cern.colt.matrix.tfloat.algo.decomposition
Class FloatSingularValueDecompositionDC

java.lang.Object
  cern.colt.matrix.tfloat.algo.decomposition.FloatSingularValueDecompositionDC

public class FloatSingularValueDecompositionDC
extends Object

For an m x n matrix A, the singular value decomposition is an m x m orthogonal matrix U, an m x n diagonal matrix S, and an n x n orthogonal matrix V so that A = U*S*V'.

The singular values, sigma[k] = S[k][k], are ordered so that sigma[0] >= sigma[1] >= ... >= sigma[min(m-1,n-1)].

This implementation uses the divide-and-conquer algorithm (Dgesdd) from JLAPACK.

Author:

Piotr Wendykier (piotr.wendykier@gmail.com)

Constructor Summary

FloatSingularValueDecompositionDC(FloatMatrix2D Arg, boolean wantUV, boolean wantWholeUV)
Constructs and returns a new singular value decomposition object; The decomposed matrices can be retrieved via instance methods of the returned decomposition object.

Method Summary

float	cond() Returns the two norm condition number, which is max(S) / min(S).
org.netlib.util.intW	getInfo() Returns the output flag
FloatMatrix2D	getS() Returns the diagonal matrix of singular values.
float[]	getSingularValues() Returns the diagonal of S, which is a one-dimensional array of singular values
FloatMatrix2D	getU() Returns the left singular vectors U.
FloatMatrix2D	getUt() Returns the transpose of the left singular vectors U.
FloatMatrix2D	getV() Returns the right singular vectors V.
FloatMatrix2D	getVt() Returns the transpose of the right singular vectors V.
float	norm2() Returns the two norm, which is max(S).
int	rank() Returns the effective numerical matrix rank, which is the number of nonnegligible singular values.
String	toString() Returns a String with (propertyName, propertyValue) pairs.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

FloatSingularValueDecompositionDC

public FloatSingularValueDecompositionDC(FloatMatrix2D Arg,
                                         boolean wantUV,
                                         boolean wantWholeUV)

Constructs and returns a new singular value decomposition object; The decomposed matrices can be retrieved via instance methods of the returned decomposition object.

Parameters:

Arg - rectangular matrix

wantUV -
true: all matrices (U, S, V') are computed
false: only S is computed

wantWholeUV -
true: all m columns of U and all n rows of V' are computed
false: the first min(m,n) columns of U and the first min(m,n) rows of V' are computed

Method Detail

cond

public float cond()

Returns the two norm condition number, which is max(S) / min(S).

getS

public FloatMatrix2D getS()

Returns the diagonal matrix of singular values.

Returns:

S

getSingularValues

public float[] getSingularValues()

Returns the diagonal of S, which is a one-dimensional array of singular values

Returns:

diagonal of S.

getU

public FloatMatrix2D getU()

Returns the left singular vectors U.

Returns:

U

getUt

public FloatMatrix2D getUt()

Returns the transpose of the left singular vectors U.

Returns:

U'

getVt

public FloatMatrix2D getVt()

Returns the transpose of the right singular vectors V.

Returns:

V'

getV

public FloatMatrix2D getV()

Returns the right singular vectors V.

Returns:

V

getInfo

public org.netlib.util.intW getInfo()

Returns the output flag

Returns:

0: successful exit
< 0: if INFO = -i, the i-th argument had an illegal value
> 0: process did not converge.

norm2

public float norm2()

Returns the two norm, which is max(S).

rank

public int rank()

Returns the effective numerical matrix rank, which is the number of nonnegligible singular values.

toString

public String toString()

Returns a String with (propertyName, propertyValue) pairs. Useful for debugging or to quickly get the rough picture. For example,

         rank          : 3
         trace         : 0
 
 

Overrides:

toString in class Object