nessai.gw.legacy
Legacy version of GWFlowProposal used in the first paper and the realted utilities and priors
Module Contents
Classes
A proposal specific to gravitational wave CBC |
Functions
|
Determine the values of the prior min and max in the rescaled |
|
Decompose an angle into a real and imaginary part |
|
Reconstruct an angle given the real and imaginary part. Assume the angle |
|
Convert right ascension, declination and (optional) luminosity distance |
|
Reconstruct an angle given the real and imaginary part |
|
Convert azimuth, zenith and (optional) luminosity distance |
|
Reconstruct an angle given the real and imaginary part |
|
Convert a variable defined on [0,1] to an angle on [-pi, pi] and |
|
Convert Cartesian coordinates to a variable defined on [0, 1] and |
|
|
|
|
|
|
|
|
|
Unformalised log probability of uniform prior |
|
Log probability of Cartesian coordinates for a uniform distibution of |
Log probability of Cartesian coordinates for a sine distibution of |
|
|
Log probability of 3d Cartesian coordinates for an isotropic distribution |
|
Log probability of the prior on the components of spin vectors |
|
Log probability of the prior on the components of spin vectors |
Attributes
- nessai.gw.legacy.logger
- nessai.gw.legacy.determine_rescaled_bounds(prior_min, prior_max, x_min, x_max, invert)
Determine the values of the prior min and max in the rescaled space. Parameters ———- prior_min : float
Mininum of the prior
- prior_maxfloat
Maximum of the prior
- x_minfloat
New minimum
- x_maxfloat
New maximum
- invertfalse or {‘upper’, ‘lower’, ‘both’}
Type of inversion
- nessai.gw.legacy.angle_to_cartesian(alpha, r=None, scale=1.0)
Decompose an angle into a real and imaginary part
- nessai.gw.legacy.cartesian_to_angle(x, y, scale=1.0, zero='centre')
Reconstruct an angle given the real and imaginary part. Assume the angle is defined on [0, 2 pi] / scale.
- Parameters
- x, yarray_like
Cartesian coordinates
- scalefloat, optional
Rescaling factor used to rescale from [0, 2pi]
- zerostr, {centre, bound}
Specifiy is zero should be the central value or lower bound
- nessai.gw.legacy.ra_dec_to_cartesian(ra, dec, dL=None)
Convert right ascension, declination and (optional) luminosity distance (defined on [0, 1]) to Cartesian coordinates.
- Parameters
- ra, dec: array_like
Right ascension and declination
- dL: array_like, optional
Corresponding luminosity distance defined on [0, 1]. If None (default) radial componment is drawn froma chi distribution with 3 degrees of freedom
- Returns
- x, y, z: array_like
Cartesian coordinates
- log_J: array_like
Determinant of the log-Jacobian
- nessai.gw.legacy.cartesian_to_ra_dec(x, y, z)
Reconstruct an angle given the real and imaginary part
- Parameters
- x, y, z: array_like
Three dimensional Cartesian coordinates
- Returns:
- ra, dec: array_like
Right ascension and declination
- dl: array_like
Luminosity distance
- log_J: array_like
Determinant of the log-Jacobian
- nessai.gw.legacy.azimuth_zenith_to_cartesian(azimuth, zenith, dL=None)
Convert azimuth, zenith and (optional) luminosity distance (defined on [0, 1]) to Cartesian coordinates.
- Parameters
- azimuth, zenith: array_like
Azimuth and zenith
- dL: array_like, optional
Corresponding luminosity distance defined on [0, 1]. If None (default) radial componment is drawn froma chi distribution with 3 degrees of freedom
- Returns
- x, y, z: array_like
Cartesian coordinates
- log_J: array_like
Determinant of the log-Jacobian
- nessai.gw.legacy.cartesian_to_azimuth_zenith(x, y, z)
Reconstruct an angle given the real and imaginary part
- Parameters
- x, y, z: array_like
Three dimensional Cartesian coordinates
- nessai.gw.legacy.zero_one_to_cartesian(theta, mode='split')
Convert a variable defined on [0,1] to an angle on [-pi, pi] and to Cartesian coordinates with a radius drawn from a chi distribution with two degrees of freedom. The lower bound is place at 0 and the upper bound at -pi/pi.
