Package jdistlib.accelerator

Class PivotedQrFactor

java.lang.Object
jdistlib.accelerator.PivotedQrFactor
public final class PivotedQrFactor extends Object
Immutable column-pivoted Householder QR factorization.

  • Constructor Summary

    Constructors
    Constructor
    Description
    PivotedQrFactor(int rows, int columns, double[] qr, double[] tau, int[] pivot)
    Creates a factor from packed Householder QR storage and a zero-based pivot.

  • Method Summary

    Modifier and Type
    Method
    Description
    int
    columns()
    Returns the number of columns in the original matrix.
    double[]
    packed()
    Returns a row-major copy containing R and the packed Householder vectors.
    int[]
    pivot()
    Returns the permutation where factor column j came from original column pivot[j].
    int
    rank()
    Estimates numerical rank using max(rows,columns)*ulp(1)*max(abs(diag(R))).
    int
    rank(double tolerance)
    Returns the number of diagonal entries of R larger than an absolute tolerance.
    int
    rows()
    Returns the number of rows in the original matrix.
    double[]
    solveLeastSquares(double[] right)
    Solves a full-column-rank least-squares problem and returns coefficients in original order.

    Methods inherited from class java.lang.Object

    clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

  • Constructor Details

    • PivotedQrFactor

      public PivotedQrFactor(int rows, int columns, double[] qr, double[] tau, int[] pivot)
      Creates a factor from packed Householder QR storage and a zero-based pivot.

  • Method Details

    • rows

      public int rows()
      Returns the number of rows in the original matrix.

    • columns

      public int columns()
      Returns the number of columns in the original matrix.

    • pivot

      public int[] pivot()
      Returns the permutation where factor column j came from original column pivot[j].

    • packed

      public double[] packed()
      Returns a row-major copy containing R and the packed Householder vectors.

    • rank

      public int rank()
      Estimates numerical rank using max(rows,columns)*ulp(1)*max(abs(diag(R))).

    • rank

      public int rank(double tolerance)
      Returns the number of diagonal entries of R larger than an absolute tolerance.

    • solveLeastSquares

      public double[] solveLeastSquares(double[] right)
      Solves a full-column-rank least-squares problem and returns coefficients in original order.