/*
* JScience - Java(TM) Tools and Libraries for the Advancement of Sciences.
* Copyright (C) 2006 - JScience (http://jscience.org/)
* All rights reserved.
*
* Permission to use, copy, modify, and distribute this software is
* freely granted, provided that this notice is preserved.
*/
package org.jscience.mathematics.vector;
import java.util.Iterator;
import java.util.List;
import javolution.context.ConcurrentContext;
import javolution.context.ObjectFactory;
import javolution.lang.MathLib;
import javolution.util.FastTable;
import org.jscience.mathematics.number.Float64;
/**
* <p> This class represents an optimized {@link Matrix matrix} implementation
* for {@link Float64 64 bits floating-point} numbers.</p>
*
* <p> Instances of this class can be created from {@link Float64Vector},
* either as rows or columns if the matrix is transposed. For example:[code]
* Float64Vector<Rational> column0 = Float64Vector.valueOf(...);
* Float64Vector<Rational> column1 = Float64Vector.valueOf(...);
* Float64Matrix<Rational> M = Float64Matrix.valueOf(column0, column1).transpose();
* [/code]</p>
*
* @author <a href="mailto:jean-marie@dautelle.com">Jean-Marie Dautelle</a>
* @version 3.3, January 2, 2007
*/
public final class Float64Matrix extends Matrix<Float64> {
/**
* Holds the number of columns n.
*/
int _n;;
/**
* Indicates if this matrix is transposed (the rows are then the columns).
*/
boolean _transposed;
/**
* Holds this matrix rows (or columns when transposed).
*/
final FastTable<Float64Vector> _rows = new FastTable<Float64Vector>();
/**
* Returns a dense matrix from a 2-dimensional array of <code>double</code>
* values. The first dimension being the row and the second being the
* column.
*
* @param values the array of <code>double</code> values.
* @return the matrix having the specified elements.
* @throws DimensionException if rows have different length.
* @see Float64Vector
*/
public static Float64Matrix valueOf(double[][] values) {
int m = values.length;
int n = values[0].length;
Float64Matrix M = Float64Matrix.newInstance(n, false);
for (int i = 0; i < m; i++) {
Float64Vector row = Float64Vector.valueOf(values[i]);
if (row.getDimension() != n)
throw new DimensionException();
M._rows.add(row);
}
return M;
}
/**
* Returns a complex matrix holding the specified row vectors
* (column vectors if {@link #transpose transposed}).
*
* @param rows the row vectors.
* @return the matrix having the specified rows.
* @throws DimensionException if the rows do not have the same dimension.
*/
public static Float64Matrix valueOf(Float64Vector... rows) {
final int n = rows[0].getDimension();
Float64Matrix M = Float64Matrix.newInstance(n, false);
for (int i = 0, m = rows.length; i < m; i++) {
Float64Vector rowi = rows[i];
if (rowi.getDimension() != n)
throw new DimensionException(
"All vectors must have the same dimension.");
M._rows.add(rowi);
}
return M;
}
/**
* Returns a complex matrix holding the row vectors from the specified
* collection (column vectors if {@link #transpose transposed}).
*
* @param rows the list of row vectors.
* @return the matrix having the specified rows.
* @throws DimensionException if the rows do not have the same dimension.
*/
public static Float64Matrix valueOf(List<Float64Vector> rows) {
final int n = rows.get(0).getDimension();
Float64Matrix M = Float64Matrix.newInstance(n, false);
Iterator<Float64Vector> iterator = rows.iterator();
for (int i = 0, m = rows.size(); i < m; i++) {
Float64Vector rowi = iterator.next();
if (rowi.getDimension() != n)
throw new DimensionException(
"All vectors must have the same dimension.");
M._rows.add(rowi);
}
return M;
}
/**
* Returns a complex matrix equivalent to the specified matrix.
*
* @param that the matrix to convert.
* @return <code>that</code> or a complex matrix holding the same elements
* as the specified matrix.
