Measurement configuration

yet_another_wizz uses configuration objects to organise its free parameters. These are implemented in the config module, the most important one, Configuration, is directly availale after importing the main module.

from yaw import Configuration

The recommended way to create a new configuration is through its constructor method, which is a keyword-only function. A minimum example is:

>>> config = Configuration.create(rmin=100, rmax=1000, zmin=0.07, zmax=1.42)

The configuration object is hierarchical and holds the parameters in its attributes. The structure matches the layout of the YAML configuration file used for batch processing:

>>> config.scales
ScalesConfig(rmin=100.0, rmax=1000.0, rweight=None, rbin_num=50)
>>> config.binning
BinningConfig(zbin_num=30, z='0.070...1.420', method='linear')
>>> config.backend
BackendConfig(thread_num=8, crosspatch=True, rbin_slop=0.01)
>>> config.cosmology
FlatLambdaCDM(name="Planck15", H0=67.74 km / (Mpc s), Om0=0.3075, Tcmb0=2.7255 K, Neff=3.046, m_nu=[0.   0.   0.06] eV, Ob0=0.0486)

The sections scales, binning, backend, and cosmology are configuration objects themselves, that are constructed automatically (see below). The attributes of these sections can be accessed through the section itself, e.g.:

>>> config.scales.rmin
100.0
>>> config.binning.zmax
1.42

Note

To protect configuration objects from accidental modification, all attributes are read-only. To modify a configuration, use the modify() method, which supports the same arguments as create(). The method creates a new instance and replaces all values by the given keyword arguments, e.g.:

>>> config.scales.rmin, config.scales.rmin
(100.0, 1000.0)

>>> new = config.modify(rmin=200, rmax=2000)
>>> new.scales.rmin, new.scales.rmax
(200.0, 2000.0)

Configuration.scales

Scales for correlation measurements are implemented in the ScalesConfig class and provided as projected physical distance in kpc. The lower and upper limits, rmin and rmax, can either be scalar or arrays of the same length to specify more than one scale.

Optional parameter are rweight and rbin_num, which are used to apply a radial weighting to the single-bin correlation measurement. If rweight is provided, the correlation function is measured in rbin_num logarithmically spaced radial bins. The pair counts are multiplied by the radius corresponding to the bin center and the summed together to approximate a radially weighted correlation measurement.

Configuration.binning

The redshift binning is either constructed automatically using the zmin and zmax (and optional zbin_num and method) parameters, or assigned manually using the zbins parameter:

>>> Configuration.create(
...     rmin=100, rmax=1000, zbins=[0.1, 0.2, 0.3, 0.4])

>>> Configuration.create(
...     rmin=100, rmax=1000, zmin=0.07, zmax=1.42,
...     zbin_num=30, method="linear")

The latter example above corresponds to the default binning, which are 30 bins, linearly spaced in redshift. Other spacings can be selected using the method parameter, see also binning.

Note

Either zmin and zmax or zbins are required to construct a valid redshift binning, otherwise a ConfigError is raised.

Configuration.backend

This section maps to the BackendConfig class, which are parameters for the backend used to compute correlations (see backend and the section on data catalogs). The most important parameter here is the thread_num parameter, which specifies the number of parallel threads to use. The crosspatch parameter specifies, whether the backend counts pairs beyond patch boundaries (crosspatch=False not supported by all backends).

Configuration.cosmology

The cosmological model that is used for distance calculation has usually a minor effect on clustring redshifts. In the configuration it is specifed through the cosmology parameter, the current default is cosmology=Planck15.

If you need use a model, a number of named models from the astropy.cosmology module are available. If you need a custom cosmological model, make sure to implemented it as subclass of yaw.cosmology.CustomCosmology and overwrite the methods that yet_another_wizz expects, e.g.:

class MyCosmology(CustomCosmology):

    def comoving_distance(self, z: ArrayLike) -> ArrayLike:
        return my_comoving_distance(z)

    def comoving_transverse_distance(self, z: ArrayLike) -> ArrayLike:
        return my_comoving_transverse_distance(z)