org.shared.array.jni

Class NativeArrayKernel

java.lang.Object

org.shared.array.jni.NativeArrayKernel

All Implemented Interfaces:

ArrayKernel, Service

public class NativeArrayKernel
extends Object
implements ArrayKernel

A native implementation of ArrayKernel.

Field Summary

Fields inherited from interface org.shared.array.kernel.ArrayKernel

C_TO_R_ABS, C_TO_R_IM, C_TO_R_RE, CA_PROD, CA_SUM, CE_ADD, CE_DIV, CE_MUL, CE_SUB, CU_ADD, CU_CONJ, CU_COS, CU_EXP, CU_FILL, CU_MUL, CU_RND, CU_SHUFFLE, CU_SIN, I_TO_R, IE_ADD, IE_MAX, IE_MIN, IE_MUL, IE_SUB, IU_ADD, IU_FILL, IU_MUL, IU_SHUFFLE, R_TO_C_IM, R_TO_C_RE, RA_ENT, RA_MAX, RA_MIN, RA_PROD, RA_SUM, RA_VAR, RD_PROD, RD_SUM, RE_ADD, RE_DIV, RE_MAX, RE_MIN, RE_MUL, RE_SUB, RI_GZERO, RI_LZERO, RI_MAX, RI_MIN, RI_SORT, RI_ZERO, RR_MAX, RR_MIN, RR_PROD, RR_SUM, RR_VAR, RU_ABS, RU_ADD, RU_ATAN, RU_COS, RU_EXP, RU_FILL, RU_INV, RU_LOG, RU_MUL, RU_POW, RU_RND, RU_SHUFFLE, RU_SIN, RU_SQR, RU_SQRT

Constructor Summary

Constructors

Constructor and Description
NativeArrayKernel() Default constructor.

Method Summary

Methods

Modifier and Type	Method and Description
double[]	caOp(int type, double[] srcV) Performs a complex accumulator operation.
void	convert(int type, Object srcV, boolean isSrcComplex, Object dstV, boolean isDstComplex) Performs a conversion operation.
void	cuOp(int type, double aRe, double aIm, double[] srcV) Applies a complex unary operation.
void	derandomize() Seeds the underlying source of randomness with a constant.
void	diag(double[] srcV, double[] dstV, int size, boolean complex) Gets the diagonal of a Matrix.
void	eigs(double[] srcV, double[] vecV, double[] valV, int size) Computes the eigenvectors and eigenvalues of a Matrix.
void	eOp(int type, Object lhsV, Object rhsV, Object dstV, boolean complex) Applies a binary operation.
int[]	find(int[] srcV, int[] srcD, int[] srcS, int[] logical) Extracts the valid indices along a dimension anchored at the given logical index.
<V> SparseArrayState<V>	insertSparse(V oldV, int[] oldD, int[] oldS, int[] oldDo, int[] oldI, V newV, int[] newLi) Inserts elements into a sparse array.
void	invert(double[] srcV, double[] dstV, int size) Computes the inverse of a Matrix.
void	iuOp(int type, int a, int[] srcV) Applies an integer unary operation.
void	map(int[] bounds, Object srcV, int[] srcD, int[] srcS, Object dstV, int[] dstD, int[] dstS) Performs a mapping operation.
void	mul(double[] lhsV, double[] rhsV, int lr, int rc, double[] dstV, boolean complex) Multiplies two Matrixs.
void	randomize() Seeds the underlying source of randomness with the current time.
double	raOp(int type, double[] srcV) Performs a real accumulator operation.
void	rdOp(int type, double[] srcV, int[] srcD, int[] srcS, double[] dstV, int... opDims) Performs a real dimension operation.
void	riOp(int type, double[] srcV, int[] srcD, int[] srcS, int[] dstV, int dim) Performs a real index operation.
void	rrOp(int type, double[] srcV, int[] srcD, int[] srcS, double[] dstV, int[] dstD, int[] dstS, int... opDims) Performs a real reduce operation.
void	ruOp(int type, double a, double[] srcV) Applies a real unary operation.
void	slice(int[] slices, Object srcV, int[] srcD, int[] srcS, Object dstV, int[] dstD, int[] dstS) Performs a slicing operation.
<V> SparseArrayState<V>	sliceSparse(int[] slices, V srcV, int[] srcD, int[] srcS, int[] srcDo, int[] srcI, int[] srcIo, int[] srcIi, V dstV, int[] dstD, int[] dstS, int[] dstDo, int[] dstI, int[] dstIo, int[] dstIi) Slices one sparse array into another.
void	svd(double[] srcV, int srcStrideRow, int srcStrideCol, double[] uV, double[] sV, double[] vV, int nRows, int nCols) Computes the singular value decomposition of a Matrix.

Methods inherited from class java.lang.Object

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

Constructor Detail

NativeArrayKernel

public NativeArrayKernel()

Default constructor. Checks the validity of native bindings.

