Click or drag to resize

DoubleCOWeightedLeastSq Class

Class DoubleCOWeightedLeastSq solves weighted least squares problems by using a Complete Orthogonal (CO) decomposition technique.
Inheritance Hierarchy
SystemObject
  CenterSpace.NMath.CoreDoubleCOWeightedLeastSq

Namespace:  CenterSpace.NMath.Core
Assembly:  NMath (in NMath.dll) Version: 7.3
Syntax
public class DoubleCOWeightedLeastSq

The DoubleCOWeightedLeastSq type exposes the following members.

Constructors
  NameDescription
Public methodDoubleCOWeightedLeastSq
Default constructor. Instances created with this constructor will be empty and unsuable until the Factor method is called.
Public methodDoubleCOWeightedLeastSq(DoubleMatrix, DoubleVector)
Constructucts a DoubleCOWeightedLeastSq instance from the given matrix and weights.
Public methodDoubleCOWeightedLeastSq(DoubleMatrix, DoubleVector, Boolean)
Constructucts a DoubleCOWeightedLeastSq instance from the given matrix and weights.
Top
Properties
  NameDescription
Public propertyA
Gets the matrix A used in the calculation.
Top
Methods
  NameDescription
Public methodClone
Creates a deep copy of this weighted least squares instance.
Public methodFactor(DoubleMatrix, DoubleVector)
Performs any factorization on the matrix A necessary before computing a solution to the weighted least squares problem.
Public methodFactor(DoubleMatrix, DoubleVector, Boolean)
Performs any factorization on the matrix A necessary before computing a solution to the weighted least squares problem.
Public methodResidualNormSqr
Computes the 2-norm squared of the residual vector.
Public methodResidualVector
Computes and returns the residual vector.
Public methodReweight
Performs necessary computations for a change of weights.
Public methodSolve
Compute the solution to the weighted least squares problem.
Top
Remarks
Use class DoubleCOWeightedLeastSq to find the minimal weighted norm solution to the overdetermined linear system:
Ax = b
That is, find the vector x that minimizes the 2-norm of the weighted residual vector (D^-1/2)*(Ax - b). Where D is a digaonal matrix with non-negative values on the diagonal. Prerequisites on the matrix A are that it has more rows than columns, and is of full rank. The Alogorithm satisfies an accuracy bound that is not affected by ill conditioning in the weight matrix D.
Reference: Complete Orthogonal Decomposition For Weighted Least Squares Patricia D. Hough and Stephen A. Vavasis SIAM J. Matrix Anal. Appl. Vol. 18, No. 2, pp 369-392, April 1997
See Also