*/
public static Float64Matrix valueOf(Matrix<Float64> that) {
if (that instanceof Float64Matrix)
return (Float64Matrix) that;
int n = that.getNumberOfColumns();
int m = that.getNumberOfRows();
Float64Matrix M = Float64Matrix.newInstance(n, false);
for (int i = 0; i < m; i++) {
Float64Vector rowi = Float64Vector.valueOf(that.getRow(i));
M._rows.add(rowi);
}
return M;
}
@Override
public int getNumberOfRows() {
return _transposed ? _n : _rows.size();
}
@Override
public int getNumberOfColumns() {
return _transposed ? _rows.size() : _n;
}
@Override
public Float64 get(int i, int j) {
return _transposed ? _rows.get(j).get(i) : _rows.get(i).get(j);
}
@Override
public Float64Vector getRow(int i) {
if (!_transposed)
return _rows.get(i);
// Else transposed.
int n = _rows.size();
int m = _n;
if ((i < 0) || (i >= m))
throw new DimensionException();
Float64Vector V = Float64Vector.newInstance(n);
for (int j = 0; j < n; j++) {
V.set(j, _rows.get(j).get(i).doubleValue());
}
return V;
}
@Override
public Float64Vector getColumn(int j) {
if (_transposed)
return _rows.get(j);
int m = _rows.size();
if ((j < 0) || (j >= _n))
throw new DimensionException();
Float64Vector V = Float64Vector.newInstance(m);
for (int i = 0; i < m; i++) {
V.set(i, _rows.get(i).get(j).doubleValue());
}
return V;
}
@Override
public Float64Vector getDiagonal() {
int m = this.getNumberOfRows();
int n = this.getNumberOfColumns();
int dimension = MathLib.min(m, n);
Float64Vector V = Float64Vector.newInstance(dimension);
for (int i = 0; i < dimension; i++) {
V.set(i, this.get(i, i).doubleValue());
}
return V;
}
@Override
public Float64Matrix opposite() {
Float64Matrix M = Float64Matrix.newInstance(_n, _transposed);
for (int i = 0, p = _rows.size(); i < p; i++) {
M._rows.add(_rows.get(i).opposite());
}
return M;
}
@Override
public Float64Matrix plus(Matrix<Float64> that) {
if (this.getNumberOfRows() != that.getNumberOfRows())
throw new DimensionException();
Float64Matrix M = Float64Matrix.newInstance(_n, _transposed);
for (int i = 0, p = _rows.size(); i < p; i++) {
M._rows.add(_rows.get(i).plus(
_transposed ? that.getColumn(i) : that.getRow(i)));
}
return M;
}
@Override
public Float64Matrix minus(Matrix<Float64> that) { // Returns more specialized type.
return this.plus(that.opposite());
}
@Override
public Float64Matrix times(Float64 k) {
Float64Matrix M = Float64Matrix.newInstance(_n, _transposed);
for (int i = 0, p = _rows.size(); i < p; i++) {
M._rows.add(_rows.get(i).times(k));
}
return M;
}
@Override
public Float64Vector times(Vector<Float64> v) {
if (v.getDimension() != this.getNumberOfColumns())
throw new DimensionException();
final int m = this.getNumberOfRows();
Float64Vector V = Float64Vector.newInstance(m);
for (int i = 0; i < m; i++) {
V.set(i, this.getRow(i).times(v).doubleValue());
}
return V;
}
@Override
public Float64Matrix times(Matrix<Float64> that) {
final int n = this.getNumberOfColumns();
final int m = this.getNumberOfRows();
final int p = that.getNumberOfColumns();
if (that.getNumberOfRows() != n)
throw new DimensionException();
// Creates a mxp matrix in transposed form (p columns vectors of size m)
Float64Matrix M = Float64Matrix.newInstance(m, true); // Transposed.
M._rows.setSize(p);
Multiply multiply = Multiply.valueOf(this, that, 0, p, M._rows);
multiply.run();
Multiply.recycle(multiply);
return M;
}
// Logic to multiply two matrices.
private static class Multiply implements Runnable {
private static final ObjectFactory<Multiply> FACTORY = new ObjectFactory<Multiply>() {
@Override
protected Multiply create() {
return new Multiply();
}
};
private Float64Matrix _left;
private Matrix<Float64> _right;
private int _rightColumnStart;
private int _rightColumnEnd;
private FastTable<Float64Vector> _columnsResult;
static Multiply valueOf(Float64Matrix left, Matrix<Float64> right,
int rightColumnStart, int rightColumnEnd,
FastTable<Float64Vector> columnsResult) {
Multiply multiply = Multiply.FACTORY.object();
multiply._left = left;
multiply._right = right;
multiply._rightColumnStart = rightColumnStart;
multiply._rightColumnEnd = rightColumnEnd;
multiply._columnsResult = columnsResult;
return multiply;
}
static void recycle(Multiply multiply) {
multiply._left = null;
multiply._right = null;
multiply._columnsResult = null;
Multiply.FACTORY.recycle(multiply);
}
public void run() {
if (_rightColumnEnd - _rightColumnStart < 32) { // Direct calculation.
