonal templates, which can be turned on in the .param file. Optionally, cosmological collider non-Gaussianity can be included (following Cabass+24). This requires a set of interpolation tables for the one-loop power spectrum (P_12), which can be provided on request.

In the data/ directory, we include measurements of each of the above data-sets for the BOSS survey. Note that power spectra and bispectra do not include the survey window function; these are computed with the estimators of Philcox 20, and Philcox 21, as in the PolyBin3D repository. We provide a sample .param file in the input/ directory; this will reproduce a full BOSS analysis similar to Philcox & Ivanov 21.

Authors

  • Oliver Philcox (Columbia / Simons Foundation)
  • Mikhail Ivanov (MIT)
  • Giovanni Cabass (Ruder Boskovic Institute, Zagreb)

Newly added features: cosmological collider non-Gaussianity and general streamlining*