org.nd4j.linalg.api.ops

Class BaseAccumulation

java.lang.Object

org.nd4j.linalg.api.ops.BaseOp

org.nd4j.linalg.api.ops.BaseAccumulation

All Implemented Interfaces:

Accumulation, Op

Direct Known Subclasses:

BaseLossFunction, Bias, CosineSimilarity, Dot, EuclideanDistance, ManhattanDistance, Max, Min, Norm1, Norm2, NormMax, Prod, Sum, Variance

public abstract class BaseAccumulation
extends BaseOp
implements Accumulation

Base class for accumulation, initiates the initial entry with respect to the child class. Also contains baseline fields for the over all field with accumulation.

Author:

Adam Gibson

Field Summary

Fields

Modifier and Type	Field and Description
protected boolean	applyFinalTransform
protected Number	finalResult
protected IComplexNumber	finalResultComplex

Fields inherited from class org.nd4j.linalg.api.ops.BaseOp

extraArgs, n, numProcessed, passThrough, x, y, z

Constructor Summary

Constructors

Constructor and Description
BaseAccumulation()
BaseAccumulation(INDArray x)
BaseAccumulation(INDArray x, INDArray y)
BaseAccumulation(INDArray x, INDArray y, INDArray z, long n) Initialize with the given input, pairwise transform, result, and number of elements
BaseAccumulation(INDArray x, INDArray y, long n)

Method Summary

Modifier and Type	Method and Description
boolean	applyFinalTransform() Whether to apply the final transform or not
double	calculateFinalResult(double accum, long n) Calculate the final result.
float	calculateFinalResult(float accum, long n)
double	combineSubResults(double first, double second) Combine sub-results, when the Accumulation operation is split and executed in parallel. Sometimes this is the same as the update operation (i.e., in sum operation), but often it is not (for example, in L2 norm, sqrt(1/n * sum_i x_i^2).
float	combineSubResults(float first, float second)
IComplexNumber	combineSubResults(IComplexNumber first, IComplexNumber second)
Number	currentResult() Essentially identical to Accumulation.getFinalResult(), maintained to avoid breaking implementations created before Accumulation ops were refactored.
double	getAndSetFinalResult(double accum) Get and set the final result. The final result is also stored in the Accumulation op itself (and can be retrieved by getFinalResult()) In some Accumulation operations, a final operation must be done on the accumulated result.
float	getAndSetFinalResult(float accum)
IComplexNumber	getAndSetFinalResult(IComplexNumber accum) Complex version of getAndSetFinalResult().
Number	getFinalResult() Get the final result (may return null if getAndSetFinalResult has not been called, or for accumulation ops on complex arrays)
IComplexNumber	getFinalResultComplex()
void	init(INDArray x, INDArray y, INDArray z, long n) Initialize the operation based on the parameters
INDArray	noOp() Returns the no op version of the input Basically when a reduce can't happen (eg: sum(0) on a row vector) you have a no op state for a given reduction.
long	numProcessed() Number processed
double	op(double origin) Transform an individual element
double	op(double origin, double other) Pairwise op (applicable with an individual element in y)
float	op(float origin) Transform an individual element
float	op(float origin, float other) Pairwise op (applicable with an individual element in y)
IComplexNumber	op(IComplexNumber origin) Transform an individual element
IComplexNumber	op(IComplexNumber origin, double other) Pairwise op (applicable with an individual element in y)
IComplexNumber	op(IComplexNumber origin, float other) Pairwise op (applicable with an individual element in y)
IComplexNumber	op(IComplexNumber origin, IComplexNumber other) Pairwise op (applicable with an individual element in y)
void	setApplyFinalTransform(boolean applyFinalTransform) Setter for final transform
void	setFinalResult(Number number)
void	setFinalResultComplex(IComplexNumber number)
void	setZ(INDArray z) set z (the solution ndarray)
IComplexNumber	zeroComplex() Complex initial value
double	zeroDouble() Initial value (used to initialize the accumulation op)
float	zeroFloat() Initial value (used to initialize the accumulation op)

Methods inherited from class org.nd4j.linalg.api.ops.BaseOp

exec, exec, extraArgs, extraArgsBuff, extraArgsDataBuff, isExecSpecial, isPassThrough, n, setN, setX, setY, toString, x, y, z

