jesterTOV.inference.transforms.transform.JesterTransform

jesterTOV.inference.transforms.transform.JesterTransform#

class JesterTransform(
eos,
tov_solver,
name_mapping=None,
keep_names=None,
ndat_TOV=100,
min_nsat_TOV=0.75,
fixed_params=None,
**kwargs,
)[source]#

Bases: NtoMTransform

Transform EOS parameters to neutron star observables (M, R, Λ).

This is the main transform class that combines an equation of state (EOS) model with a TOV solver to produce neutron star observables from microscopic EOS parameters.

The transform can be created either by: 1. Passing EOS and TOV solver instances directly 2. Using from_config() classmethod with configuration dict/object

Parameters:

  • eos (Interpolate_EOS_model) – EOS model instance (MetaModel, MetaModelCSE, Spectral, etc.)

  • tov_solver (TOVSolverBase) – TOV solver instance (GRTOVSolver, AnisotropyTOVSolver, ScalarTensorTOVSolver)

  • name_mapping (tuple[list[str], list[str]] | None) – Tuple of (input_names, output_names). If None, constructed from EOS and TOV required parameters.

  • keep_names (list[str] | None) – Parameter names to preserve in output. If None, keeps all inputs.

  • ndat_TOV (int) – Number of central pressure points for M-R-Λ curves (default: 100)

  • min_nsat_TOV (float) – Minimum density for TOV integration in units of nsat (default: 0.75)

  • **kwargs (Any) – Additional parameters (for compatibility)

Variables:

  • eos (Interpolate_EOS_model) – The equation of state model

  • tov_solver (TOVSolverBase) – The TOV equation solver

  • eos_params (list[str]) – Parameters required by the EOS

  • tov_params (list[str]) – Parameters required by the TOV solver

  • keep_names (list[str]) – Parameters to preserve in output

Examples

>>> # Direct instantiation
>>> from jesterTOV.eos.metamodel import MetaModel_EOS_model
>>> from jesterTOV.tov.gr import GRTOVSolver
>>> eos = MetaModel_EOS_model(crust_name="DH")
>>> solver = GRTOVSolver()
>>> transform = JesterTransform(eos=eos, tov_solver=solver)

>>> # From configuration
>>> from jesterTOV.inference.config.schema import MetamodelCSEEOSConfig, TOVConfig
>>> eos_config = MetamodelCSEEOSConfig(type="metamodel_cse", nb_CSE=8)
>>> tov_config = TOVConfig(type="gr")
>>> transform = JesterTransform.from_config(eos_config, tov_config)

>>> # Transform parameters to observables
>>> params = {"E_sat": -16.0, "K_sat": 230.0, ...}
>>> result = transform.forward(params)
>>> print(result["masses_EOS"])  # Neutron star masses in M☉
__init__(
eos,
tov_solver,
name_mapping=None,
keep_names=None,
ndat_TOV=100,
min_nsat_TOV=0.75,
fixed_params=None,
**kwargs,
)[source]#

Methods

__init__(eos, tov_solver[, name_mapping, ...])

construct_eos_and_solve_tov(params)

Construct EOS from parameters and solve TOV equations.

forward(x)

Transform parameters and preserve keep_names.

from_config(eos_config, tov_config[, ...])

Create transform from configuration objects.

get_eos_type()

Return EOS type identifier.

get_parameter_names()

Return combined list of sampled EOS and TOV parameter names.

propagate_name(x)

Propagate parameter names through the transform.

Attributes

transform_func

name_mapping

construct_eos_and_solve_tov(params)[source]#

Construct EOS from parameters and solve TOV equations.

This is the core transformation method that: 1. Constructs EOS from parameters 2. Solves TOV equations for M-R-Λ family 3. Returns observables with constraint checking

Parameters:

params (dict[str, Float]) – Input parameters (EOS + TOV parameters)

Return type:

dict[str, Float | Float[Array, 'n']]

Returns:

dict[str, Float | Float[Array, “ n”]] – Dictionary containing: - masses_EOS : Neutron star masses [M☉] - radii_EOS : Neutron star radii [km] - Lambdas_EOS : Tidal deformabilities - logpc_EOS : Log10 central pressures - n, p, h, e, dloge_dlogp, cs2 : EOS quantities - Constraint violation counts

forward(x)[source]#

Transform parameters and preserve keep_names.

This overrides NtoMTransform.forward() to:

  1. Merge fixed parameters into x before the EOS/TOV pipeline runs.

  2. Preserve parameters specified in self.keep_names.

  3. Add fixed parameters to the output so they appear in the result.

Parameters:

x (dict[str, Float]) – Input parameter dictionary (sampled parameters only)

Return type:

dict[str, Float]

Returns:

dict[str, Float] – Transformed parameters with keep_names and fixed_params included

classmethod from_config(
eos_config,
tov_config,
keep_names=None,
max_nbreak_nsat=None,
fixed_params=None,
)[source]#

Create transform from configuration objects.

This factory method instantiates the appropriate EOS and TOV solver based on the separate configurations, then creates the transform.

Parameters:

  • eos_config (EOSConfig) – EOS configuration (MetamodelEOSConfig, MetamodelCSEEOSConfig, or SpectralEOSConfig)

  • tov_config (TOVConfig) – TOV solver configuration

  • keep_names (list[str] | None) – Parameters to preserve in output

  • max_nbreak_nsat (float | None) – Maximum nbreak value (for MetaModelCSE optimization)

  • fixed_params (dict[str, float] | None) – Parameters pinned to constant values, excluded from the sampling space but injected into every forward() call.

Return type:

JesterTransform

Returns:

JesterTransform – Configured transform instance

Raises:

ValueError – If EOS or TOV type is unknown

get_eos_type()[source]#

Return EOS type identifier.

Return type:

str

Returns:

str – EOS class name (e.g., ‘MetaModel_EOS_model’)

get_parameter_names()[source]#

Return combined list of sampled EOS and TOV parameter names.

Fixed parameters (those pinned via Fixed(...) in the prior file) are excluded — they are not part of the sampling space.

Return type:

list[str]

Returns:

list[str] – Sampled parameter names required by this transform

Contents