idl.tmt.representation.matrix
Class TmtMatrix

java.lang.Object
  |
  +--idl.tmt.representation.matrix.TmtMatrix

All Implemented Interfaces:

java.io.Serializable

public class TmtMatrix

extends java.lang.Object

implements java.io.Serializable

Encapsulates matrix representation and manipulation via the colt library.

This class contains many constructors, allowing the initialization of matrix operations from a variety of data structures, allowing both sparse and dense representation.

To insulate users from the details of colt, a TmTMatrix contains functionality for basic data analysis. For example, the class can compute, store, and provide access to the SVD of the matrix it contains. Likewise it can pre- or post-multiply itself with another TmtMatrix.

Author:

miles

See Also:

Serialized Form

Field Summary

private boolean	isSparse
private SingularValueDecomposition	mySVD
protected DoubleMatrix2D	theMatrix

Constructor Summary

TmtMatrix(double[][] denseMatrix)
constructor for a dense matrix from a 2D array of doubles

TmtMatrix(DoubleMatrix2D newMatrix)
constructor to build a matrix from a colt-formatted version

TmtMatrix(int rows, int cols, java.util.ArrayList rowInd, java.util.ArrayList colInd, java.util.ArrayList values)
constructor for a sparse matrix from data stored in memory

TmtMatrix(int rows, int cols, boolean sparse)
constructor for a basic zero matrix (dense)

TmtMatrix(java.lang.String fileName, boolean sparse)
constructor for reading a matrix (sparse or dense) from disk when we know the number of rows and columns beforehand

TmtMatrix(java.lang.String fileName, int rows, int cols, boolean sparse)
constructor for reading a matrix (sparse or dense) from disk when we know the number of rows and columns beforehand

TmtMatrix(TmtMatrix m)

Method Summary

TmtMatrix	addMatrix(TmtMatrix b) returns the matrix sum of this matrix with another matrix
int	columnCount() access to the number of columns in the matrix
TmtVector	getColumn(int i) get a column from the matrix
TmtMatrix	getColumnCentered() returns column-centered version of the matrix
TmtMatrix	getColumnNormalized() returns column-normalized (unit length) version of the matrix
TmtMatrix	getColumns(int start, int end) construct a new matrix containing a subset of the columns of this matrix
TmtMatrix	getCorrelationMatrix() computes and returns the correlation matrix of this matrix
TmtMatrix	getLeftSingularVectors() returns a TmtMatrix containing the left singular vectors of this matrix
DoubleMatrix2D	getMatrix()
TmtMatrix	getRightSingularVectors() returns a TmtMatrix containing the right singular vectors of this matrix
TmtVector	getRow(int i) get a row from the matrix
TmtMatrix	getRows(int start, int end) construct a new matrix containing a subset of the columns of this matrix
TmtMatrix	getSingularValues() returns a diagonal TmtMatrix containing the singular values of this matrix
double	getValue(int row, int column)
TmtMatrix	postMultiplyBy(TmtMatrix b) postmultiply this matrix by another matrix B
TmtMatrix	preMultiplyBy(TmtMatrix b) premuliply this matrix by another matrix B
int	rowCount() access to the number of rows in the matrix
TmtMatrix	scalarMultiply(double d) returns the product of this matrix times a scalar
void	showMatrix() prints the matrix to stdout
void	svd() computes the singular value decomposition for this matrix, storing the results as a member of this matrix's class
Instances	toWekaInstances() converts this matrix into a set of Weka Instances for further processing.
TmtMatrix	transpose() creates a new matrix that is the transpose of this one
double	zSum() returns the sum of all elements of the matrix

Methods inherited from class java.lang.Object

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

Field Detail

theMatrix

protected DoubleMatrix2D theMatrix

mySVD

private SingularValueDecomposition mySVD

isSparse

private boolean isSparse

Constructor Detail

TmtMatrix

public TmtMatrix(int rows,
                 int cols,
                 boolean sparse)

constructor for a basic zero matrix (dense)

Parameters:

rows -

cols -

sparse -

TmtMatrix

public TmtMatrix(double[][] denseMatrix)

constructor for a dense matrix from a 2D array of doubles

Parameters:

denseMatrix -

TmtMatrix

public TmtMatrix(int rows,
                 int cols,
                 java.util.ArrayList rowInd,
                 java.util.ArrayList colInd,
                 java.util.ArrayList values)
          throws BadDimensionException

constructor for a sparse matrix from data stored in memory

Parameters:

rows - the number of rows

cols - the number of columns

rowInd - an arrayList of row indices of nonzeros

colInd - an arrayList of column indicies of nonzeros

values - an arrayList of the nonzero values

TmtMatrix

public TmtMatrix(java.lang.String fileName,
                 int rows,
                 int cols,
                 boolean sparse)

constructor for reading a matrix (sparse or dense) from disk when we know the number of rows and columns beforehand

