cern.colt.matrix.tfloat.algo

Interface FloatBlas

All Known Implementing Classes:

SmpFloatBlas

public interface FloatBlas

Subset of the BLAS (Basic Linear Algebra System); High quality "building block" routines for performing basic vector and matrix operations. Because the BLAS are efficient, portable, and widely available, they're commonly used in the development of high quality linear algebra software.

Mostly for compatibility with legacy notations. Most operations actually just delegate to the appropriate methods directly defined on matrices and vectors.

This class implements the BLAS functions for operations on matrices from the matrix package. It follows the spirit of the Draft Proposal for Java BLAS Interface, by Roldan Pozo of the National Institute of Standards and Technology. Interface definitions are also identical to the Ninja interface. Because the matrix package supports sections, the interface is actually simpler.

Currently, the following operations are supported:

1. BLAS Level 1: Vector-Vector operations
• ddot : dot product of two vectors
• daxpy : scalar times a vector plus a vector
• drotg : construct a Givens plane rotation
• drot : apply a plane rotation
• dcopy : copy vector X into vector Y
• dswap : interchange vectors X and Y
• dnrm2 : Euclidean norm of a vector
• dasum : sum of absolute values of vector components
• dscal : scale a vector by a scalar
• idamax: index of element with maximum absolute value
2. 2.BLAS Level 2: Matrix-Vector operations
• dgemv : matrix-vector multiply with general matrix
• dger : rank-1 update on general matrix
• dsymv : matrix-vector multiply with symmetric matrix
• dtrmv : matrix-vector multiply with triangular matrix
3. 3.BLAS Level 3: Matrix-Matrix operations
   • dgemm : matrix-matrix multiply with general matrices

Version:

0.9, 16/04/2000

Author:

wolfgang.hoschek@cern.ch

Method Summary

Methods

Modifier and Type	Method and Description
void	assign(FloatMatrix2D A, FloatFunction function) Assigns the result of a function to each cell; x[row,col] = function(x[row,col]).
void	assign(FloatMatrix2D x, FloatMatrix2D y, FloatFloatFunction function) Assigns the result of a function to each cell; x[row,col] = function(x[row,col],y[row,col]).
float	dasum(FloatMatrix1D x) Returns the sum of absolute values; |x[0]| + |x[1]| + ...
void	daxpy(float alpha, FloatMatrix1D x, FloatMatrix1D y) Combined vector scaling; y = y + alpha*x.
void	daxpy(float alpha, FloatMatrix2D A, FloatMatrix2D B) Combined matrix scaling; B = B + alpha*A.
void	dcopy(FloatMatrix1D x, FloatMatrix1D y) Vector assignment (copying); y = x.
void	dcopy(FloatMatrix2D A, FloatMatrix2D B) Matrix assignment (copying); B = A.
float	ddot(FloatMatrix1D x, FloatMatrix1D y) Returns the dot product of two vectors x and y, which is Sum(x[i]*y[i]).
void	dgemm(boolean transposeA, boolean transposeB, float alpha, FloatMatrix2D A, FloatMatrix2D B, float beta, FloatMatrix2D C) Generalized linear algebraic matrix-matrix multiply; C = alpha*A*B + beta*C.
void	dgemv(boolean transposeA, float alpha, FloatMatrix2D A, FloatMatrix1D x, float beta, FloatMatrix1D y) Generalized linear algebraic matrix-vector multiply; y = alpha*A*x + beta*y.
void	dger(float alpha, FloatMatrix1D x, FloatMatrix1D y, FloatMatrix2D A) Performs a rank 1 update; A = A + alpha*x*y'.
float	dnrm2(FloatMatrix1D x) Return the 2-norm; sqrt(x[0]^2 + x[1]^2 + ...).
void	drot(FloatMatrix1D x, FloatMatrix1D y, float c, float s) Applies a givens plane rotation to (x,y); x = c*x + s*y; y = c*y - s*x.
void	drotg(float a, float b, float[] rotvec) Constructs a Givens plane rotation for (a,b).
void	dscal(float alpha, FloatMatrix1D x) Vector scaling; x = alpha*x.
void	dscal(float alpha, FloatMatrix2D A) Matrix scaling; A = alpha*A.
void	dswap(FloatMatrix1D x, FloatMatrix1D y) Swaps the elements of two vectors; y <==> x.
void	dswap(FloatMatrix2D x, FloatMatrix2D y) Swaps the elements of two matrices; B <==> A.
void	dsymv(boolean isUpperTriangular, float alpha, FloatMatrix2D A, FloatMatrix1D x, float beta, FloatMatrix1D y) Symmetric matrix-vector multiplication; y = alpha*A*x + beta*y.
void	dtrmv(boolean isUpperTriangular, boolean transposeA, boolean isUnitTriangular, FloatMatrix2D A, FloatMatrix1D x) Triangular matrix-vector multiplication; x = A*x or x = A'*x.
int	idamax(FloatMatrix1D x) Returns the index of largest absolute value; i such that |x[i]| == max(|x[0]|,|x[1]|,...)..

