`altar.models.cudalinear.cudaLinear`

Module Contents

Classes

class altar.models.cudalinear.cudaLinear.cudaLinear(name, locator, **kwds)

Bases: altar.cuda.models.cudaBayesian.cudaBayesian

Cuda implementation of a linear model A linear model is defined as data = G theta

dataobs

default

doc = the observed data

green

doc = the name of the file with the Green functions

GF

gGF

gDataPred

cublas_handle

initialize(self, application): Initialize the state of the model given a {problem} specification

_forwardModel(self, theta, prediction, batch, observation=None): Linear Forward Model prediction= G theta

cuEvalLikelihood(self, theta, likelihood, batch): to be loaded by super class cuEvalLikelihood which already decides where the local likelihood is added to

loadGF(self): Load the data in the input files into memory

prepareGF(self): copy green function to gpu and merge cd with green function