Method Detail

randomize

public final void randomize()

Description copied from interface: ArrayKernel

Seeds the underlying source of randomness with the current time.

Specified by:

randomize in interface ArrayKernel

derandomize

public final void derandomize()

Description copied from interface: ArrayKernel

Seeds the underlying source of randomness with a constant.

Specified by:

derandomize in interface ArrayKernel

map

public final void map(int[] bounds,
       Object srcV,
       int[] srcD,
       int[] srcS,
       Object dstV,
       int[] dstD,
       int[] dstS)

Description copied from interface: ArrayKernel

Performs a mapping operation.

Specified by:

map in interface ArrayKernel

Parameters:

bounds - the mapping bounds.

srcV - the source values.

srcD - the source dimensions.

srcS - the source strides.

dstV - the destination values.

dstD - the destination dimensions.

dstS - the destination strides.

slice

public final void slice(int[] slices,
         Object srcV,
         int[] srcD,
         int[] srcS,
         Object dstV,
         int[] dstD,
         int[] dstS)

Description copied from interface: ArrayKernel

Performs a slicing operation.

Specified by:

slice in interface ArrayKernel

Parameters:

slices - the slicing specification.

srcV - the source values.

srcD - the source dimensions.

srcS - the source strides.

dstV - the destination values.

dstD - the destination dimensions.

dstS - the destination strides.

rrOp

public final void rrOp(int type,
        double[] srcV,
        int[] srcD,
        int[] srcS,
        double[] dstV,
        int[] dstD,
        int[] dstS,
        int... opDims)

Description copied from interface: ArrayKernel

Performs a real reduce operation.

Specified by:

rrOp in interface ArrayKernel

Parameters:

type - the operation type.

srcV - the source values.

srcD - the source dimensions.

srcS - the source strides.

dstV - the destination values.

dstD - the destination dimensions.

dstS - the destination strides.

opDims - the dimensions of interest.

riOp

public final void riOp(int type,
        double[] srcV,
        int[] srcD,
        int[] srcS,
        int[] dstV,
        int dim)

Description copied from interface: ArrayKernel

Performs a real index operation.

Specified by:

riOp in interface ArrayKernel

Parameters:

type - the operation type.

srcV - the source values.

srcD - the source dimensions.

srcS - the source strides.

dstV - the destination values.

dim - the dimension of interest.

rdOp

public final void rdOp(int type,
        double[] srcV,
        int[] srcD,
        int[] srcS,
        double[] dstV,
        int... opDims)

Description copied from interface: ArrayKernel

Performs a real dimension operation.

Specified by:

rdOp in interface ArrayKernel

Parameters:

type - the operation type.

srcV - the source values.

srcD - the source dimensions.

srcS - the source strides.

dstV - the destination values.

opDims - the dimensions of interest.

raOp

public final double raOp(int type,
          double[] srcV)

Description copied from interface: ArrayKernel

Performs a real accumulator operation.

Specified by:

raOp in interface ArrayKernel

Parameters:

type - the operation type.

srcV - the array.

Returns:

the accumulated result.

caOp

public final double[] caOp(int type,
            double[] srcV)

Description copied from interface: ArrayKernel

Performs a complex accumulator operation.

Specified by:

caOp in interface ArrayKernel

Parameters:

type - the operation type.

srcV - the array.

Returns:

the accumulated result.

ruOp

public final void ruOp(int type,
        double a,
        double[] srcV)

Description copied from interface: ArrayKernel

Applies a real unary operation.

Specified by:

ruOp in interface ArrayKernel

Parameters:

type - the operation type.

a - the argument, if any.

srcV - the array.

cuOp

public final void cuOp(int type,
        double aRe,
        double aIm,
        double[] srcV)

Description copied from interface: ArrayKernel

Applies a complex unary operation.

Specified by:

cuOp in interface ArrayKernel

Parameters:

type - the operation type.

aRe - the real part of the argument, if any.

aIm - the imaginary part of the argument, if any.

srcV - the array.

iuOp

public final void iuOp(int type,
        int a,
        int[] srcV)

Description copied from interface: ArrayKernel

Applies an integer unary operation.

Specified by:

iuOp in interface ArrayKernel

Parameters:

type - the operation type.

a - the argument, if any.

srcV - the array.

eOp

public final void eOp(int type,
       Object lhsV,
       Object rhsV,
       Object dstV,
       boolean complex)

Description copied from interface: ArrayKernel

Applies a binary operation.

Specified by:

eOp in interface ArrayKernel

Parameters:

type - the operation type.

lhsV - the left hand side values.

rhsV - the right hand side values.

dstV - the destination values.

complex - whether the operation is complex-valued.

convert

public final void convert(int type,
           Object srcV,
           boolean isSrcComplex,
           Object dstV,
           boolean isDstComplex)

Description copied from interface: ArrayKernel

Performs a conversion operation.

