Example: Basic covariance computation

Users should use the tjpcov.covariance_calculator.CovarianceCalculator class with a configuration file or dictionary input.

An example configuration file could be the following. In this case, one will compute the Gaussian covariance with NaMaster and the SSC covariance. The configuration specific for each covariance is defined in their own sections. Extra sections (e.g. GaussianFsky) are ignored.

tjpcov:
  sacc_file: tests/benchmarks/32_DES_tjpcov_bm/cls_cov_as_txpipe.fits

  cosmo: tests/data/cosmo_desy1.yaml

  # allowed covariance type: gauss, ng (to be implemented)
  cov_type: [FourierGaussianNmt, FourierSSCHaloModel]


  IA: 0.5

  outdir: tests/tmp/

  mask_file:
    DESgc__0: ./tests/benchmarks/32_DES_tjpcov_bm/catalogs/mask_DESgc__0.fits.gz
    DESwl__0: ./tests/benchmarks/32_DES_tjpcov_bm/catalogs/DESwlMETACAL_mask_zbin0_ns32.fits.gz
    DESwl__1: ./tests/benchmarks/32_DES_tjpcov_bm/catalogs/DESwlMETACAL_mask_zbin1_ns32.fits.gz

  mask_names:
    DESgc__0: mask_DESgc0
    DESwl__0: mask_DESwl0
    DESwl__1: mask_DESwl1

  # Survey params:
  # 5 lens bins
  Ngal_DESgc__0: 26

  Ngal_DESwl__0: 26
  Ngal_DESwl__1: 26
  # # constant bin sigma_e
  sigma_e_DESwl__0: 0.26
  sigma_e_DESwl__1: 0.26

  # linear bias for lenses constant for redshift bin (example notebook)
  bias_DESgc__0: 1.48

GaussianFsky:
  fsky: 0.5

NaMaster:
  module: pymaster
  f:
    n_iter: 0
    lmax_sht: -1
  w:
    lmax_mask: -1
    l_toeplitz: -1
    l_exact: -1
    dl_band: -1
    n_iter: 3
    recompute: True
  cw:
    l_toeplitz: -1
    l_exact: -1
    dl_band: -1
    n_iter: 3
    recompute: True

cache:
  'workspaces':
    '00':
      !!python/tuple [mask_DESgc0, mask_DESgc0]: ./tests/benchmarks/32_DES_tjpcov_bm/DESgc_DESgc/w__mask_DESgc__mask_DESgc.fits
    '02':
      !!python/tuple [mask_DESgc0, mask_DESwl0]: ./tests/benchmarks/32_DES_tjpcov_bm/DESgc_DESwl/w__mask_DESgc__mask_DESwl0.fits
      !!python/tuple [mask_DESgc0, mask_DESwl1]: ./tests/benchmarks/32_DES_tjpcov_bm/DESgc_DESwl/w__mask_DESgc__mask_DESwl1.fits
    '22':
      !!python/tuple [mask_DESwl0, mask_DESwl0]: ./tests/benchmarks/32_DES_tjpcov_bm/DESwl_DESwl/w__mask_DESwl0__mask_DESwl0.fits
      !!python/tuple [mask_DESwl0, mask_DESwl1]: ./tests/benchmarks/32_DES_tjpcov_bm/DESwl_DESwl/w__mask_DESwl0__mask_DESwl1.fits
      !!python/tuple [mask_DESwl1, mask_DESwl1]: ./tests/benchmarks/32_DES_tjpcov_bm/DESwl_DESwl/w__mask_DESwl1__mask_DESwl1.fits

In order to generate the covariance there are two options:

1. Using the run_tjpcov.py script.

$ python run_tjpcov.py config.yml

This will create a summary_statistics.sacc file containing the final covariance in the output directory defined in the configuration file (i.e. tests/tmp/). You can use the -o or --output arguments to use a different file name (the path will be the same).

2. Interactively

If you just want to get the final covariance:

from tjpcov.covariance_calculator import CovarianceCalculator
cc = CovarianceCalculator(config_yml)
cov = cc.get_covariance()

If you just want to get the sacc file with the covariance (this will also save a copy of the sacc file):

from tjpcov.covariance_calculator import CovarianceCalculator
cc = CovarianceCalculator(config_yml)
s = cc.create_sacc_cov()