Class Tensor2D

java.lang.Object

deepnetts.tensor.TensorBase

deepnetts.tensor.Tensor2D

All Implemented Interfaces:

Tensor, Serializable

public final class Tensor2D
extends TensorBase

A 2D tensor / matrix with specified number of rows and columns..

See Also:

• Serialized Form

Field Summary

Fields inherited from class TensorBase

numDimensions, shape, values, vectorizationImpl

Constructor Summary

Constructors

Constructor	Description
Tensor2D(float[][] vals)	Creates a 2D tensor / matrix filled with given 2d array.
Tensor2D(int rows, int cols)	Creates a tensor with specified number of rows and columns (matrix).
Tensor2D(int rows, int cols, float[] values)

Method Summary

Modifier and Type	Method	Description
final void	add(float value, int row, int col)	Adds specified value to matrix value at position x, y
final Tensor2D	add(Tensor1D toAdd)
Object	clone()
final int	cols()	Returns a number of columns in this matrix.
TensorBase	copy()
void	createColsCache()
void	createRowsAndColsCache()
void	createRowsCache()
static Tensor2D	fromFile(String filePath)
final float	get(int row, int col)	Returns a value at specified row, col position in this tensor.
float[]	getCol(int col, float[] colBuff)
float[]	getColsCache(int colIdx)
float[]	getRow(int rowIdx, float[] rowBuff)
float[]	getRowsCache(int rowIdx)
Tensor2D	getTransposed()
Tensor1D	matMul(Tensor1D vectorB, Tensor1D result)	Performs matrix multiplication with this 2d tensor and vector , and stores results in result tensor.
Tensor2D	matMul(Tensor2D matrixB, Tensor2D result)	Performs dot product operation on this tensor and matrib, and stores results in result tensor.
Tensor2D	matMulMT(Tensor2D matrixB, Tensor2D result)
Tensor1D	matMulVectorized(Tensor1D vector, Tensor1D result)
Tensor2D	matMulVectorized(Tensor2D matrixB, Tensor2D result)
void	outerProductAccumulate(Tensor2D inputs, Tensor2D result)
final int	rows()	Returns a number of rows in this matrix.
final void	set(float val, int row, int col)	Set value at specified [row, col] position in this tensor.
void	setCol(int colIdx, float[] colData)
void	setRow(int rowIdx, float[] rowData)
void	setThreadPool(DeepNettsThreadPool threadPool)
void	setValuesFromStringTransposed(String values)
Tensor1D	sqrSumByCol(Tensor1D sumSqr)
final void	sub(float val, int row, int col)
Tensor1D	sumByColInto(Tensor1D sumTensor)
void	toFile(String filePath)
Tensor2D	transposeInto(Tensor2D transposed)

Methods inherited from class TensorBase

add, add, addInto, apply, apply, containsNaN, copyFrom, copyFrom, copyFromGPU, copyToGPU, createAcceleratorBridge, div, div, div, equals, equals, fill, getAcceleratorBridge, getOrCreateAccBridge, getValues, hashCode, mean, multiply, multiply, multiplyElementWise, numDimensions, numElements, randomize, replace, setShape, setValues, setValuesFromString, shape, sqr, sqrt, std, sub, sub, sub, subInplace, sum, sumAbs, sumSqr, toString, toTsr, valuesAsString

Methods inherited from class Object

finalize, getClass, notify, notifyAll, wait, wait, wait

Constructor Details

Tensor2D

public Tensor2D(int rows,
 int cols)

Creates a tensor with specified number of rows and columns (matrix).

Parameters:

rows - number of rows

cols - number of columns

Tensor2D

public Tensor2D(int rows,
 int cols,
 float[] values)

Tensor2D

public Tensor2D(float[][] vals)

Creates a 2D tensor / matrix filled with given 2d array. First dimension corresponds to number of rows, while second dimension corresponds to number of columns.

Parameters:

vals - values to store into matrix

Method Details

cols

public final int cols()

Returns a number of columns in this matrix.

Returns:

rows

public final int rows()

Returns a number of rows in this matrix.

Returns:

get

public final float get(int row,
 int col)

Returns a value at specified row, col position in this tensor.

Parameters:

row -

col -

Returns:

value at [row, col]

set

public final void set(float val,
 int row,
 int col)

Set value at specified [row, col] position in this tensor.

Parameters:

val - value to set

row - tensor's row index

col - tensor's col index

setRow

public void setRow(int rowIdx,
 float[] rowData)

setCol

public void setCol(int colIdx,
 float[] colData)

createRowsCache

public void createRowsCache()

createColsCache

public void createColsCache()

createRowsAndColsCache

public void createRowsAndColsCache()

getRow

public float[] getRow(int rowIdx,
 float[] rowBuff)

getCol

public float[] getCol(int col,
 float[] colBuff)

add

public final void add(float value,
 int row,
 int col)

Adds specified value to matrix value at position x, y

Parameters:

value -

row -

col -

add

public final Tensor2D add(Tensor1D toAdd)

sub

public final void sub(float val,
 int row,
 int col)

copy

public TensorBase copy()

Overrides:

copy in class TensorBase

clone

public Object clone()

Overrides:

clone in class TensorBase

getTransposed

public Tensor2D getTransposed()

transposeInto

public Tensor2D transposeInto(Tensor2D transposed)

setValuesFromStringTransposed

public void setValuesFromStringTransposed(String values)

Overrides:

setValuesFromStringTransposed in class TensorBase

fromFile

public static Tensor2D fromFile(String filePath)
                         throws IOException

Throws:

IOException

toFile

public void toFile(String filePath)

sqrSumByCol

public Tensor1D sqrSumByCol(Tensor1D sumSqr)

matMul

public Tensor2D matMul(Tensor2D matrixB,
 Tensor2D result)

Performs dot product operation on this tensor and matrib, and stores results in result tensor.

Parameters:

matrixB -

result -

Returns:

matMul

public Tensor1D matMul(Tensor1D vectorB,
 Tensor1D result)

Performs matrix multiplication with this 2d tensor and vector , and stores results in result tensor.

Parameters:

vectorB -

result -

Returns:

getRowsCache

public float[] getRowsCache(int rowIdx)

getColsCache

public float[] getColsCache(int colIdx)

matMulVectorized

public Tensor1D matMulVectorized(Tensor1D vector,
 Tensor1D result)

matMulVectorized

public Tensor2D matMulVectorized(Tensor2D matrixB,
 Tensor2D result)

outerProductAccumulate

public void outerProductAccumulate(Tensor2D inputs,
 Tensor2D result)

matMulMT

public Tensor2D matMulMT(Tensor2D matrixB,
 Tensor2D result)

setThreadPool

public void setThreadPool(DeepNettsThreadPool threadPool)

sumByColInto

public Tensor1D sumByColInto(Tensor1D sumTensor)