rxmc.observation.FixedCovarianceObservation

class rxmc.observation.FixedCovarianceObservation(x: ndarray, y: ndarray, covariance: ndarray)

Bases: Observation

A class to represent an observation with fixed covariance. That is, the covariance matrix for the Multivariate Gaussian likelihood for a model prediction ym is known a priori and does not change with the model prediction.

The simplest such case is when the covariance is a diagonal matrix containing the reported statistical variances for each data point in y.

In the case that the covariance is a vector, it is interpreted as the diagonal of the covariance matrix, and the likelihood reduces to the standard form using the chi-squared statistic. In the case that the covariance is a full matrix, this corresponds to the generalised chi-squared statistic.

__init__(x: ndarray, y: ndarray, covariance: ndarray)

Initializes the FixedCovarianceObservation with data and a fixed covariance matrix.

Parameters:

• x (np.ndarray) – The independent variable data.

• y (np.ndarray) – The dependent variable data.

• covariance (np.ndarray) – The fixed covariance matrix associated with the observation.

Methods

__init__(x, y, covariance)	Initializes the FixedCovarianceObservation with data and a fixed covariance matrix.
covariance(y)	Returns the fixed covariance matrix for the observation, which is constant and does not depend on y.
num_pts_within_interval(ylow, yhigh[, xlim])	Returns the number of points in y that fall between ylow and yhigh, useful for calculating emperical coverages
residual(ym)

covariance(y)

Returns the fixed covariance matrix for the observation, which is constant and does not depend on y.

Parameters:

y (np.ndarray) – The dependent variable data for which to compute the covariance.