FastTable<Float64Vector> rows = _left.getRows();
final int m = rows.size();
for (int j = _rightColumnStart; j < _rightColumnEnd; j++) {
Vector<Float64> thatColj = _right.getColumn(j);
Float64Vector column = Float64Vector.newInstance(m);
_columnsResult.set(j, column);
for (int i = 0; i < m; i++) {
column.set(i, rows.get(i).times(thatColj)
.doubleValue());
}
}
} else { // Concurrent/Recursive calculation.
int halfIndex = (_rightColumnStart + _rightColumnEnd) >> 1;
Multiply firstHalf = Multiply.valueOf(_left, _right,
_rightColumnStart, halfIndex, _columnsResult);
Multiply secondHalf = Multiply.valueOf(_left, _right,
halfIndex, _rightColumnEnd, _columnsResult);
ConcurrentContext.enter();
try {
ConcurrentContext.execute(firstHalf);
ConcurrentContext.execute(secondHalf);
} finally {
ConcurrentContext.exit();
}
Multiply.recycle(firstHalf);
Multiply.recycle(secondHalf);
}
}
}
private FastTable<Float64Vector> getRows() {
if (!_transposed)
return _rows;
FastTable<Float64Vector> rows = FastTable.newInstance();
for (int i = 0; i < _n; i++) {
rows.add(this.getRow(i));
}
return rows;
}
@Override
public Float64Matrix inverse() {
if (!isSquare())
throw new DimensionException("Matrix not square");
return Float64Matrix.valueOf(LUDecomposition.valueOf(this).inverse());
}
@Override
public Float64 determinant() {
return LUDecomposition.valueOf(this).determinant();
}
@Override
public Float64Matrix transpose() {
Float64Matrix M = Float64Matrix.newInstance(_n, !_transposed);
M._rows.addAll(this._rows);
return M;
}
@Override
public Float64 cofactor(int i, int j) {
if (_transposed) {
int k = i;
i = j;
j = k; // Swaps i,j
}
int m = _rows.size();
Float64Matrix M = Float64Matrix.newInstance(m - 1, _transposed);
for (int k1 = 0; k1 < m; k1++) {
if (k1 == i)
continue;
Float64Vector row = _rows.get(k1);
Float64Vector V = Float64Vector.newInstance(_n - 1);
M._rows.add(V);
for (int k2 = 0, k = 0; k2 < _n; k2++) {
if (k2 == j)
continue;
V.set(k++, row.get(k2).doubleValue());
}
}
return M.determinant();
}
@Override
public Float64Matrix adjoint() {
Float64Matrix M = Float64Matrix.newInstance(_n, _transposed);
int m = _rows.size();
for (int i = 0; i < m; i++) {
Float64Vector row = Float64Vector.newInstance(_n);
M._rows.add(row);
for (int j = 0; j < _n; j++) {
Float64 cofactor = _transposed ? cofactor(j, i)
: cofactor(i, j);
row.set(j, ((i + j) % 2 == 0) ? cofactor.doubleValue()
: cofactor.opposite().doubleValue());
}
}
return M.transpose();
}
@Override
public Float64Matrix tensor(Matrix<Float64> that) {
return Float64Matrix.valueOf(DenseMatrix.valueOf(this).tensor(that));
}
@Override
public Float64Vector vectorization() {
return Float64Vector.valueOf(DenseMatrix.valueOf(this).vectorization());
}
@Override
public Float64Matrix copy() {
Float64Matrix M = Float64Matrix.newInstance(_n, _transposed);
for (Float64Vector row : _rows) {
M._rows.add(row.copy());
}
return M;
}
///////////////////////
// Factory creation. //
///////////////////////
static Float64Matrix newInstance(int n, boolean transposed) {
Float64Matrix M = FACTORY.object();
M._n = n;
M._transposed = transposed;
return M;
}
private static ObjectFactory<Float64Matrix> FACTORY = new ObjectFactory<Float64Matrix>() {
@Override
protected Float64Matrix create() {
return new Float64Matrix();
}
@Override
protected void cleanup(Float64Matrix matrix) {
matrix._rows.reset();
}
};
private Float64Matrix() {
}
private static final long serialVersionUID = 1L;
}
|