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.nd4j.linalg.api.ops.Accumulation

update, update, update, update, update, update, update, update, update

Methods inherited from interface org.nd4j.linalg.api.ops.Op

exec, exec, extraArgs, extraArgsBuff, extraArgsDataBuff, isExecSpecial, isPassThrough, n, name, opForDimension, opForDimension, opNum, setN, setX, setY, x, y, z

Field Detail

finalResult

protected Number finalResult

finalResultComplex

protected IComplexNumber finalResultComplex

applyFinalTransform

protected boolean applyFinalTransform

Constructor Detail

BaseAccumulation

public BaseAccumulation()

BaseAccumulation

public BaseAccumulation(INDArray x,
                        INDArray y,
                        INDArray z,
                        long n)

Initialize with the given input, pairwise transform, result, and number of elements

Parameters:

x - the input

y - the pairwise transform

z - the result

n - the number of elements

BaseAccumulation

public BaseAccumulation(INDArray x,
                        INDArray y,
                        long n)

BaseAccumulation

public BaseAccumulation(INDArray x)

BaseAccumulation

public BaseAccumulation(INDArray x,
                        INDArray y)

Method Detail

noOp

public INDArray noOp()

Description copied from interface: Accumulation

Returns the no op version of the input Basically when a reduce can't happen (eg: sum(0) on a row vector) you have a no op state for a given reduction. For most accumulations, this should return x but certain transformations should return say: the absolute value

Specified by:

noOp in interface Accumulation

Returns:

the no op version of the input

applyFinalTransform

public boolean applyFinalTransform()

Description copied from interface: Accumulation

Whether to apply the final transform or not

Specified by:

applyFinalTransform in interface Accumulation

Returns:

setApplyFinalTransform

public void setApplyFinalTransform(boolean applyFinalTransform)

Description copied from interface: Accumulation

Setter for final transform

Specified by:

setApplyFinalTransform in interface Accumulation

op

public IComplexNumber op(IComplexNumber origin,
                         double other)

Description copied from interface: Op

Pairwise op (applicable with an individual element in y)

Specified by:

op in interface Accumulation

Specified by:

op in interface Op

Parameters:

origin - the origin number

other - the other number

Returns:

the transformed output

op

public IComplexNumber op(IComplexNumber origin,
                         float other)

Description copied from interface: Op

Pairwise op (applicable with an individual element in y)

Specified by:

op in interface Accumulation

Specified by:

op in interface Op

Parameters:

origin - the origin number

other - the other number

Returns:

the transformed output

op

public IComplexNumber op(IComplexNumber origin,
                         IComplexNumber other)

Description copied from interface: Op

Pairwise op (applicable with an individual element in y)

Specified by:

op in interface Accumulation

Specified by:

op in interface Op

Parameters:

origin - the origin number

other - the other number

Returns:

the transformed output

op

public float op(float origin,
                float other)

Description copied from interface: Op

Pairwise op (applicable with an individual element in y)

Specified by:

op in interface Op

Parameters:

origin - the origin number

other - the other number

Returns:

the transformed output

op

public double op(double origin,
                 double other)

Description copied from interface: Op

Pairwise op (applicable with an individual element in y)

Specified by:

op in interface Op

Parameters:

origin - the origin number

other - the other number

Returns:

the transformed output

op

public double op(double origin)

Description copied from interface: Op

Transform an individual element

Specified by:

op in interface Op

Parameters:

origin - the origin element

Returns:

the new element

op

public float op(float origin)

Description copied from interface: Op

Transform an individual element

Specified by:

op in interface Op

Parameters:

origin - the origin element

Returns:

the new element

op

public IComplexNumber op(IComplexNumber origin)

Description copied from interface: Op

Transform an individual element

Specified by:

op in interface Accumulation

Specified by:

op in interface Op

Parameters:

origin - the origin element

Returns:

the new element

zeroDouble

public double zeroDouble()

Description copied from interface: Accumulation

Initial value (used to initialize the accumulation op)

Specified by:

zeroDouble in interface Accumulation

Returns:

the initial value

zeroFloat

public float zeroFloat()

Description copied from interface: Accumulation

Initial value (used to initialize the accumulation op)

Specified by:

zeroFloat in interface Accumulation

zeroComplex

public IComplexNumber zeroComplex()

