Package org.apache.sysds.runtime.compress.lib

Class CLALibLeftMultBy

java.lang.Object

org.apache.sysds.runtime.compress.lib.CLALibLeftMultBy

public final class CLALibLeftMultBy
extends Object

Method Summary

Modifier and Type	Method	Description
static MatrixBlock	leftMultByMatrix(CompressedMatrixBlock right, MatrixBlock left, MatrixBlock ret, int k)	Left multiplication with two CompressedMatrixBlock following the equation: ret = left %*% right
static MatrixBlock	leftMultByMatrixTransposed(CompressedMatrixBlock right, CompressedMatrixBlock left, MatrixBlock ret, int k)	Left multiplication with two CompressedMatrixBlock following the equation: ret = t(left) %*% right
static MatrixBlock	leftMultByMatrixTransposed(CompressedMatrixBlock right, MatrixBlock left, MatrixBlock ret, int k)	Left multiplication with a CompressedMatrixBlock on the right following the equation: ret = t(left) %*% right
static double[]	rowSum(MatrixBlock mb, int rl, int ru, int cl, int cu)

Methods inherited from class java.lang.Object

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

Method Detail

leftMultByMatrixTransposed

public static MatrixBlock leftMultByMatrixTransposed(CompressedMatrixBlock right,
                                                     MatrixBlock left,
                                                     MatrixBlock ret,
                                                     int k)

Left multiplication with a CompressedMatrixBlock on the right following the equation: ret = t(left) %*% right

Parameters:

right - A CompressedMatrixBlock on the right side of the multiplication.

left - A not transposed MatrixBlock.

ret - The result output matrix, this allocation of the object can be used if appropriate, otherwise a new matrix Block is allocated to be returned. This argument can also be null.

k - The number of threads allowed to be used

Returns:

The result of the matrix multiplication

leftMultByMatrixTransposed

public static MatrixBlock leftMultByMatrixTransposed(CompressedMatrixBlock right,
                                                     CompressedMatrixBlock left,
                                                     MatrixBlock ret,
                                                     int k)

Left multiplication with two CompressedMatrixBlock following the equation: ret = t(left) %*% right

Parameters:

right - A CompressedMatrixBlock on the right side of the multiplication.

left - A not transposed CompressedMatrixBlock, but logically inside the function it is considered transposed.

ret - The result output matrix, this allocation of the object can be used if appropriate, otherwise a new matrix Block is allocated to be returned. This argument can also be null.

k - The number of threads allowed to be used

Returns:

The result of the matrix multiplication

leftMultByMatrix

public static MatrixBlock leftMultByMatrix(CompressedMatrixBlock right,
                                           MatrixBlock left,
                                           MatrixBlock ret,
                                           int k)

Left multiplication with two CompressedMatrixBlock following the equation: ret = left %*% right

Parameters:

right - A CompressedMatrixBlock on the right side of the multiplication.

left - A MatrixBlock on the left side of the equation

ret - The result output matrix, this allocation of the object can be used if appropriate, otherwise a new matrix Block is allocated to be returned. This argument can also be null.

k - The number of threads allowed to be used

Returns:

The result of the matrix multiplication

rowSum

public static double[] rowSum(MatrixBlock mb,
                              int rl,
                              int ru,
                              int cl,
                              int cu)