Parallel Colt 0.7.2

cern.colt.matrix.tfloat
Class FloatFactory2D

java.lang.Object
  extended by cern.colt.PersistentObject
      extended by cern.colt.matrix.tfloat.FloatFactory2D
All Implemented Interfaces:
Serializable, Cloneable

public class FloatFactory2D
extends PersistentObject

Factory for convenient construction of 2-d matrices holding float cells. Also provides convenient methods to compose (concatenate) and decompose (split) matrices from/to constituent blocks.

 

Construction Use idioms like FloatFactory2D.dense.make(4,4) to construct dense matrices, FloatFactory2D.sparse.make(4,4) to construct sparse matrices.
Construction with initial values Use other make methods to construct matrices with given initial values.
Appending rows and columns Use methods appendColumns, appendRows and repeat to append rows and columns.
General block matrices Use methods compose and decompose to work with general block matrices.
Diagonal matrices Use methods diagonal(vector), diagonal(matrix) and identity to work with diagonal matrices.
Diagonal block matrices Use method composeDiagonal to work with diagonal block matrices.
Random Use methods random and sample to construct random matrices.

 

If the factory is used frequently it might be useful to streamline the notation. For example by aliasing:

  FloatFactory2D F = FloatFactory2D.dense;
  F.make(4,4);
  F.descending(10,20);
  F.random(4,4);
  ...
 

Version:
1.0, 09/24/99
Author:
wolfgang.hoschek@cern.ch
See Also:
Serialized Form

Field Summary
static FloatFactory2D dense
          A factory producing dense matrices.
static FloatFactory2D rowCompressed
          A factory producing sparse row compressed matrices.
static FloatFactory2D sparse
          A factory producing sparse hash matrices.
 
Fields inherited from class cern.colt.PersistentObject
serialVersionUID
 
Method Summary
 FloatMatrix2D appendColumn(FloatMatrix2D A, FloatMatrix1D b)
           
 FloatMatrix2D appendColumns(FloatMatrix2D A, FloatMatrix2D B)
          C = A||B; Constructs a new matrix which is the column-wise concatenation of two other matrices.
 FloatMatrix2D appendRow(FloatMatrix2D A, FloatMatrix1D b)
           
 FloatMatrix2D appendRows(FloatMatrix2D A, FloatMatrix2D B)
          C = A||B; Constructs a new matrix which is the row-wise concatenation of two other matrices.
 FloatMatrix2D ascending(int rows, int columns)
          Constructs a matrix with cells having ascending values.
 FloatMatrix2D compose(FloatMatrix2D[][] parts)
          Constructs a block matrix made from the given parts.
 FloatMatrix2D composeBidiagonal(FloatMatrix2D A, FloatMatrix2D B)
           
 FloatMatrix2D composeDiagonal(FloatMatrix2D A, FloatMatrix2D B)
          Constructs a diagonal block matrix from the given parts (the direct sum of two matrices).
 FloatMatrix2D composeDiagonal(FloatMatrix2D A, FloatMatrix2D B, FloatMatrix2D C)
          Constructs a diagonal block matrix from the given parts.
 void decompose(FloatMatrix2D[][] parts, FloatMatrix2D matrix)
          Splits a block matrix into its constituent blocks; Copies blocks of a matrix into the given parts.
 void demo1()
          Demonstrates usage of this class.
 void demo2()
          Demonstrates usage of this class.
 FloatMatrix2D descending(int rows, int columns)
          Constructs a matrix with cells having descending values.
 FloatMatrix2D diagonal(float[] vector)
          Constructs a new diagonal matrix whose diagonal elements are the elements of vector.
 FloatMatrix2D diagonal(FloatMatrix1D vector)
          Constructs a new diagonal matrix whose diagonal elements are the elements of vector.
 FloatMatrix1D diagonal(FloatMatrix2D A)
          Constructs a new vector consisting of the diagonal elements of A .
 FloatMatrix2D identity(int rowsAndColumns)
          Constructs an identity matrix (having ones on the diagonal and zeros elsewhere).
 FloatMatrix2D make(float[][] values)
          Constructs a matrix with the given cell values.
 FloatMatrix2D make(float[] values, int rows)
          Construct a matrix from a one-dimensional column-major packed array, ala Fortran.
 FloatMatrix2D make(int rows, int columns)
          Constructs a matrix with the given shape, each cell initialized with zero.
 FloatMatrix2D make(int rows, int columns, float initialValue)
          Constructs a matrix with the given shape, each cell initialized with the given value.
 FloatMatrix2D random(int rows, int columns)
          Constructs a matrix with uniformly distributed values in (0,1) (exclusive).
 FloatMatrix2D repeat(FloatMatrix2D A, int rowRepeat, int columnRepeat)
          C = A||A||..||A; Constructs a new matrix which is duplicated both along the row and column dimension.
 FloatMatrix2D reshape(FloatMatrix1D a, int rows, int columns)
           
