Skip to content

Latest commit

 

History

History
124 lines (80 loc) · 3.69 KB

README.rst

File metadata and controls

124 lines (80 loc) · 3.69 KB
Raw

tILe-C

https://travis-ci.org/simonsobs/tilec.svg?branch=master https://coveralls.io/repos/github/ACTCollaboration/tilec/badge.svg?branch=master

tILe-C is ILC in tiles. It is both a library for CMB foregrounds and harmonic ILC as well as a set of pipeline scripts designed primarily for component separation of high-resolution ground-based CMB maps that might have inhomogenous, anisotropic and inhomogenously anisotropic noise.

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

Dependencies

  • Python>=2.7 or Python>=3.4
  • numpy, scipy, matplotlib
  • pixell

Development and Contributing

This code is currently being developed by Mathew Madhavacheril, Colin Hill and Sigurd Naess. Contributions are welcome and encouraged through pull requests.

Installation

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

pip install -e . --user

Pointing to data

You can start running the pipeline scripts as soon as you set up a configuration file that tells these scripts where relevant data products are stored. To do this, copy the template in input/paths_template.yml to paths.yml. The former is version controlled and the latter is not. Edit the latter to conform to where you have stored the relevant data on the system you have just cloned this repository to.

The data model itself is implemented in tilec/datamodel.py. To interface with the data products of a new experiment (or version), you are encouraged to modify that module and submit your changes as a pull request to be incorporated into the master branch here. Modifications are preferred to be in the form of subclasses of tilec.datamodel.DataModel. See the ACT and Planck examples there.

You can then add descriptions of ``arrays'' to `data.yml' that point to implemented data models. The overall analysis is tied to a `mask' whose WCS geometry specifies what pixel regions of the maps in each array are loaded and coadded together. The mask is also assumed to be apodized and is applied right before any FFT is performed.

Running ILC

1. Covariance

The first step is to generate an empirical covariance matrix. The pipeline script for this is ``bin/cov.py''. The call options are:

python bin/cov.py <version> <region>
<comma,separated,list,of,array,short,names>
[--signal-bin-width I] [--signal-interp-order I] [--dfact I]
[--rfit-bin-width I] [--rfit-wnoise-width I] [--rfit-lmin I]

An example call is:

python bin/cov.py v0.1 deep56 a1,a2,p1,p2

This will make a directory named v0.1 in the default save location, a sub-directory named deep56, use masks corresponding to deep56, generate the non-redundant parts of a TILe-C hybrid covariance matrix involving arrays with short names a1,a2,p1,p2 and save it to that sub-directory along with a copy of the configuration options.

2. Coadd

python bin/ilc.py <version> <region>
<comma,separated,list,of,array,short,names>
<comma,separated,list,of,solutions>

Each solution is of the form x-y-... where x is solved for and the optionally provided y-,... are deprojected. The x,y,z,... can be belong to any of cmb,tsz,cib.

e.g.

python bin/ilc.py v0.1 deep56 a1,p1
cmb,cmb-tsz,cmb-cib,cmb-tsz-cib,tsz,tsz-cmb,tsz-cib

In the above example, only 2 of the arrays in the original covariance matrix are used.