Description copied from interface: Accumulation

Complex initial value

Specified by:

zeroComplex in interface Accumulation

Returns:

the complex initial value

numProcessed

public long numProcessed()

Description copied from interface: Op

Number processed

Specified by:

numProcessed in interface Op

Overrides:

numProcessed in class BaseOp

Returns:

the number of elements accumulated

init

public void init(INDArray x,
                 INDArray y,
                 INDArray z,
                 long n)

Description copied from interface: Op

Initialize the operation based on the parameters

Specified by:

init in interface Op

Overrides:

init in class BaseOp

Parameters:

x - the input

y - the pairwise transform ndarray

z - the resulting ndarray

n - the number of elements

combineSubResults

public double combineSubResults(double first,
                                double second)

Description copied from interface: Accumulation

Combine sub-results, when the Accumulation operation is split and executed in parallel.
Sometimes this is the same as the update operation (i.e., in sum operation), but often it is not (for example, in L2 norm, sqrt(1/n * sum_i x_i^2).

Specified by:

combineSubResults in interface Accumulation

Parameters:

first - The first sub-value to be combined

second - The second sub-value to be combined

Returns:

The combined result

combineSubResults

public float combineSubResults(float first,
                               float second)

Specified by:

combineSubResults in interface Accumulation

See Also:

Accumulation.combineSubResults(double, double)

combineSubResults

public IComplexNumber combineSubResults(IComplexNumber first,
                                        IComplexNumber second)

Specified by:

combineSubResults in interface Accumulation

See Also:

Accumulation.combineSubResults(double, double)

getAndSetFinalResult

public double getAndSetFinalResult(double accum)

Description copied from interface: Accumulation

Get and set the final result.
The final result is also stored in the Accumulation op itself (and can be retrieved by getFinalResult()) In some Accumulation operations, a final operation must be done on the accumulated result. For example, division by n in a mean operation: the accumulated result is the sum, whereas the final result is the sum/n.

Specified by:

getAndSetFinalResult in interface Accumulation

Parameters:

accum - The accumulated result

Returns:

The final result

getAndSetFinalResult

public float getAndSetFinalResult(float accum)

Specified by:

getAndSetFinalResult in interface Accumulation

See Also:

Accumulation.getAndSetFinalResult(double)

getAndSetFinalResult

public IComplexNumber getAndSetFinalResult(IComplexNumber accum)

Description copied from interface: Accumulation

Complex version of getAndSetFinalResult().

Specified by:

getAndSetFinalResult in interface Accumulation

See Also:

Accumulation.getAndSetFinalResult(double)

calculateFinalResult

public double calculateFinalResult(double accum,
                                   long n)

Description copied from interface: Accumulation

Calculate the final result. unlike Accumulation.getAndSetFinalResult(double), the result is merely calculated and returned, not stored in the Accumulation op itself.

Specified by:

calculateFinalResult in interface Accumulation

Parameters:

accum - The accumulated result

n - the number of elements accumulated

Returns:

calculated final result

calculateFinalResult

public float calculateFinalResult(float accum,
                                  long n)

Specified by:

calculateFinalResult in interface Accumulation

See Also:

#calculateFinalResult(double, int)

currentResult

public Number currentResult()

Description copied from interface: Accumulation

Essentially identical to Accumulation.getFinalResult(), maintained to avoid breaking implementations created before Accumulation ops were refactored. May be removed in a future release.

Specified by:

currentResult in interface Accumulation

setFinalResult

public void setFinalResult(Number number)

Specified by:

setFinalResult in interface Accumulation

setFinalResultComplex

public void setFinalResultComplex(IComplexNumber number)

Specified by:

setFinalResultComplex in interface Accumulation

getFinalResult

public Number getFinalResult()

Description copied from interface: Accumulation

Get the final result (may return null if getAndSetFinalResult has not been called, or for accumulation ops on complex arrays)

Specified by:

getFinalResult in interface Accumulation

getFinalResultComplex

public IComplexNumber getFinalResultComplex()

Specified by:

getFinalResultComplex in interface Accumulation

setZ

public void setZ(INDArray z)

Description copied from interface: Op

set z (the solution ndarray)

Specified by:

setZ in interface Op

Overrides:

setZ in class BaseOp