 FloatMatrix2D sample(FloatMatrix2D matrix, float value, float nonZeroFraction)
          Modifies the given matrix to be a randomly sampled matrix.
 FloatMatrix2D sample(int rows, int columns, float value, float nonZeroFraction)
          Constructs a randomly sampled matrix with the given shape.
 
Methods inherited from class cern.colt.PersistentObject
clone
 
Methods inherited from class java.lang.Object
equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Field Detail

dense

public static final FloatFactory2D dense
A factory producing dense matrices.


sparse

public static final FloatFactory2D sparse
A factory producing sparse hash matrices.


rowCompressed

public static final FloatFactory2D rowCompressed
A factory producing sparse row compressed matrices.

Method Detail

appendColumns

public FloatMatrix2D appendColumns(FloatMatrix2D A,
                                   FloatMatrix2D B)
C = A||B; Constructs a new matrix which is the column-wise concatenation of two other matrices.
         0 1 2
         3 4 5
         appendColumns
         6 7
         8 9
         -->
         0 1 2 6 7 
         3 4 5 8 9
 
 


appendColumn

public FloatMatrix2D appendColumn(FloatMatrix2D A,
                                  FloatMatrix1D b)

appendRows

public FloatMatrix2D appendRows(FloatMatrix2D A,
                                FloatMatrix2D B)
C = A||B; Constructs a new matrix which is the row-wise concatenation of two other matrices.
         0 1 
         2 3 
         4 5
         appendRows
         6 7
         8 9
         -->
         0 1 
         2 3 
         4 5
         6 7
         8 9
 
 


appendRow

public FloatMatrix2D appendRow(FloatMatrix2D A,
                               FloatMatrix1D b)

ascending

public FloatMatrix2D ascending(int rows,
                               int columns)
Constructs a matrix with cells having ascending values. For debugging purposes. Example:
         0 1 2 
         3 4 5
 
 


reshape

public FloatMatrix2D reshape(FloatMatrix1D a,
                             int rows,
                             int columns)

compose

public FloatMatrix2D compose(FloatMatrix2D[][] parts)
Constructs a block matrix made from the given parts. The inverse to method decompose(FloatMatrix2D[][], FloatMatrix2D).

All matrices of a given column within parts must have the same number of columns. All matrices of a given row within parts must have the same number of rows. Otherwise an IllegalArgumentException is thrown. Note that nulls within parts[row,col] are an exception to this rule: they are ignored. Cells are copied. Example:

Code Result
 FloatMatrix2D[][] parts1 = { { null, make(2, 2, 1), null }, { make(4, 4, 2), null, make(4, 3, 3) }, { null, make(2, 2, 4), null } };
 System.out.println(compose(parts1));
 
8 x 9 matrix
0 0 0 0 1 1 0 0 0
0 0 0 0 1 1 0 0 0
2 2 2 2 0 0 3 3 3
2 2 2 2 0 0 3 3 3
2 2 2 2 0 0 3 3 3
2 2 2 2 0 0 3 3 3
0 0 0 0 4 4 0 0 0
0 0 0 0 4 4 0 0 0
 FloatMatrix2D[][] parts3 = { { identity(3), null, }, { null, identity(3).viewColumnFlip() }, { identity(3).viewRowFlip(), null } };
 System.out.println("\n" + make(parts3));
 