Parameters:

fileName -

rows -

cols -

sparse -

TmtMatrix

public TmtMatrix(java.lang.String fileName,
                 boolean sparse)

constructor for reading a matrix (sparse or dense) from disk when we know the number of rows and columns beforehand

Parameters:

fileName -

rows -

cols -

sparse -

TmtMatrix

public TmtMatrix(DoubleMatrix2D newMatrix)

constructor to build a matrix from a colt-formatted version

Parameters:

newMatrix - a matrix from colt

TmtMatrix

public TmtMatrix(TmtMatrix m)

Method Detail

showMatrix

public void showMatrix()

prints the matrix to stdout

rowCount

public int rowCount()

access to the number of rows in the matrix

Returns:

the total number of rows in the matrix

columnCount

public int columnCount()

access to the number of columns in the matrix

Returns:

the total number of columns in the matrix

transpose

public TmtMatrix transpose()

creates a new matrix that is the transpose of this one

Returns:

a NEW matrix initialized as the transpose of this matrix

getColumn

public TmtVector getColumn(int i)

get a column from the matrix

Parameters:

i - the (0-based) index of the column to retrieve

Returns:

a TmtVector containing the column

getRow

public TmtVector getRow(int i)

get a row from the matrix

Parameters:

i - the (0-based) index of the row to retrieve

Returns:

a TmtVector containing the row

getValue

public double getValue(int row,
                       int column)

getColumns

public TmtMatrix getColumns(int start,
                            int end)

construct a new matrix containing a subset of the columns of this matrix

Parameters:

start - the (0-indexed) starting point of the subset

end - the end point

Returns:

a TmtMatrix containing the desired columns

getRows

public TmtMatrix getRows(int start,
                         int end)

construct a new matrix containing a subset of the columns of this matrix

Parameters:

start - the (0-indexed) starting point of the subset

end - the end point

Returns:

a TmtMatrix containing the desired columns

getColumnCentered

public TmtMatrix getColumnCentered()

returns column-centered version of the matrix

Returns:

version of this matrix centered around column means

getColumnNormalized

public TmtMatrix getColumnNormalized()

returns column-normalized (unit length) version of the matrix

Returns:

version of this matrix with columns normalized to unit length

zSum

public double zSum()

returns the sum of all elements of the matrix

Returns:

the sum in double format

getMatrix

public DoubleMatrix2D getMatrix()

addMatrix

public TmtMatrix addMatrix(TmtMatrix b)

returns the matrix sum of this matrix with another matrix

Parameters:

b - the matrix to add

Returns:

the matrix sum

scalarMultiply

public TmtMatrix scalarMultiply(double d)

returns the product of this matrix times a scalar

Parameters:

d - the scalar to multiply by

Returns:

the product matrix

preMultiplyBy

public TmtMatrix preMultiplyBy(TmtMatrix b)

premuliply this matrix by another matrix B

Parameters:

b - the matrix to premultiply by

Returns:

the matrix product B . thisMatrix

postMultiplyBy

public TmtMatrix postMultiplyBy(TmtMatrix b)

postmultiply this matrix by another matrix B

Parameters:

b - the matrix to postmultiply by

Returns:

the matrix product thisMatrix . B

getCorrelationMatrix

public TmtMatrix getCorrelationMatrix()

computes and returns the correlation matrix of this matrix

Returns:

the p X p correlation matrix of this n X p matrix

svd

public void svd()

computes the singular value decomposition for this matrix, storing the results as a member of this matrix's class

getLeftSingularVectors

public TmtMatrix getLeftSingularVectors()

returns a TmtMatrix containing the left singular vectors of this matrix

Returns:

the left singular vectors in TmtFormat

getRightSingularVectors

public TmtMatrix getRightSingularVectors()

returns a TmtMatrix containing the right singular vectors of this matrix

Returns:

the right singular vectors in TmtFormat

getSingularValues

public TmtMatrix getSingularValues()

returns a diagonal TmtMatrix containing the singular values of this matrix

Returns:

the singular values in diagonal TmtFormat

toWekaInstances

public Instances toWekaInstances()

converts this matrix into a set of Weka Instances for further processing. This version of the method assumes that the attribute names are unknown, assigning them and the dataset name dummy values.

Returns:

an object of type Instances, with instances on the rows and attributes on the columns