Function test_sparse_matrix_vector_multiplies

dlib/test/sparse_vector.cpp:19–66

    dlib::logger dlog("test.sparse_vector");

    void test_sparse_matrix_vector_multiplies()
    {
        dlib::rand rnd;

        const long size = 30;

        for (int iter = 0; iter < 10; ++iter)
        {
            print_spinner();

            std::vector<sample_pair> edges;
            std::vector<ordered_sample_pair> oedges;
            matrix<double> M(size,size);
            M = 0;
            for (long i = 0; i < M.size()/3; ++i)
            {
                const long r = rnd.get_random_32bit_number()%M.nr();
                const long c = rnd.get_random_32bit_number()%M.nc();
                const double d = rnd.get_random_gaussian()*10;
                M(r,c) += d;
                oedges.push_back(ordered_sample_pair(r,c,d));
            }

            matrix<double> SM(size,size);
            SM = 0;
            for (long i = 0; i < SM.size()/3; ++i)
            {
                const long r = rnd.get_random_32bit_number()%SM.nr();
                const long c = rnd.get_random_32bit_number()%SM.nc();
                const double d = rnd.get_random_gaussian()*10;
                SM(r,c) += d;
                if (r != c)
                    SM(c,r) += d;
                edges.push_back(sample_pair(r,c,d));
            }

            const matrix<double> v = randm(size,1);

            matrix<double> result;

            sparse_matrix_vector_multiply(oedges, v, result);
            DLIB_TEST_MSG(length(M*v - result) < 1e-12, length(M*v - result));

            sparse_matrix_vector_multiply(edges, v, result);
            DLIB_TEST_MSG(length(SM*v - result) < 1e-12, length(SM*v - result));

        }
    }

// ----------------------------------------------------------------------------------------