9 x 6 matrix
1 0 0 0 0 0
0 1 0 0 0 0
0 0 1 0 0 0
0 0 0 0 0 1
0 0 0 0 1 0
0 0 0 1 0 0
0 0 1 0 0 0
0 1 0 0 0 0
1 0 0 0 0 0
 FloatMatrix2D A = ascending(2, 2);
 FloatMatrix2D B = descending(2, 2);
 FloatMatrix2D _ = null;
 
 FloatMatrix2D[][] parts4 = { { A, _, A, _ }, { _, A, _, B } };
 System.out.println("\n" + make(parts4));
 
4 x 8 matrix
1 2 0 0 1 2 0 0
3 4 0 0 3 4 0 0
0 0 1 2 0 0 3 2
0 0 3 4 0 0 1 0
 FloatMatrix2D[][] parts2 = { { null, make(2, 2, 1), null }, { make(4, 4, 2), null, make(4, 3, 3) }, { null, make(2, 3, 4), null } };
 System.out.println("\n" + Factory2D.make(parts2));
 
IllegalArgumentException
A[0,1].cols != A[2,1].cols
(2 != 3)

Throws:
IllegalArgumentException - subject to the conditions outlined above.

composeDiagonal

public FloatMatrix2D composeDiagonal(FloatMatrix2D A,
                                     FloatMatrix2D B)
Constructs a diagonal block matrix from the given parts (the direct sum of two matrices). That is the concatenation
         A 0
         0 B
 
 
(The direct sum has A.rows()+B.rows() rows and A.columns()+B.columns() columns). Cells are copied.

Returns:
a new matrix which is the direct sum.

composeDiagonal

public FloatMatrix2D composeDiagonal(FloatMatrix2D A,
                                     FloatMatrix2D B,
                                     FloatMatrix2D C)
Constructs a diagonal block matrix from the given parts. The concatenation has the form
         A 0 0
         0 B 0
         0 0 C
 
 
from the given parts. Cells are copied.


composeBidiagonal

public FloatMatrix2D composeBidiagonal(FloatMatrix2D A,
                                       FloatMatrix2D B)

decompose

public void decompose(FloatMatrix2D[][] parts,
                      FloatMatrix2D matrix)
Splits a block matrix into its constituent blocks; Copies blocks of a matrix into the given parts. The inverse to method compose(FloatMatrix2D[][]).

All matrices of a given column within parts must have the same number of columns. All matrices of a given row within parts must have the same number of rows. Otherwise an IllegalArgumentException is thrown. Note that nulls within parts[row,col] are an exception to this rule: they are ignored. Cells are copied. Example:

Code matrix --> parts
         FloatMatrix2D matrix = ... ;
         FloatMatrix2D _ = null;
         FloatMatrix2D A,B,C,D;
         A = make(2,2); B = make (4,4);
         C = make(4,3); D = make (2,2);
         FloatMatrix2D[][] parts = 
         {
            { _, A, _ },
            { B, _, C },
            { _, D, _ }
         };
         decompose(parts,matrix);
         System.out.println("\nA = "+A);
         System.out.println("\nB = "+B);
         System.out.println("\nC = "+C);
         System.out.println("\nD = "+D);
 
 
8 x 9 matrix
9 9 9 9 1 1 9 9 9
9 9 9 9 1 1 9 9 9
2 2 2 2 9 9 3 3 3
2 2 2 2 9 9 3 3 3
2 2 2 2 9 9 3 3 3
2 2 2 2 9 9 3 3 3
9 9 9 9 4 4 9 9 9
9 9 9 9 4 4 9 9 9

A = 2 x 2 matrix
1 1
1 1

B = 4 x 4 matrix
2 2 2 2
2 2 2 2
2 2 2 2
2 2 2 2

C = 4 x 3 matrix
3 3 3
3 3 3
3 3 3
3 3 3

D = 2 x 2 matrix
4 4
4 4

Throws:
IllegalArgumentException - subject to the conditions outlined above.

demo1

public void demo1()
Demonstrates usage of this class.


demo2

public void demo2()
Demonstrates usage of this class.


