Description of __mult

__mult_dim		Programmers Guide
		Contents \| Index \| Search

Category:algorithms

Component type:function

Prototype

template <class MatA, class MatB, class MatC> void __mult_dim(const MatA& A, const MatB& B, MatC& C, twod_tag) ;

Description

The actual specialization of the algorithm used depends of the types of matrices used. If all the matrices are dense and rectangular the blocked algorithm is used (when --with-blais is specified in the configure). Otherwise the traversal depends on matrix A. Therefore if one is multiplying a sparse matrix by a dense, one would want the sparse matrix as the A argument. Typically, for performance reasons, one would not want to use a sparse matrix for C.

Note: ignore the twod_tag argument and the underscores in the name of this function.

Definition

mtl.h

Requirements on types

the value types for each of the matrices must be compatible
the multiplication operator must be defined for MatA::value_type
the addition operator must be defined for MatA::value_type

Preconditions

A.nrows() == C.nrows()
A.ncols() == B.nrows()
B.ncols() == C.ncols()

Complexity

Example

Notes

See also