Package de.labathome.trustconstr.matrix

Class DenseMatrix

java.lang.Object

de.labathome.trustconstr.matrix.Matrix

de.labathome.trustconstr.matrix.DenseMatrix

public final class DenseMatrix
extends Matrix

Dense matrix backed by a column-major double[] buffer of length rows * cols. Element (i, j) is stored at data[i + j * rows].

Column-major is the layout LAPACK and BLAS expect, so data() can be passed directly to dgesv / dgeqrf / dgesvd / dpotrf / dsyr / dsyr2 with no copy. The hot kernel mtimes(Matrix) dispatches to BLAS dgemm when both operands are DenseMatrix; everything else falls back to a triple loop.

Subclass of Matrix so existing call sites that accept a generic Matrix keep working unchanged. As the migration progresses, callers narrow their declared types to DenseMatrix where the dense specialisation matters (BLAS dgemm fast path or data() pass-through to LAPACK).

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static final class	DenseMatrix.Factory	Backwards-compatible nested factory class.

Constructor Summary

Constructors

Constructor	Description
DenseMatrix(int rows, int cols)	Allocate a fresh rows x cols zero-filled DenseMatrix.

Method Summary

Modifier and Type	Method	Description
DenseMatrix	abs()
DenseMatrix	absInPlace()	In-place absolute value.
int	cols()
static DenseMatrix	column(double... v)
DenseMatrix	copy()
static DenseMatrix	copyFromMatrix(Matrix m)	Densify any Matrix: if m is already a DenseMatrix, return a copy(); otherwise materialise via getAsDouble(long, long).
double[]	data()	The underlying column-major buffer.
DenseMatrix	divide(double alpha)
DenseMatrix	divide(Matrix elementwise)
static DenseMatrix	eye(int n)
static DenseMatrix	fromColumnMajor(int rows, int cols, double[] data)	Wrap an already-column-major buffer without copying.
static DenseMatrix	fromRows(double[][] rows)	Repack a row-major double[][] into a fresh column-major DenseMatrix.
double	get(int i, int j)
double	getAsDouble(long row, long col)
long	getColumnCount()
long	getRowCount()
DenseMatrix	minus(Matrix other)
DenseMatrix	mtimes(Matrix other)
double	norm2()
double	normInf()
DenseMatrix	plus(Matrix other)
static DenseMatrix	row(double... v)
int	rows()
DenseMatrix	scaleInPlace(double alpha)	In-place scaling.
DenseMatrix	selectColumns(int... colsToKeep)	Pick an arbitrary subset of columns into a fresh DenseMatrix.
DenseMatrix	selectRows(int... rowsToKeep)	Pick an arbitrary subset of rows into a fresh DenseMatrix.
void	set(int i, int j, double v)
void	setAsDouble(double v, long row, long col)
DenseMatrix	subMatrix(int r0, int r1, int c0, int c1)	Inclusive submatrix slice rows[r0..r1] x cols[c0..c1].
DenseMatrix	times(double alpha)
DenseMatrix	times(Matrix elementwise)
double[]	toColumnArray()
double[][]	toDoubleArray()
double[]	toRowMajor()
String	toString()
DenseMatrix	transpose()
static DenseMatrix	zeros(int rows, int cols)

Methods inherited from class de.labathome.trustconstr.matrix.Matrix

doubleValue, equalsContent, fill, getSize, isColumnVector, isRowVector, isSparse, mapAdd, minus, normF, plus, rank, selectColumns, solve, subMatrix

Methods inherited from class java.lang.Object

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

Constructor Details

DenseMatrix

public DenseMatrix(int rows,
 int cols)

Allocate a fresh rows x cols zero-filled DenseMatrix.

Parameters:

rows - number of rows (must be non-negative)

cols - number of columns (must be non-negative)

Method Details

rows

public int rows()

Returns:

number of rows (primitive int accessor)

cols

public int cols()

Returns:

number of columns (primitive int accessor)

data

public double[] data()

The underlying column-major buffer. Mutable: callers may overwrite entries directly (used by BLAS rank updates and LAPACK in-place ops).

Returns:

the live column-major buffer of length rows * cols; entry (i, j) at index i + j * rows

get

public double get(int i,
 int j)

Parameters:

i - 0-based row index

j - 0-based column index

Returns:

entry at (i, j)

set

public void set(int i,
 int j,
 double v)

Parameters:

i - 0-based row index

j - 0-based column index

v - value to store at (i, j)

getRowCount

public long getRowCount()

Specified by:

getRowCount in class Matrix

Returns:

number of rows

getColumnCount

public long getColumnCount()

Specified by:

getColumnCount in class Matrix

Returns:

number of columns

getAsDouble

public double getAsDouble(long row,
 long col)

Specified by:

getAsDouble in class Matrix

Parameters:

row - 0-based row index

col - 0-based column index

Returns:

entry at (row, col)

setAsDouble

public void setAsDouble(double v,
 long row,
 long col)

Specified by:

setAsDouble in class Matrix

Parameters:

v - value to store

row - 0-based row index

col - 0-based column index

copy

public DenseMatrix copy()

Specified by:

copy in class Matrix

Returns:

a deep copy of this matrix

toDoubleArray

public double[][] toDoubleArray()

