org.ejml.alg.block

Class BlockVectorOps

java.lang.Object

org.ejml.alg.block.BlockVectorOps

public class BlockVectorOps
extends Object

Math operations for inner vectors (row and column) inside of block matrices:

scale: b_i = α*a_i
div: _i = a_i/α
add: c_i = α*a_i + βB_i
dot: c = sum a_i*b_i

All submatrices must be block aligned. All offsets and end indexes are relative to the beginning of each submatrix.

Author:

Peter Abeles

Constructor Summary

Constructors

Constructor and Description
BlockVectorOps()

Method Summary

Methods

Modifier and Type	Method and Description
static void	add_row(int blockLength, D1Submatrix64F A, int rowA, double alpha, D1Submatrix64F B, int rowB, double beta, D1Submatrix64F C, int rowC, int offset, int end) Row vector add: add: ci = α*ai + βBi where 'a', 'b', and 'c' are row vectors within the row block vectors of A, B, and C respectively.
static void	div_row(int blockLength, D1Submatrix64F A, int rowA, double alpha, D1Submatrix64F B, int rowB, int offset, int end) Row vector divide: div: bi = ai/α where 'a' and 'b' are row vectors within the row block vector A and B.
static double	dot_row_col(int blockLength, D1Submatrix64F A, int rowA, D1Submatrix64F B, int colB, int offset, int end) vector dot/inner product from one row vector and one column vector: dot: c = sum ai*bi where 'a' is a row vector 'b' is a column vectors within the row block vector A and B, and 'c' is a scalar.
static double	dot_row(int blockLength, D1Submatrix64F A, int rowA, D1Submatrix64F B, int rowB, int offset, int end) Row vector dot/inner product: dot: c = sum ai*bi where 'a' and 'b' are row vectors within the row block vector A and B, and 'c' is a scalar.
static void	scale_row(int blockLength, D1Submatrix64F A, int rowA, double alpha, D1Submatrix64F B, int rowB, int offset, int end) Row vector scale: scale: bi = α*ai where 'a' and 'b' are row vectors within the row block vector A and B.

Methods inherited from class java.lang.Object

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

Constructor Detail

BlockVectorOps

public BlockVectorOps()

Method Detail

scale_row

public static void scale_row(int blockLength,
             D1Submatrix64F A,
             int rowA,
             double alpha,
             D1Submatrix64F B,
             int rowB,
             int offset,
             int end)

Row vector scale:
scale: b_i = α*a_i
where 'a' and 'b' are row vectors within the row block vector A and B.

Parameters:

A - submatrix. Not modified.

rowA - which row in A the vector is contained in.

alpha - scale factor.

B - submatrix that the results are written to. Modified.

offset - Index at which the vectors start at.

end - Index at which the vectors end at.

div_row

public static void div_row(int blockLength,
           D1Submatrix64F A,
           int rowA,
           double alpha,
           D1Submatrix64F B,
           int rowB,
           int offset,
           int end)

Row vector divide:
div: b_i = a_i/α
where 'a' and 'b' are row vectors within the row block vector A and B.

Parameters:

A - submatrix. Not modified.

rowA - which row in A the vector is contained in.

alpha - scale factor.

B - submatrix that the results are written to. Modified.

offset - Index at which the vectors start at.

end - Index at which the vectors end at.

add_row

public static void add_row(int blockLength,
           D1Submatrix64F A,
           int rowA,
           double alpha,
           D1Submatrix64F B,
           int rowB,
           double beta,
           D1Submatrix64F C,
           int rowC,
           int offset,
           int end)

Row vector add:
add: c_i = α*a_i + βB_i
where 'a', 'b', and 'c' are row vectors within the row block vectors of A, B, and C respectively.

Parameters:

blockLength - Length of each inner matrix block.

A - submatrix. Not modified.

rowA - which row in A the vector is contained in.

alpha - scale factor of A

B - submatrix. Not modified.

rowB - which row in B the vector is contained in.

beta - scale factor of B

C - submatrix where the results are written to. Modified.

rowC - which row in C is the vector contained.

offset - Index at which the vectors start at.

end - Index at which the vectors end at.

dot_row

public static double dot_row(int blockLength,
             D1Submatrix64F A,
             int rowA,
             D1Submatrix64F B,
             int rowB,
             int offset,
             int end)

Row vector dot/inner product:
dot: c = sum a_i*b_i
where 'a' and 'b' are row vectors within the row block vector A and B, and 'c' is a scalar.

Parameters:

A - submatrix. Not modified.

rowA - which row in A the vector is contained in.

B - submatrix. Not modified.

rowB - which row in B the vector is contained in.

offset - Index at which the vectors start at.

end - Index at which the vectors end at.

Returns:

Results of the dot product.

dot_row_col

public static double dot_row_col(int blockLength,
                 D1Submatrix64F A,
                 int rowA,
                 D1Submatrix64F B,
                 int colB,
                 int offset,
                 int end)

vector dot/inner product from one row vector and one column vector:
dot: c = sum a_i*b_i
where 'a' is a row vector 'b' is a column vectors within the row block vector A and B, and 'c' is a scalar.

Parameters:

A - block row vector. Not modified.

rowA - which row in A the vector is contained in.

B - block column vector. Not modified.

colB - which column in B is the vector contained in.

offset - Index at which the vectors start at.

end - Index at which the vectors end at.

Returns:

Results of the dot product.