spey.multiparameter.contour.ContourResult#

class spey.multiparameter.contour.ContourResult(theta_mle: ndarray, nll_min: float, threshold: float, delta: float, contour_points: ndarray, from_radial: ndarray, parameter_names: List[str] | None, confidence_level: float, dof: int)[source]#

Container for the output of find_contour().

theta_mle#

Maximum-likelihood estimate \(\hat\theta\), shape (k,).

Type:

np.ndarray

nll_min#

Minimum negative log-likelihood \(\mathrm{NLL}(\hat\theta)\).

Type:

float

threshold#

NLL value that defines the contour boundary, \(T = \mathrm{NLL}(\hat\theta) + \Delta_\alpha / 2\).

Type:

float

delta#

Chi-squared quantile \(\Delta_\alpha = F^{-1}_{\chi^2_k}(1-\alpha)\).

Type:

float

contour_points#

Points on the contour boundary, shape (n_points, k). Every row \(\theta^*\) satisfies \(|\mathrm{NLL}(\theta^*) - T| \lesssim \varepsilon_\text{tol}\).

Type:

np.ndarray

from_radial#

Boolean mask of shape (n_points,). True for points produced by the radial search; False for points added by the constrained RATTLE walk.

Type:

np.ndarray

parameter_names#

Names of the \(k\) parameters in the same order as the columns of contour_points, or None when the model does not provide names.

Type:

Optional[List[str]]

confidence_level#

The confidence level \(1-\alpha\), e.g. 0.95.

Type:

float

dof#

Degrees of freedom \(k\) (number of model parameters).

Type:

int

__init__(theta_mle: ndarray, nll_min: float, threshold: float, delta: float, contour_points: ndarray, from_radial: ndarray, parameter_names: List[str] | None, confidence_level: float, dof: int) None#

Methods

__init__(theta_mle, nll_min, threshold, ...)

Attributes