MatrixProduct.h

浏览该文件的文档.

/*
 * Copyright (c) The Shogun Machine Learning Toolbox
 * Written (w) 2014 Khaled Nasr
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are met:
 *
 * 1. Redistributions of source code must retain the above copyright notice, this
 *    list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright notice,
 *    this list of conditions and the following disclaimer in the documentation
 *    and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 * The views and conclusions contained in the software and documentation are those
 * of the authors and should not be interpreted as representing official policies,
 * either expressed or implied, of the Shogun Development Team.
 */

#ifndef MATRIX_PRODUCT_IMPL_H_
#define MATRIX_PRODUCT_IMPL_H_

#include <shogun/lib/config.h>
#include <shogun/lib/SGMatrix.h>

#ifdef HAVE_EIGEN3
#include <shogun/mathematics/eigen3.h>
#endif // HAVE_EIGEN3

#ifdef HAVE_VIENNACL
#include <shogun/lib/GPUMatrix.h>
#include <viennacl/linalg/prod.hpp>
#include <viennacl/matrix.hpp>
#endif // HAVE_VIENNACL

namespace shogun
{

namespace linalg
{

namespace implementation
{

template <enum Backend, class Matrix>
struct matrix_product
{
    typedef typename Matrix::Scalar T;

    static void compute(Matrix A, Matrix B, Matrix C,
        bool transpose_A, bool transpose_B, bool overwrite);
};

#ifdef HAVE_EIGEN3

template <class Matrix>
struct matrix_product<Backend::EIGEN3, Matrix>
{
    typedef typename Matrix::Scalar T;

    typedef SGMatrix<T> ReturnType;

    typedef Eigen::Matrix<T,Eigen::Dynamic,Eigen::Dynamic> MatrixXt;

    static ReturnType compute(SGMatrix<T> A, SGMatrix<T> B,
        bool transpose_A, bool transpose_B)
    {
        REQUIRE(A.matrix, "Matrix A is not initialized!\n");
        REQUIRE(B.matrix, "Matrix B is not initialized!\n");
        REQUIRE(A.num_cols == B.num_rows, "Number of columns for A (%d) and "
                "number of rows for B (%d) should be equal!\n", A.num_cols, B.num_rows);

        ReturnType retMatrix(A.num_rows, B.num_cols);
        compute(A, B, retMatrix, transpose_A, transpose_B, true);

        return retMatrix;
    }

    static void compute(SGMatrix<T> A, SGMatrix<T> B, SGMatrix<T> C,
        bool transpose_A, bool transpose_B, bool overwrite)
    {
        Eigen::Map<MatrixXt> A_eig = A;
        Eigen::Map<MatrixXt> B_eig = B;
        Eigen::Map<MatrixXt> C_eig = C;

        if (overwrite)
        {
            if (transpose_A && transpose_B)
                C_eig = A_eig.transpose() * B_eig.transpose();

            else if (transpose_A)
                C_eig = A_eig.transpose() * B_eig;

            else if (transpose_B)
                C_eig = A_eig * B_eig.transpose();

            else
                C_eig = A_eig * B_eig;
        }
        else
        {
            if (transpose_A && transpose_B)
                C_eig += A_eig.transpose() * B_eig.transpose();

            else if (transpose_A)
                C_eig += A_eig.transpose() * B_eig;

            else if (transpose_B)
                C_eig += A_eig * B_eig.transpose();

            else
                C_eig += A_eig * B_eig;
        }
    }
};
#endif // HAVE_EIGEN3

#ifdef HAVE_VIENNACL

template <class Matrix>
struct matrix_product<Backend::VIENNACL, Matrix>
{
    typedef typename Matrix::Scalar T;

    typedef CGPUMatrix<T> ReturnType;

    static ReturnType compute(CGPUMatrix<T> A, CGPUMatrix<T> B,
        bool transpose_A, bool transpose_B)
    {
        REQUIRE(A.matrix, "Matrix A is not initialized!\n");
        REQUIRE(B.matrix, "Matrix B is not initialized!\n");
        REQUIRE(A.num_cols == B.num_rows, "Number of columns for A (%d) and "
                "number of rows for B (%d) should be equal!\n", A.num_cols, B.num_rows);

        ReturnType retMatrix(A.num_rows, B.num_cols);
        compute(A, B, retMatrix, transpose_A, transpose_B, true);

        return retMatrix;
    }

    static void compute(CGPUMatrix<T> A, CGPUMatrix<T> B, CGPUMatrix<T> C,
        bool transpose_A, bool transpose_B, bool overwrite)
    {
        if (overwrite)
        {
            if (transpose_A && transpose_B)
                C.vcl_matrix() = viennacl::linalg::prod(
                    viennacl::trans(A.vcl_matrix()), viennacl::trans(B.vcl_matrix()));

            else if (transpose_A)
                C.vcl_matrix() = viennacl::linalg::prod(
                    viennacl::trans(A.vcl_matrix()), B.vcl_matrix());

            else if (transpose_B)
                C.vcl_matrix() = viennacl::linalg::prod(
                    A.vcl_matrix(), viennacl::trans(B.vcl_matrix()));

            else
                C.vcl_matrix() = viennacl::linalg::prod(A.vcl_matrix(), B.vcl_matrix());
        }
        else
        {
            if (transpose_A && transpose_B)
                C.vcl_matrix() += viennacl::linalg::prod(
                    viennacl::trans(A.vcl_matrix()), viennacl::trans(B.vcl_matrix()));

            else if (transpose_A)
                C.vcl_matrix() += viennacl::linalg::prod(
                    viennacl::trans(A.vcl_matrix()), B.vcl_matrix());

            else if (transpose_B)
                C.vcl_matrix() += viennacl::linalg::prod(
                    A.vcl_matrix(), viennacl::trans(B.vcl_matrix()));

            else
                C.vcl_matrix() += viennacl::linalg::prod(A.vcl_matrix(), B.vcl_matrix());
        }
    }
};

#endif // HAVE_VIENNACL

}

}

}
#endif // MATRIX_PRODUCT_IMPL_H_