Class
NumCosmoGalaxyShapeFactor
Description [src]
abstract class NumCosmo.GalaxyShapeFactor : GObject.Object
{
/* No available fields */
}Shape likelihood-factor calculator for the weak-lensing pipeline.
The calculator owns the whole HSM measurement engine: the intrinsic
ellipticity $\chi_I$ is drawn from the NcGalaxyShapePop resolved from the
NcmMSet, mapped deterministically by the reduced-shear transformation with
calibration bias $\tilde g = (1+m)\,g + c$, and observed with additive
per-galaxy Gaussian pixel noise,
$\epsilon_\mathrm{obs} = f_{\tilde g}(\chi_I) + n$, $n \sim N_2(0, \sigma_\mathrm{noise}^2)$.
The likelihood factor is the marginal over the latent intrinsic shape, $$P(\epsilon_\mathrm{obs} \mid z, \ldots) = \int_{|\chi_I| < 1} d^2\chi_I\, P_\mathrm{pop}(\chi_I)\, N_2!\big(\epsilon_\mathrm{obs} - f_{\tilde g}(\chi_I); \sigma_\mathrm{noise}^2\big),$$ evaluated per galaxy at each source redshift $z$ (the outer $z$-integral belongs to the orchestrator). How this two-dimensional marginal is computed is the only axis subclasses vary: each subclass is one evaluation strategy implementing the eval_marginal / eval_ln_marginal hooks. Everything else - generation, per-galaxy geometry caches (projected radius, optzs, critical surface-density nodes), frame bookkeeping and data IO - lives here, written once.
This is a calculator, not a model: it holds no fitted parameters and does
not live in the NcmMSet; the models it needs (cosmology, halo position and
profile, surface mass density, shape population) are resolved from the mset
passed to each method.
Functions
nc_galaxy_shape_factor_clear
Decreases the reference count of gsf by one, and sets the pointer *gsf to
NULL.
Instance methods
nc_galaxy_shape_factor_apply_shear
Applies the reduced shear g to the intrinsic ellipticity E, storing the
resulting observed ellipticity in E_obs. The transformation depends on the
NcGalaxyWLObsEllipConv configured in gsf.
nc_galaxy_shape_factor_apply_shear_inv
Applies the inverse shear transformation using g to recover the intrinsic
ellipticity E from the observed ellipticity E_obs. The transformation
depends on the NcGalaxyWLObsEllipConv configured in gsf.
nc_galaxy_shape_factor_check_obs
Checks that obs‘s ellipticity convention matches gsf‘s own
NcGalaxyWLObsEllipConv. Validation the shape factor owns (it is the one
that knows what convention it expects), not its callers.
nc_galaxy_shape_factor_data_get
Gets the observed ellipticity and the per-galaxy measurement quantities.
nc_galaxy_shape_factor_data_set
Sets the observed ellipticity and the per-galaxy measurement quantities.
nc_galaxy_shape_factor_direct_estimate
Computes the inverse-variance-weighted direct estimate of the reduced shear from the observed ellipticities. The per-galaxy intrinsic scatter enters through the population model’s rms (see nc_galaxy_shape_pop_e_rms()).
nc_galaxy_shape_factor_eval_at_nodes
Evaluates the shape likelihood P(epsilon_obs | z_j, data) at every node,
writing the result into out. Requires a prior call to nc_galaxy_shape_factor_prepare_data_array_at_nodes().
nc_galaxy_shape_factor_eval_ln_marginal
Logarithmic counterpart of nc_galaxy_shape_factor_eval_marginal().
nc_galaxy_shape_factor_eval_marginal
Evaluates the intrinsic-ellipticity marginal
P(epsilon_obs | g) = int d^2chi_I P_pop(chi_I) N_2(epsilon_obs - f_g(chi_I); std_noise^2)
using the subclass evaluation strategy. The shear and the observed
ellipticity must be expressed in the same frame; data supplies the
per-galaxy noise (std_noise) and the population fragment (pop_data).