- Parameters
- theta: array_like
Array of values bound on [0, 1]
- Returns
- x, y: array_like
Cartesian coordinates
- log_J: array_like
Determinant of the log-Jacobian
- nessai.gw.legacy.cartesian_to_zero_one(x, y)
Convert Cartesian coordinates to a variable defined on [0, 1] and a corresponding radius.
- Parameters
- x, y: array_like
Cartesian coordinates
- Returns
- theta: array_like
Variable defined on [0,1]
- radius: array_like
Corresponding radius
- log_J: array_like
Determinant of log-Jacobian
- nessai.gw.legacy.transform_from_precessing_parameters(theta_jn, phi_jl, theta_1, theta_2, phi_12, a_1, a_2, m1, m2, f_ref, phase)
- nessai.gw.legacy.transform_to_precessing_parameters(iota, s1x, s1y, s1z, s2x, s2y, s2z, m1, m2, f_ref, phase)
- nessai.gw.legacy.rescale_and_logit(x, xmin, xmax)
- nessai.gw.legacy.rescale_and_sigmoid(x, xmin, xmax)
- nessai.gw.legacy.log_uniform_prior(x, xmin=- 1, xmax=1)
Unformalised log probability of uniform prior
- nessai.gw.legacy.log_2d_cartesian_prior(x, y, k=np.pi)
Log probability of Cartesian coordinates for a uniform distibution of angles on [0, k] and a radial component drawn from a chi distribution with two degrees of freedom.
- nessai.gw.legacy.log_2d_cartesian_prior_sine(x, y)
Log probability of Cartesian coordinates for a sine distibution of angles and a radial component drawn from a chi distribution with two degrees of freedom.
- nessai.gw.legacy.log_3d_cartesian_prior(x, y, z)
Log probability of 3d Cartesian coordinates for an isotropic distribution of angles and a radial component drawn from a chi distribution with three degrees of freedom.
- nessai.gw.legacy.log_spin_prior(s1x, s1y, s1z, s2x, s2y, s2z, k1=0.99, k2=0.99)
Log probability of the prior on the components of spin vectors assume the distribution of a_i is uniform.
- nessai.gw.legacy.log_spin_prior_uniform(s1x, s1y, s1z, s2x, s2y, s2z, k1=0.99, k2=0.99)
Log probability of the prior on the components of spin vectors assume the distribution the components in uniform within the 2-ball.
- class nessai.gw.legacy.LegacyGWFlowProposal(model, reparameterisations={}, **kwargs)
Bases:
nessai.proposal.FlowProposalA proposal specific to gravitational wave CBC
- set_reparameterisations(self, reparameterisations)
Set the relevant reparamterisation flags
- setup_angle(self, name, radial_name=False, scale=1.0, zero='bound')
Add an angular parameter to the list of reparameterisations
- property inversion_parameters(self)
Returns a list of parameters to which an inversion (normal or log)
- add_inversion(self, name)
Setup inversion
- add_log_inversion(self, name)
Setup log inversion
- add_angle_conversion(self, name, mode='split')
- configure_time(self)
Configure the time parameter if present
- configure_sky(self)
Configure the sky parameters
- configure_angles(self)
Configure angles
- set_rescaling(self)
Set the rescaling functions
- check_state(self, x)
Check the state of the rescaling before training
- update_rescaled_bounds(self, rescaled_names=None, xmin=None, xmax=None)
- setup_uniform_distance_parameter(self, scale_factor=1000, **kwargs)
Set up the uniform distance parameter dc3
- Parameters
- scale_factorfloat, (optional)
Factor used to rescale comoving distance
- kwargs
Keyword arguments parsed to ComovingDistanceConverter
- convert_to_dl(self, dc3)
Convert from uniform distance parameter dc3 to luminosity distance
- convert_to_dc3(self, dl)
Convert to uniform distance parameter dc3
- setup_spin_logit(self, fuzz_factor=0.01)
- configure_spin_conversion(self, m1=20, m2=20, phase=0, f_ref=20, scale_factor=1, use_cbrt=True)
- rescale(self, x, compute_radius=False, test=None)
Rescale from the x space to the x prime space
- inverse_rescale(self, x_prime)
Rescale from the x prime space to the x space
- log_prior(self, x)
Modified log prior that handles radial parameters
- compute_rescaled_bounds(self, name)
- x_prime_log_prior(self, x_prime)
Priors redefined in the x_prime space