descending

public FloatMatrix2D descending(int rows,
                                int columns)
Constructs a matrix with cells having descending values. For debugging purposes. Example:
         5 4 3 
         2 1 0
 
 


diagonal

public FloatMatrix2D diagonal(FloatMatrix1D vector)
Constructs a new diagonal matrix whose diagonal elements are the elements of vector. Cells values are copied. The new matrix is not a view. Example:
         5 4 3 -->
         5 0 0
         0 4 0
         0 0 3
 
 

Returns:
a new matrix.

diagonal

public FloatMatrix2D diagonal(float[] vector)
Constructs a new diagonal matrix whose diagonal elements are the elements of vector. Cells values are copied. The new matrix is not a view. Example:
         5 4 3 -->
         5 0 0
         0 4 0
         0 0 3
 
 

Returns:
a new matrix.

diagonal

public FloatMatrix1D diagonal(FloatMatrix2D A)
Constructs a new vector consisting of the diagonal elements of A . Cells values are copied. The new vector is not a view. Example:
         5 0 0 9
         0 4 0 9
         0 0 3 9
         --> 5 4 3
 
 

Parameters:
A - the matrix, need not be square.
Returns:
a new vector.

identity

public FloatMatrix2D identity(int rowsAndColumns)
Constructs an identity matrix (having ones on the diagonal and zeros elsewhere).


make

public FloatMatrix2D make(float[][] values)
Constructs a matrix with the given cell values. values is required to have the form values[row][column] and have exactly the same number of columns in every row.

The values are copied. So subsequent changes in values are not reflected in the matrix, and vice-versa.

Parameters:
values - The values to be filled into the new matrix.
Throws:
IllegalArgumentException - if for any 1 <= row < values.length: values[row].length != values[row-1].length .

make

public FloatMatrix2D make(float[] values,
                          int rows)
Construct a matrix from a one-dimensional column-major packed array, ala Fortran. Has the form matrix.get(row,column) == values[row + column*rows]. The values are copied.

Parameters:
values - One-dimensional array of floats, packed by columns (ala Fortran).
rows - the number of rows.
Throws:
IllegalArgumentException - values.length must be a multiple of rows .

make

public FloatMatrix2D make(int rows,
                          int columns)
Constructs a matrix with the given shape, each cell initialized with zero.


make

public FloatMatrix2D make(int rows,
                          int columns,
                          float initialValue)
Constructs a matrix with the given shape, each cell initialized with the given value.


random

public FloatMatrix2D random(int rows,
                            int columns)
Constructs a matrix with uniformly distributed values in (0,1) (exclusive).


repeat

public FloatMatrix2D repeat(FloatMatrix2D A,
                            int rowRepeat,
                            int columnRepeat)
C = A||A||..||A; Constructs a new matrix which is duplicated both along the row and column dimension. Example:
         0 1
         2 3
         repeat(2,3) -->
         0 1 0 1 0 1
         2 3 2 3 2 3
         0 1 0 1 0 1
         2 3 2 3 2 3
 
 


sample

public FloatMatrix2D sample(int rows,
                            int columns,
                            float value,
                            float nonZeroFraction)
Constructs a randomly sampled matrix with the given shape. Randomly picks exactly Math.round(rows*columns*nonZeroFraction) cells and initializes them to value, all the rest will be initialized to zero. Note that this is not the same as setting each cell with probability nonZeroFraction to value. Note: The random seed is a constant.

Throws:
IllegalArgumentException - if nonZeroFraction < 0 || nonZeroFraction > 1.
See Also:
FloatRandomSampler

sample

public FloatMatrix2D sample(FloatMatrix2D matrix,
                            float value,
                            float nonZeroFraction)
Modifies the given matrix to be a randomly sampled matrix. Randomly picks exactly Math.round(rows*columns*nonZeroFraction) cells and initializes them to value, all the rest will be initialized to zero. Note that this is not the same as setting each cell with probability nonZeroFraction to value. Note: The random seed is a constant.

Throws:
IllegalArgumentException - if nonZeroFraction < 0 || nonZeroFraction > 1.
See Also:
FloatRandomSampler

Parallel Colt 0.7.2

Jump to the Parallel Colt Homepage