Overrides:

toDoubleArray in class Matrix

Returns:

a fresh row-major double[][] copy of the contents

toColumnArray

public double[] toColumnArray()

Overrides:

toColumnArray in class Matrix

Returns:

contents of an n x 1 column vector as a fresh double[n]; throws if the matrix is not a column vector

toRowMajor

public double[] toRowMajor()

Returns:

contents repacked as a fresh row-major flat double[rows*cols] (entry (i, j) at index i*cols + j)

mtimes

public DenseMatrix mtimes(Matrix other)

Specified by:

mtimes in class Matrix

Parameters:

other - right operand

Returns:

matrix product this * other

plus

public DenseMatrix plus(Matrix other)

Specified by:

plus in class Matrix

Parameters:

other - right operand

Returns:

element-wise sum this + other

minus

public DenseMatrix minus(Matrix other)

Specified by:

minus in class Matrix

Parameters:

other - right operand

Returns:

element-wise difference this - other

times

public DenseMatrix times(double alpha)

Specified by:

times in class Matrix

Parameters:

alpha - scalar

Returns:

alpha * this

times

public DenseMatrix times(Matrix elementwise)

Specified by:

times in class Matrix

Parameters:

elementwise - same-shape factor

Returns:

Hadamard product this .* elementwise

divide

public DenseMatrix divide(double alpha)

Specified by:

divide in class Matrix

Parameters:

alpha - scalar

Returns:

this / alpha

divide

public DenseMatrix divide(Matrix elementwise)

Specified by:

divide in class Matrix

Parameters:

elementwise - same-shape divisor

Returns:

Hadamard division this ./ elementwise

transpose

public DenseMatrix transpose()

Specified by:

transpose in class Matrix

Returns:

the transpose of this matrix

abs

public DenseMatrix abs()

Specified by:

abs in class Matrix

Returns:

element-wise absolute value as a fresh matrix

absInPlace

public DenseMatrix absInPlace()

In-place absolute value.

Overrides:

absInPlace in class Matrix

Returns:

this after replacing every entry by its absolute value

scaleInPlace

public DenseMatrix scaleInPlace(double alpha)

In-place scaling.

Parameters:

alpha - scalar multiplier

Returns:

this after multiplying every entry by alpha

norm2

public double norm2()

Specified by:

norm2 in class Matrix

Returns:

Frobenius norm

normInf

public double normInf()

Specified by:

normInf in class Matrix

Returns:

infinity-norm (largest absolute entry)

subMatrix

public DenseMatrix subMatrix(int r0,
 int r1,
 int c0,
 int c1)

Inclusive submatrix slice rows[r0..r1] x cols[c0..c1]. Always returns a fresh DenseMatrix; no view aliasing.

Parameters:

r0 - first row (inclusive)

r1 - last row (inclusive)

c0 - first column (inclusive)

c1 - last column (inclusive)

Returns:

(r1-r0+1) x (c1-c0+1) slice

selectColumns

public DenseMatrix selectColumns(int... colsToKeep)

Pick an arbitrary subset of columns into a fresh DenseMatrix.

Parameters:

colsToKeep - column indices to keep, in the desired output order

Returns:

rows x colsToKeep.length matrix

selectRows

public DenseMatrix selectRows(int... rowsToKeep)

Pick an arbitrary subset of rows into a fresh DenseMatrix.

Parameters:

rowsToKeep - row indices to keep, in the desired output order

Returns:

rowsToKeep.length x cols matrix

toString

public String toString()

Overrides:

toString in class Object

zeros

public static DenseMatrix zeros(int rows,
 int cols)

Parameters:

rows - number of rows (must be non-negative)

cols - number of columns (must be non-negative)

Returns:

a fresh rows x cols matrix filled with zeros

eye

public static DenseMatrix eye(int n)

Parameters:

n - dimension (must be non-negative)

Returns:

the n x n identity matrix

column

public static DenseMatrix column(double... v)

Parameters:

v - entries of the resulting column vector (copied)

Returns:

an n x 1 column vector

row

public static DenseMatrix row(double... v)

Parameters:

v - entries of the resulting row vector (copied)

Returns:

a 1 x n row vector

fromRows

public static DenseMatrix fromRows(double[][] rows)

Repack a row-major double[][] into a fresh column-major DenseMatrix.

Parameters:

rows - row-major source (rows[i][j] = entry at (i,j))

Returns:

a fresh DenseMatrix with the same shape and values

fromColumnMajor

public static DenseMatrix fromColumnMajor(int rows,
 int cols,
 double[] data)

Wrap an already-column-major buffer without copying. The caller surrenders ownership of data: subsequent mutations through this matrix will be visible to anyone else holding the array.

Parameters:

rows - number of rows

cols - number of columns

data - column-major buffer of length rows * cols; the entry at (i, j) is data[i + j * rows]

Returns:

a DenseMatrix that aliases data

copyFromMatrix

public static DenseMatrix copyFromMatrix(Matrix m)

Densify any Matrix: if m is already a DenseMatrix, return a copy(); otherwise materialise via getAsDouble(long, long).

Parameters:

m - source matrix

Returns:

a fresh dense copy