nessai.gw.utils

Utilities specific to the gw subpackage.

Module Contents

Classes

DistanceConverter	Base object for converting from a distance parameter to a uniform parameter.
NullDistanceConverter	Converter that applies the identity transformation.
PowerLawConverter	Convert from a distance parameter sampled from a power law to a uniform
ComovingDistanceConverter	Object to convert luminosity distance with a prior that is uniform

Functions

get_distance_converter(prior)

Get a distance converter from a type of prior.

Attributes

logger

nessai.gw.utils.logger

class nessai.gw.utils.DistanceConverter

Bases: abc.ABC

Base object for converting from a distance parameter to a uniform parameter.

See nessai.gw.reparameterisations.DistanceReparameterisation for more details on how the distance converters are used.

has_conversion = False

Indicates if the converter class includes a conversion. This is used when defining the prior in the X-prime space. For example, the NullDistanceConverter doesn’t apply a conversion to a uniform parameter, so the prime prior cannot be defined.

has_jacobian = False

Indicates if the transform applied by the converter has a tractable jacobian.

abstract to_uniform_parameter(self, d)

Converter to parameter that has uniform prior.

Parameters

darray_like

Distance.

Returns

d, log_jarray_like

Distance and the log Jacobian determinant

abstract from_uniform_parameter(self, d)

Convert from a parameter that has a uniform prior.

Parameters

darray_like

Scaled distance.

Returns

d, log_jarray_like

Distance and the log Jacobian determinant

class nessai.gw.utils.NullDistanceConverter(**kwargs)

Bases: DistanceConverter

Converter that applies the identity transformation.

Used for cases where the prior on distance is not specified.

has_jacobian = True

to_uniform_parameter(self, d)

Applies the identity transformation.

Parameters

darray_like

Distance.

Returns

d, log_jarray_like

Distance and the log Jacobian determinant, which will always be zero.

from_uniform_parameter(self, d)

Applies the identity transformation.

Parameters

darray_like

Distance.

Returns

d, log_jarray_like

Distance and the log Jacobian determinant, which will always be zero.

class nessai.gw.utils.PowerLawConverter(power=None, scale=1000.0, **kwargs)

Bases: DistanceConverter

Convert from a distance parameter sampled from a power law to a uniform parameter.

Assumes d is proportional to \(d^{(\text{power} + 1)}\) following the convention in Bilby.

Parameters

powerfloat

Power to use for the power-law.

scalefloat

Factor used to rescale distance prior to converting to the uniform parameter.

has_conversion = True

has_jacobian = True

to_uniform_parameter(self, d)

Convert distance to a parameter with a uniform prior.

Parameters

darray_like

Distance.

Returns

d, log_jarray_like

Distance and the log Jacobian determinant.

from_uniform_parameter(self, d)

Convert to distance from a parameter that has a uniform prior.

Parameters

darray_like

scaled distance.

Returns

d, log_jarray_like

Distance and the log Jacobian determinant.

class nessai.gw.utils.ComovingDistanceConverter(d_min=None, d_max=None, units='Mpc', cosmology='Planck15', scale=1000.0, pad=0.05, n_interp=500)

Bases: DistanceConverter

Object to convert luminosity distance with a prior that is uniform in co-moving volume to a parameter with a uniform prior.

The uniform parameter is a scaled version of the co-moving distance cubed. This transformation does not have a tractable Jacobian determinant and therefore returns zero.

Parameters

d_min, d_maxfloat

Minimum and maximum distances.

unitsstr, optional

Units used for the distance, must be compatible with astropy units.

cosmologystr, optional

Cosmology used for conversion, must be compatible with astropy. Default is Planck15.

scalefloat, optional

Rescaling applied to distance after converting to co-moving distance.

padfloat, optional

Padding used for min and max of interpolation range: min = (1 - pad) * d_min and max = (1 + pad) * d_max

n_interpint, optional

Length of vector used for generating the look up table. For a range of [100, 5000] 200 seems to the minimum for the conversion to be invertible up to 6 decimal places. The recommended setting is at 500.

has_conversion = True

has_jacobian = False

to_uniform_parameter(self, d)

Convert luminosity distance to a parameter with a uniform prior.

Parameters

darray_like

Distance.

Returns

d, log_jarray_like

Distance and the log Jacobian determinant, which will always be zero.

from_uniform_parameter(self, d)

Convert from a uniform parameter to luminosity distance.

Parameters

darray_like

Scaled distance.

Returns

d, log_jarray_like

Distance and the log Jacobian determinant, which will always be zero.

nessai.gw.utils.get_distance_converter(prior)

Get a distance converter from a type of prior.

If the prior is unknown nessai.gw.utils.NullDistanceConverter is returned which has the identity rescaling.

Parameters

priorstr, {‘uniform-comoving-volume’, ‘power-law’}

The prior that is being used for the distance parameter.

Returns

nessai.gw.utils.DistanceConverter

The corresponding distance converter.