nc_galaxy_shape_factor_gen
Generates a galaxy shape: the intrinsic ellipticity is drawn from the
NcGalaxyShapePop in mset, sheared with the reduced shear at the galaxy’s
(ra, dec, z) with the calibration bias applied, and observed with additive
Gaussian noise. The per-galaxy measurement quantities (std_noise, c1, c2, m)
and the frame data->coord must be set beforehand (see
nc_galaxy_shape_factor_data_set()). The shape is generated in the celestial
frame and mapped to data->coord (see NcWLEllipticityFrame); since the
intrinsic shape and noise are isotropic the generation is frame-symmetric,
while the additive bias (c1, c2) is taken to be already expressed in
data->coord.
nc_galaxy_shape_factor_get_crit_hash
Hash of the models the critical surface mass density depends on: the cosmology, the surface-mass-density model and the lens redshift (halo_position).
nc_galaxy_shape_factor_integ
Creates a new shape integrand P(epsilon_obs | z, data). The marginalization hook (eval_marginal or eval_ln_marginal) is resolved once here, out of the per-evaluation path.
nc_galaxy_shape_factor_lndet_jac
Computes the natural logarithm of the absolute value of the Jacobian determinant of the transformation from intrinsic to observed ellipticity.
nc_galaxy_shape_factor_prepare
Factory-level prepare: refreshes the geometry/profile caches shared by
every galaxy (only recomputing whichever half actually went stale, judged
by hashing the models each half depends on), refreshes the cached model
references NcGalaxyShapeFactor‘s own per-Data updates use
(nc_galaxy_shape_factor_update_data_radius(),
nc_galaxy_shape_factor_update_data_optzs(),
nc_galaxy_shape_factor_update_data_pop()), then calls the subclass’s own
prepare hook. Cheap/no-op when nothing relevant changed — safe and
correct to call unconditionally every orchestrator prepare() cycle.
nc_galaxy_shape_factor_prepare_data_array
Prepares the per-galaxy caches used by the integrand: lens geometry (projected radius, tangential-frame rotation of the observed ellipticity and calibration bias) and the reduced-shear optimization cache. Also prepares the population fragments and runs the subclass prepare hook.
nc_galaxy_shape_factor_prepare_data_array_at_nodes
Pre-computes cached quantities (Sigma_crit per node, Sigma(R)) for the fixed quadrature path. This unconditionally (re)computes every per-galaxy node cache: deciding when a recompute is needed is the caller’s responsibility.
nc_galaxy_shape_factor_update_data_at_nodes_crit
Refreshes data‘s per-node critical surface density cache, using the lens
context cached by the last nc_galaxy_shape_factor_prepare() call. This
cache depends only on self->lens_ctx/self->cosmo/
self->surface_mass_density (not on density_profile or the projected
radius), and lens_ctx is refreshed whenever
nc_galaxy_shape_factor_get_optzs_hash() changes (see prepare()’s own doc
comment) — so call this whenever that hash changed, or whenever the node
grid itself was rebuilt (different/resized z_nodes).
nc_galaxy_shape_factor_update_data_at_nodes_sigma
Refreshes data‘s cached surface mass density Sigma(R) used by the
fixed-nodes path, using data‘s own cached projected radius as an input.
Unconditional: call only when nc_galaxy_shape_factor_get_radius_hash() or
nc_galaxy_shape_factor_get_optzs_hash() changed, and only after
nc_galaxy_shape_factor_update_data_radius() in the same cycle if that one
also needs to run — this assumes data‘s cached projected radius is
already current (same ordering contract as update_data_optzs()).
nc_galaxy_shape_factor_update_data_optzs
Refreshes data‘s cached reduced-shear optimization context, using the
profile cache from the last nc_galaxy_shape_factor_prepare() call.