Specified by:

convert in interface ArrayKernel

Parameters:

type - the operation type.

srcV - the source values.

isSrcComplex - whether the source is complex-valued.

dstV - the destination values.

isDstComplex - whether the destination is complex-valued.

mul

public final void mul(double[] lhsV,
       double[] rhsV,
       int lr,
       int rc,
       double[] dstV,
       boolean complex)

Description copied from interface: ArrayKernel

Multiplies two Matrixs. They are assumed to have storage order Array.IndexingOrder.FAR.

Specified by:

mul in interface ArrayKernel

Parameters:

lhsV - the left hand side values.

rhsV - the right hand side values.

lr - the row count of the result.

rc - the column count of the result.

dstV - the destination values.

complex - whether the operation is complex-valued.

diag

public final void diag(double[] srcV,
        double[] dstV,
        int size,
        boolean complex)

Description copied from interface: ArrayKernel

Gets the diagonal of a Matrix.

Specified by:

diag in interface ArrayKernel

Parameters:

srcV - source values.

dstV - destination values.

size - the matrix size.

complex - whether the operation is complex-valued.

svd

public final void svd(double[] srcV,
       int srcStrideRow,
       int srcStrideCol,
       double[] uV,
       double[] sV,
       double[] vV,
       int nRows,
       int nCols)

Description copied from interface: ArrayKernel

Computes the singular value decomposition of a Matrix.

Specified by:

svd in interface ArrayKernel

Parameters:

srcV - the source values.

srcStrideRow - the source row stride.

srcStrideCol - the source column stride.

uV - the input vectors.

sV - the gain controls.

vV - the output vectors.

nRows - the number of rows.

nCols - the number of columns.

eigs

public final void eigs(double[] srcV,
        double[] vecV,
        double[] valV,
        int size)

Description copied from interface: ArrayKernel

Computes the eigenvectors and eigenvalues of a Matrix.

Specified by:

eigs in interface ArrayKernel

Parameters:

srcV - the source values.

vecV - the eigenvectors.

valV - the eigenvalues.

size - the matrix size.

invert

public final void invert(double[] srcV,
          double[] dstV,
          int size)

Description copied from interface: ArrayKernel

Computes the inverse of a Matrix.

Specified by:

invert in interface ArrayKernel

Parameters:

srcV - the source values.

dstV - the destination values.

size - the matrix size.

find

public final int[] find(int[] srcV,
         int[] srcD,
         int[] srcS,
         int[] logical)

Description copied from interface: ArrayKernel

Extracts the valid indices along a dimension anchored at the given logical index.

Specified by:

find in interface ArrayKernel

Parameters:

srcV - the source values.

srcD - the source dimensions.

srcS - the source strides.

logical - the logical index.

Returns:

the valid indices.

insertSparse

public final <V> SparseArrayState<V> insertSparse(V oldV,
                                   int[] oldD,
                                   int[] oldS,
                                   int[] oldDo,
                                   int[] oldI,
                                   V newV,
                                   int[] newLi)

Description copied from interface: ArrayKernel

Inserts elements into a sparse array.

Specified by:

insertSparse in interface ArrayKernel

Type Parameters:

V - the storage array type.

Parameters:

oldV - the old values.

oldD - the old dimensions.

oldS - the old strides.

oldDo - the old dimension offsets.

oldI - the old physical indices. Invariant: Sorted in ascending order, and does not contain duplicates.

newV - the new values.

newLi - the new physical indices, which need not be sorted in ascending order.

Returns:

the SparseArrayState.

sliceSparse

public final <V> SparseArrayState<V> sliceSparse(int[] slices,
                                  V srcV,
                                  int[] srcD,
                                  int[] srcS,
                                  int[] srcDo,
                                  int[] srcI,
                                  int[] srcIo,
                                  int[] srcIi,
                                  V dstV,
                                  int[] dstD,
                                  int[] dstS,
                                  int[] dstDo,
                                  int[] dstI,
                                  int[] dstIo,
                                  int[] dstIi)

Description copied from interface: ArrayKernel

Slices one sparse array into another.

Specified by:

sliceSparse in interface ArrayKernel

Type Parameters:

V - the storage array type.

Parameters:

slices - the slicing specification.

srcV - the source values.

srcD - the source dimensions.

srcS - the source strides.

srcDo - the source dimension offsets.

srcI - the source physical indices. Invariant: Sorted in ascending order, and does not contain duplicates.

srcIo - the source indirection offsets.

srcIi - the source indirections.

dstV - the destination values.

dstD - the destination dimensions.

dstS - the destination strides.

dstDo - the destination dimension offsets.

dstI - the destination physical indices. Invariant: Sorted in ascending order, and does not contain duplicates.

dstIo - the destination indirection offsets.

dstIi - the destination indirections.

Returns:

the SparseArrayState.