NormalsSolverVCE

This program accumulates normalEquation and solves the total combined system. The relative weigthing between the individual normals is determined iteratively by means of variance component estimation (VCE). For a detailed description of the used algorithm see normalEquation.

Besides the estimated parameter vector (outputfileSolution) the estimated accuracies (outputfileSigmax) and the full covariance matrix (outputfileCovariance) can be saved. Also the combined normal system can be written to outputfileNormalEquation.

The outputfileContribution is a matrix with rows for each estimated parameter and columns for each normalEquation and indicates the contribution of the individual normals to the estimated parameters. Each row sum up to one.

See also NormalsBuild.

Name	Type	Annotation
outputfileSolution	filename	parameter vector
outputfileSigmax	filename	standard deviations of the parameters (sqrt of the diagonal of the inverse normal equation)
outputfileCovariance	filename	full covariance matrix
outputfileContribution	filename	contribution of normal system components to the solution vector
outputfileVarianceFactors	filename	estimated variance factors as vector
outputfileNormalEquation	filename	the combined normal equation system
normalEquation	normalEquation
inputfileApproxSolution	filename	to accelerate convergence
rightHandSideNumberVCE	uint	the right hand side number for estimation of variance factors
normalsBlockSize	uint	block size for distributing the normal equations, 0: one block
maxIterationCount	uint	maximum number of iterations for variance component estimation

This program is parallelized.