Method Detail

assign

void assign(FloatMatrix2D A,
          FloatFunction function)

Assigns the result of a function to each cell; x[row,col] = function(x[row,col]).

Parameters:

A - the matrix to modify.

function - a function object taking as argument the current cell's value.

See Also:

FloatFunctions

assign

void assign(FloatMatrix2D x,
          FloatMatrix2D y,
          FloatFloatFunction function)

Assigns the result of a function to each cell; x[row,col] = function(x[row,col],y[row,col]).

Parameters:

x - the matrix to modify.

y - the secondary matrix to operate on.

function - a function object taking as first argument the current cell's value of this, and as second argument the current cell's value of y,

Throws:

IllegalArgumentException - if x.columns() != y.columns() || x.rows() != y.rows()

See Also:

FloatFunctions

dasum

float dasum(FloatMatrix1D x)

Returns the sum of absolute values; |x[0]| + |x[1]| + ... . In fact equivalent to x.aggregate(cern.jet.math.Functions.plus, cern.jet.math.Functions.abs) .

Parameters:

x - the first vector.

daxpy

void daxpy(float alpha,
         FloatMatrix1D x,
         FloatMatrix1D y)

Combined vector scaling; y = y + alpha*x. In fact equivalent to y.assign(x,cern.jet.math.Functions.plusMult(alpha)).

Parameters:

alpha - a scale factor.

x - the first source vector.

y - the second source vector, this is also the vector where results are stored.

Throws:

IllegalArgumentException - x.size() != y.size()..

daxpy

void daxpy(float alpha,
         FloatMatrix2D A,
         FloatMatrix2D B)

Combined matrix scaling; B = B + alpha*A. In fact equivalent to B.assign(A,cern.jet.math.Functions.plusMult(alpha)).

Parameters:

alpha - a scale factor.

A - the first source matrix.

B - the second source matrix, this is also the matrix where results are stored.

Throws:

IllegalArgumentException - if A.columns() != B.columns() || A.rows() != B.rows().

dcopy

void dcopy(FloatMatrix1D x,
         FloatMatrix1D y)

Vector assignment (copying); y = x. In fact equivalent to y.assign(x).

Parameters:

x - the source vector.

y - the destination vector.

Throws:

IllegalArgumentException - x.size() != y.size().

dcopy

void dcopy(FloatMatrix2D A,
         FloatMatrix2D B)

Matrix assignment (copying); B = A. In fact equivalent to B.assign(A).

Parameters:

A - the source matrix.

B - the destination matrix.

Throws:

IllegalArgumentException - if A.columns() != B.columns() || A.rows() != B.rows().

ddot

float ddot(FloatMatrix1D x,
         FloatMatrix1D y)

Returns the dot product of two vectors x and y, which is Sum(x[i]*y[i]). In fact equivalent to x.zDotProduct(y).

Parameters:

x - the first vector.

y - the second vector.

Returns:

the sum of products.

Throws:

IllegalArgumentException - if x.size() != y.size().

dgemm

void dgemm(boolean transposeA,
         boolean transposeB,
         float alpha,
         FloatMatrix2D A,
         FloatMatrix2D B,
         float beta,
         FloatMatrix2D C)

Generalized linear algebraic matrix-matrix multiply; C = alpha*A*B + beta*C. In fact equivalent to A.zMult(B,C,alpha,beta,transposeA,transposeB). Note: Matrix shape conformance is checked after potential transpositions.

Parameters:

transposeA - set this flag to indicate that the multiplication shall be performed on A'.

transposeB - set this flag to indicate that the multiplication shall be performed on B'.

alpha - a scale factor.

A - the first source matrix.

B - the second source matrix.

beta - a scale factor.

C - the third source matrix, this is also the matrix where results are stored.

Throws:

IllegalArgumentException - if B.rows() != A.columns().

IllegalArgumentException - if C.rows() != A.rows() || C.columns() != B.columns().

IllegalArgumentException - if A == C || B == C.

dgemv

void dgemv(boolean transposeA,
         float alpha,
         FloatMatrix2D A,
         FloatMatrix1D x,
         float beta,
         FloatMatrix1D y)

Generalized linear algebraic matrix-vector multiply; y = alpha*A*x + beta*y. In fact equivalent to A.zMult(x,y,alpha,beta,transposeA). Note: Matrix shape conformance is checked after potential transpositions.

