1. Fabian Köhlinger
  2. qe_public



This repository contains code for a quadratic estimator algorithm in order to estimate the weak gravitational lensing shear power spectrum in multipole band powers and in tomographic bins (i.e. redshift slices along the line-of-sight). The algorithm was proposed in Hu & White 2001 (ApJ, 554, 67), further described in Lin et al. 2012 (ApJ, 761, 15) and it was expanded in Koehlinger et al. 2016 (MNRAS, 456, 1508) to allow for a tomographic analysis, including its application to shear catalogues from CFHTLenS. Most recently it was used in Koehlinger et al. 2017 (MNRAS; arXiv:1706.02892) on shear catalogues from KiDS-450.

The last two papers also include a self-consistent description of the algorithm. Please cite them if results based on this code are published!

The repository contains the following folders: quadratic_estimator, containing the main modules for the algorithm plotting, containing optional (and yet not very user-friendly) plotting routines * input, containing an example parameter-file for the main algorithm

The code is written in Python (2.7.X) and requires the following packages in addition to other standard packages:

  • numpy
  • scipy
  • multiprocessing
  • astropy
  • h5py

The main module of the code is quadratic_estimator.py (contained in the same folder) and it can be used with the command:

python /your/path/to/quadratic_estimator.py /your/path/to/input/your_algorithm_parameters.ini > terminal_output.out

All settings for the algorithm (e.g. input catalogues, output folders, etc.) have to be adjusted in a file your_algorithm_parameters.ini. In the folder input we provide the example file default.ini and we hope that all options are self-consistently and sufficiently explained in there.

The major output of the code will be stored within the specified output folder(s):

  • band_powers_<EE, BB, EB>_<z_i>x<z_j>.dat -- a simple text-file containing the measured E-mode, B-mode, and/or EB-mode band powers for each (unique) redshift bin correlation <z_i> x <z_j> from lowest multipole band to highest multipole band (as given in the file below)
  • multipoles_<EE, BB, EB>.dat -- a simple text-file (with header!) containing the naive multipole bin centers and the multipole bin ranges for each band type as specified in 'your_algorithm_parameters.ini'-file (assumed to be the same for each redshift bin correlation!)
  • band_window_matrix_nell<number>.dat -- if requested, the matrix containing the band window functions of each specified band power of type <EE, BB, EB> for each unique redshift bin correlation
  • last_Fisher_matrix.dat -- if nothing better is available, the INVERSE of this matrix can serve as a very idealised approximation to the covariance of the band powers and it comes 'for free'
  • *.log-file -- copies all set parameters from the *.ini-file and lists other useful quantities related to the performance of the algorithm
  • all other files are saved out of legacy reasons or for further diagnostics only and their purpose should become more obvious when digging into the code...

The folder plotting contains an experimental script for plotting the results (e.g. for ALL subfields of a survey) in publication quality. Major settings can be adjusted and must be supplied with a parameter file, an example of all options are listed in default.plt. Unfortunately, the script is in its current form not very user-friendly and still very tailored to the requirements of our previous analyses, so please consider it rather as a blueprint for your own (better) plotting scripts.

If you want to contribute to this project (e.g. a more efficient implementation using more advanced linear algebra tricks or the bruteforce way of porting it to C and/or GPUs...), please use a dedicated branch per feature/bugfix.

For questions/comments/issues/requests please use the issue-tracking system!

Last but not least you might want to play around with some real data, hence here are links to a personal choice of public shear catalogues (follow their acknowledgement policy if you use them!):