Unconditional: call only when nc_galaxy_shape_factor_get_optzs_hash()
actually changed, and only after
nc_galaxy_shape_factor_update_data_radius() in the same cycle if that one
also needs to run — this assumes data‘s cached projected radius is
already current.
nc_galaxy_shape_factor_update_data_pop
Resolves the NcGalaxyShapePop model’s current parameters into data‘s
population fragment. Unconditional: call only when
nc_galaxy_shape_factor_get_pop_hash() actually changed.
nc_galaxy_shape_factor_update_data_radius
Refreshes data‘s cached projected radius, tangential-frame rotation of
the observed ellipticity, and rotated calibration bias, using the
geometry cache from the last nc_galaxy_shape_factor_prepare() call.
Unconditional: call only when nc_galaxy_shape_factor_get_radius_hash()
actually changed.
Properties
NumCosmo.GalaxyShapeFactor:ellip-conv
Weak lensing observables ellipticity convention NcGalaxyWLObsEllipConv.
Signals
Signals inherited from GObject (1)
GObject::notify
The notify signal is emitted on an object when one of its properties has its value set through g_object_set_property(), g_object_set(), et al.
Class structure
struct NumCosmoGalaxyShapeFactorClass {
void (* data_init) (
NcGalaxyShapeFactor* gsf,
NcmMSet* mset,
NcGalaxyShapeFactorData* data
);
void (* prepare) (
NcGalaxyShapeFactor* gsf,
NcmMSet* mset
);
gdouble (* eval_marginal) (
NcGalaxyShapeFactor* gsf,
NcGalaxyShapePop* pop,
NcGalaxyShapeFactorData* data,
const gdouble g_1,
const gdouble g_2,
const gdouble epsilon_obs_1,
const gdouble epsilon_obs_2
);
gdouble (* eval_ln_marginal) (
NcGalaxyShapeFactor* gsf,
NcGalaxyShapePop* pop,
NcGalaxyShapeFactorData* data,
const gdouble g_1,
const gdouble g_2,
const gdouble epsilon_obs_1,
const gdouble epsilon_obs_2
);
}No description available.
Class members
data_init: void (* data_init) ( NcGalaxyShapeFactor* gsf, NcmMSet* mset, NcGalaxyShapeFactorData* data )No description available.
prepare: void (* prepare) ( NcGalaxyShapeFactor* gsf, NcmMSet* mset )No description available.
eval_marginal: gdouble (* eval_marginal) ( NcGalaxyShapeFactor* gsf, NcGalaxyShapePop* pop, NcGalaxyShapeFactorData* data, const gdouble g_1, const gdouble g_2, const gdouble epsilon_obs_1, const gdouble epsilon_obs_2 )No description available.
eval_ln_marginal: gdouble (* eval_ln_marginal) ( NcGalaxyShapeFactor* gsf, NcGalaxyShapePop* pop, NcGalaxyShapeFactorData* data, const gdouble g_1, const gdouble g_2, const gdouble epsilon_obs_1, const gdouble epsilon_obs_2 )No description available.
Virtual methods
NumCosmo.GalaxyShapeFactorClass.eval_ln_marginal
Logarithmic counterpart of nc_galaxy_shape_factor_eval_marginal().
NumCosmo.GalaxyShapeFactorClass.eval_marginal
Evaluates the intrinsic-ellipticity marginal
P(epsilon_obs | g) = int d^2chi_I P_pop(chi_I) N_2(epsilon_obs - f_g(chi_I); std_noise^2)
using the subclass evaluation strategy. The shear and the observed
ellipticity must be expressed in the same frame; data supplies the
per-galaxy noise (std_noise) and the population fragment (pop_data).
NumCosmo.GalaxyShapeFactorClass.prepare
Factory-level prepare: refreshes the geometry/profile caches shared by
every galaxy (only recomputing whichever half actually went stale, judged
by hashing the models each half depends on), refreshes the cached model
references NcGalaxyShapeFactor‘s own per-Data updates use
(nc_galaxy_shape_factor_update_data_radius(),
nc_galaxy_shape_factor_update_data_optzs(),
nc_galaxy_shape_factor_update_data_pop()), then calls the subclass’s own
prepare hook. Cheap/no-op when nothing relevant changed — safe and
correct to call unconditionally every orchestrator prepare() cycle.