1

我正在开发一个需要使用 2d Boost.MultiArray 的程序。我设法初始化它并用数据填充它。但我不明白如何获取 size 的子数组i，j如果 multiarray 是 size m，n。哪里i<=m和j<=n。谁能帮我？
代码：

matrix_type matrix(boost::extents[width][height]);
        read_matrix_from_file(file_content, matrix);
for (int rank = 1; rank < workers; rank++) {
            auto subarray_size = (rest > 0) ? lines_per_worker + 1 : lines_per_worker;
            rest--;

            typedef boost::multi_array_types::index_range range;


            size_t finish_line = subarray_size + bias - 1;
            finish_line = (finish_line==bias)? finish_line+1:finish_line;


            matrix_type::array_view<2>::type
                    current_process_batch = matrix[boost::indices[range(bias, subarray_size + bias - 1)][range(0, width)]];
}
4

1 回答 1

1

文档https://www.boost.org/doc/libs/1_73_0/libs/multi_array/doc/user.html#sec_generators

Boost.MultiArray 提供了创建现有数组组件的子视图的工具。它允许您创建一个子视图，该子视图保留与原始数组相同的维数，或者一个维数也比原始数组少的子视图。

子视图的创建是通过调用 operator[] 并传递给它一个 index_gen 类型来实现的。index_gen 通过将 index_range 对象传递给它的 operator[] 来填充。index_range 和 index_gen 类型在 multi_array_types 命名空间中定义，并作为每个数组类型的嵌套成员。与 boost::extents 类似，该库默认构造对象 boost::indices。您可以通过在包含库头之前定义 BOOST_MULTI_ARRAY_NO_GENERATORS 来抑制此对象。下面是一个简单的子视图创建示例。

样本：

住在科利鲁

#include <boost/multi_array.hpp>
#include <iostream>

template <typename Ma>
auto dump(Ma const& r) -> std::enable_if_t<1 == Ma::dimensionality> {
    for (auto c: r)
        std::cout << " " << c;
    std::cout << "\n";
}

template <typename Ma>
auto dump(Ma const& ma) -> std::enable_if_t<2 == Ma::dimensionality> {
    for (auto const& r: ma)
        dump(r);
}

int main() {using range = boost::multi_array_types::index_range;
    using boost::extents;
    using boost::indices;

    auto mn = extents[7][4];
    boost::multi_array<int, 2> ma(mn);

    // fill with numbers from 10..37
    std::iota(ma.data(), ma.data() + ma.num_elements(), 10);

    dump(ma);

    std::cout << "Slice [3..5][1..3]\n";
    dump(ma[ indices[range(3,5)][range(1,3)] ]);

    std::cout << "Slice [2..4][2..3]\n";
    dump(ma[ indices[range(2,4)][range(2,3)] ]);

    std::cout << "Slice [1,3,5][2..3]\n";
    dump(ma[ indices[range(1,7,2)][range(2,3)] ]);

    std::cout << "Degenerate views (reduced dimensionality):\n";
    std::cout << "Slice [1,3,5][2]\n";
    dump(ma[ indices[range(1,7,2)][2] ]);

    std::cout << "Slice [0][1,3]\n";
    dump(ma[ indices[0][range(1,4,2)] ]);
}

印刷

 10 11 12 13
 14 15 16 17
 18 19 20 21
 22 23 24 25
 26 27 28 29
 30 31 32 33
 34 35 36 37
Slice [3..5][1..3]
 23 24
 27 28
Slice [2..4][2..3]
 20
 24
Slice [1,3,5][2..3]
 16
 24
 32
Degenerate views (reduced dimensionality):
Slice [1,3,5][2]
 16 24 32
Slice [0][1,3]
 11 13

更新

回覆。评论：

如果你想避免退化的尺寸，这里是如何：

住在科利鲁

std::cout << "Degenerate views (reduced dimensionality):\n";
std::cout << "Slice [1,3,5][2]\n";
dump(ma[ indices[range(1,7,2)][2] ]);

std::cout << "Slice [0][1,3]\n";
dump(ma[ indices[0][range(1,4,2)] ]);

// RE: Comment
std::cout << "NON-Degenerate views (reduced dimensionality):\n";
std::cout << "Slice [1,3,5][2]\n";
dump(ma[ indices[range(1,7,2)][range(2,3)] ]);

std::cout << "Slice [0][1,3]\n";
dump(ma[ indices[range(0,1)][range(1,4,2)] ]);

印刷

Degenerate views (reduced dimensionality):
Slice [1,3,5][2]
 16 24 32
Slice [0][1,3]
 11 13
NON-Degenerate views (reduced dimensionality):
Slice [1,3,5][2]
 16
 24
 32
Slice [0][1,3]
 11 13

基本上，不要使用文字[i][j]，但是[range(i,i+1)][range(j,j+1)]

于 2020-05-30T19:43:39.073 回答