Parameters:

transposeA - set this flag to indicate that the multiplication shall be performed on A'.

alpha - a scale factor.

A - the source matrix.

x - the first source vector.

beta - a scale factor.

y - the second source vector, this is also the vector where results are stored.

Throws:

IllegalArgumentException - A.columns() != x.size() || A.rows() != y.size())..

dger

void dger(float alpha,
        FloatMatrix1D x,
        FloatMatrix1D y,
        FloatMatrix2D A)

Performs a rank 1 update; A = A + alpha*x*y'. Example:

         A = { {6,5}, {7,6} }, x = {1,2}, y = {3,4}, alpha = 1 -->
         A = { {9,9}, {13,14} }
 
 

Parameters:

alpha - a scalar.

x - an m element vector.

y - an n element vector.

A - an m by n matrix.

dnrm2

float dnrm2(FloatMatrix1D x)

Return the 2-norm; sqrt(x[0]^2 + x[1]^2 + ...). In fact equivalent to (float)Math.sqrt(Algebra.DEFAULT.norm2(x)).

Parameters:

x - the vector.

drot

void drot(FloatMatrix1D x,
        FloatMatrix1D y,
        float c,
        float s)

Applies a givens plane rotation to (x,y); x = c*x + s*y; y = c*y - s*x.

Parameters:

x - the first vector.

y - the second vector.

c - the cosine of the angle of rotation.

s - the sine of the angle of rotation.

drotg

void drotg(float a,
         float b,
         float[] rotvec)

Constructs a Givens plane rotation for (a,b). Taken from the LINPACK translation from FORTRAN to Java, interface slightly modified. In the LINPACK listing DROTG is attributed to Jack Dongarra

Parameters:

a - rotational elimination parameter a.

b - rotational elimination parameter b.

rotvec - Must be at least of length 4. On output contains the values {a,b,c,s}.

dscal

void dscal(float alpha,
         FloatMatrix1D x)

Vector scaling; x = alpha*x. In fact equivalent to x.assign(cern.jet.math.Functions.mult(alpha)).

Parameters:

alpha - a scale factor.

x - the first vector.

dscal

void dscal(float alpha,
         FloatMatrix2D A)

Matrix scaling; A = alpha*A. In fact equivalent to A.assign(cern.jet.math.Functions.mult(alpha)).

Parameters:

alpha - a scale factor.

A - the matrix.

dswap

void dswap(FloatMatrix1D x,
         FloatMatrix1D y)

Swaps the elements of two vectors; y <==> x. In fact equivalent to y.swap(x).

Parameters:

x - the first vector.

y - the second vector.

Throws:

IllegalArgumentException - x.size() != y.size().

dswap

void dswap(FloatMatrix2D x,
         FloatMatrix2D y)

Swaps the elements of two matrices; B <==> A.

Parameters:

x - the first matrix.

y - the second matrix.

Throws:

IllegalArgumentException - if A.columns() != B.columns() || A.rows() != B.rows().

dsymv

void dsymv(boolean isUpperTriangular,
         float alpha,
         FloatMatrix2D A,
         FloatMatrix1D x,
         float beta,
         FloatMatrix1D y)

Symmetric matrix-vector multiplication; y = alpha*A*x + beta*y. Where alpha and beta are scalars, x and y are n element vectors and A is an n by n symmetric matrix. A can be in upper or lower triangular format.

Parameters:

isUpperTriangular - is A upper triangular or lower triangular part to be used?

alpha - scaling factor.

A - the source matrix.

x - the first source vector.

beta - scaling factor.

y - the second vector holding source and destination.

dtrmv

void dtrmv(boolean isUpperTriangular,
         boolean transposeA,
         boolean isUnitTriangular,
         FloatMatrix2D A,
         FloatMatrix1D x)

Triangular matrix-vector multiplication; x = A*x or x = A'*x. Where x is an n element vector and A is an n by n unit, or non-unit, upper or lower triangular matrix.

Parameters:

isUpperTriangular - is A upper triangular or lower triangular?

transposeA - set this flag to indicate that the multiplication shall be performed on A'.

isUnitTriangular - true --> A is assumed to be unit triangular; false --> A is not assumed to be unit triangular

A - the source matrix.

x - the vector holding source and destination.

idamax

int idamax(FloatMatrix1D x)

Returns the index of largest absolute value; i such that |x[i]| == max(|x[0]|,|x[1]|,...)..

Parameters:

x - the vector to search through.

Returns:

the index of largest absolute value (-1 if x is empty).