flipcosmo 1.2.2

Field Level Inference Package

flip: Field Level Inference Package

flip is a Python package that uses the maximum likelihood method to fit the growth rate based on the velocity and density fields. The first part of the software is the computation of a covariance matrix from a model power spectrum and the considered coordinates. This part is generalized to work for any linear power spectrum models, both for velocities, densities, and cross-terms, and it is optimized with Hankel transform for any model. In the second part, the covariance is used to create a likelihood by multiplying it by velocities or densities. Finally, this package includes some integrated fitters such as Minuit and MCMC (with emcee) to fit the growth rate of structures.

Quick install

git clone https://github.com/corentinravoux/flip.git
cd flip
pip install .

For now, the package requires you to install manually cosmoprimo: pip install git+https://github.com/cosmodesi/cosmoprimo

Required packages

Mandatory: numpy, scipy, matplotlib, cosmoprimo, iminuit, emcee, sympy

Optional: classy, pyccl, pypower, GPy, tensorflow

Examples

For an example with velocity fit check out:

For density only:

For a joint fit:

Need help?

Documentation available on ReadTheDoc

How to cite

The full description of the core concepts of this package is given here. This package was started on the previous work of @bastiencarreres, detail in this article. Please cite both paper when using the package.