`altar.data.DataL2`

Module Contents

Classes

class altar.data.DataL2.DataL2(name, locator, **kwds)

Bases: altar.component

The observed data with L2 norm

data_file

doc = the name of the file with the observations

observations

doc = the number of observed data

cd_file

doc = the name of the file with the data covariance matrix

cd_std

doc = the constant covariance for data

merge_cd_with_data

doc = whether to merge cd with data

norm

default

doc = the norm to use when computing the data log likelihood

normalization = 0

ifs

samples

dataobs

dataobs_batch

cd

cd_inv

error

initialize(self, application): Initialize data obs from model

evalLikelihood(self, prediction, likelihood, residual=True, batch=None): compute the datalikelihood for prediction (samples x observations)

dataobsBatch(self): Get a batch of duplicated dataobs

loadData(self): load data and covariance

initializeCovariance(self, cd): For a given data covariance cd, compute L2 likelihood normalization, inverse of cd in Cholesky decomposed form, and merge cd with data observation, d-> L*d with cd^{-1} = L L* :param cd: :return:

updateCovariance(self, cp): Update data covariance with cp, cd -> cd + cp :param cp: a matrix with shape (obs, obs) :return:

computeNormalization(self, observations, cd): Compute the normalization of the L2 norm

computeCovarianceInverse(self, cd): Compute the inverse of the data covariance matrix