An open API service indexing awesome lists of open source software.

https://github.com/simonsobs/hmvec

Vectorized halo model code written entirely in pure Python / numpy
https://github.com/simonsobs/hmvec

Last synced: about 1 year ago
JSON representation

Vectorized halo model code written entirely in pure Python / numpy

Awesome Lists containing this project

README

          

=======
hmvec
=======

`hmvec` is a fast pure Python/numpy vectorized general halo model and HOD code. Currently, it includes support for (a) 3d power spectra involving NFW, Battaglia electron density profiles and galaxy HODs and (b) 2d power spectra including tSZ, cosmic shear, galaxy-galaxy lensing and CMB lensing.

It calculates a vectorized FFT for a given profile over all points in mass and
redshift, but it currently does have one double loop over mass and redshift
to interpolate the profile Fourier transforms to the target wavenumbers. Every
other part of the code is vectorized. Heavy memory usage can be an issue -- watch out for that!

* Free software: BSD license
* Documentation: in the works

Dependencies
------------

* Python>=2.7 or Python>=3.4
* numpy, scipy, matplotlib
* camb (Python package, recommend using `dev` branch)

Credits
-------

The theory used here follows the approach outlined in the appendix of
arxiv:1810.13423. This code has greatly benefited from comparisons with the
implementation written by Moritz
Munchmeyer and Matt Johnson for that paper. Some of the HOD functions are copied (and
modified) from there.

Installation
------------

Clone this repository and install with symbolic links as follows
so that changes you make to the code are immediately reflected.

.. code-block:: console

pip install -e . --user

Usage
-----

One can quickly get the matter power spectrum for desired wavenumbers and
redshifts after specifying the mass grid to integrate over. Note that
the analytic NFW profile is initialized by default.

.. code-block:: python

import hmvec as hm
zs = np.linspace(0.,3.,20)
ms = np.geomspace(2e10,1e17,200)
ks = np.geomspace(1e-4,100,1001)
hcos = hm.HaloModel(zs,ks,ms=ms)
pmm_1h = hcos.get_power_1halo(name="nfw")
pmm_2h = hcos.get_power_2halo(name="nfw")

You can add functions that implement a profile of your choice. An electron
profile from Battaglia 2016 has also been implemented. It needs to
be FFTd numerically to get the electron power spectrum, which is done as follows:

.. code-block:: python

hcos.add_battaglia_profile("electron",family="AGN",xmax=20,nxs=5000)
pee_1h = hcos.get_power_1halo(name="electron")
pee_2h = hcos.get_power_2halo(name="electron")

Cross-spectra can also be calculated:

.. code-block:: python

pme_1h = hcos.get_power_1halo("nfw","electron")
pme_2h = hcos.get_power_2halo("nfw","electron")

An HOD can be added as follows:

.. code-block:: python

hcos.add_hod(name="g",mthresh=10**10.5+zs*0.)

and galaxy spectra and cross-spectra with matter and electrons can be
calculated just as above by specifying the chosen name for the HOD.
If the galaxy number density `ngal` is provided instead of `mthresh`,
the latter will be found iteratively.

Cosmic Shear / CMB Lensing autospectrum
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

`HaloModel` inherits from `cosmology.Cosmology` which contains some
convenient functions involving Limber integrals. To get a cosmic shear
power spectrum for example, you first build the total matter power
spectrum and pass it to the relevant member function of `cosmology.Cosmology`,

.. code-block:: python

pmm_1h = hcos.get_power_1halo(name="nfw")
pmm_2h = hcos.get_power_2halo(name="nfw")
Pmm = pmm_1h + pmm_2h

ells = np.linspace(100,600,10)
Cls = hcos.C_kk(ells,zs,ks,Pmm,lzs1=2.5,lzs2=2.5)

Galaxy-galaxy lensing / Galaxy-CMB lensing
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Similarly, one can obtain cross-spectra for galaxy-galaxy lensing
and galaxy-CMB lensing,

.. code-block:: python

hcos.add_hod(name="g",mthresh=10**10.5+zs*0.)
pgm_1h = hcos.get_power_1halo("nfw","electron")
pgm_2h = hcos.get_power_2halo("nfw","electron")
Pgm = pgm_1h + pgm_2h

ells = np.linspace(100,600,10)
Cls = hcos.C_kg(ells,zs,ks,Pgm,gzs=0.8,lzs=2.5)