从BGL的图提取邻接矩阵 [英] Extract the adjacency matrix from a BGL graph

问题描述

使用的 Boost图库我要寻找一种方式为从底层图重新$ P $由要么psented提取邻接矩阵 提高::的adjacency_list 或的boost :: adjacency_matrix 。我想，要使用这个矩阵提高::数字:: uBLAS库求解线性方程组的系统。

Using the Boost Graph Library I am looking for a way to extract the adjacency matrix from an underlying graph represented by either boost::adjacency_list or boost::adjacency_matrix. I'd like to use this matrix in conjunction with boost::numeric::ublas to solve a system of simultaneous linear equations.

下面是一个最小的例子，让你去：

Here is a minimal example to get you going:

#include <boost/graph/adjacency_list.hpp>
#include <boost/graph/adjacency_matrix.hpp>

using namespace boost;

typedef boost::adjacency_list< listS, vecS, directedS > ListGraph;
typedef boost::adjacency_matrix< directedS > MatrixGraph;

int main(){ 

  ListGraph lg; 
  add_edge (0, 1, lg); 
  add_edge (0, 3, lg); 
  add_edge (1, 2, lg); 
  add_edge (2, 3, lg); 

  //How do I get the adjacency matrix underlying lg?

  MatrixGraph mg(3); 
  add_edge (0, 1, mg); 
  add_edge (0, 3, mg); 
  add_edge (1, 2, mg); 
  add_edge (2, 3, mg); 

  //How do I get the adjacency matrix underlying mg?

}

如果任何人都可以想出一个有效的方式来获得邻接矩阵，我将非常感激。理想的是，溶液与uBLAS库兼容。我不知道是否有避免在整个图形迭代的方式。

If anyone could come up with an efficient way to obtain the adjacency matrix I would be much obliged. Ideally the solution is compatible with uBLAS. I wonder if there is a way to avoid iteration through the entire graph.

推荐答案

要的adjacency_list转换成adjacency_matrix最简单的方法是使用的boost :: copy_graph

The easiest way to convert adjacency_list into adjacency_matrix is to use boost::copy_graph

您code为 MatrixGraph毫克应作如下修改

Your code for MatrixGraph mg should be modified as follows

#include <boost/graph/copy.hpp>
#include <cassert>

using namespace boost;

typedef boost::adjacency_list< listS, vecS, directedS > ListGraph;
typedef boost::adjacency_matrix< directedS > MatrixGraph;

int main(){

    ListGraph lg;
    add_edge(0, 1, lg);
    add_edge(0, 3, lg);
    add_edge(1, 2, lg);
    add_edge(2, 3, lg);

    //How do I get the adjacency matrix underlying lg?

    //How do I get the adjacency matrix underlying mg?   
    MatrixGraph mg( num_vertices(lg));
    boost::copy_graph(lg, mg);
}

现在，使用具有的uBLAS或类似的邻接矩阵，你可以写一个简单的访问类，使语法的差别的更柔顺。继续previous片段，我们得到：

Now, to use adjacency matrix with ublas or similar, you can write a simple "access" class to make syntax more compliant with ublas. Continuing previous snippet we get:

template <class Graph>
class MatrixAccessor
{
public:
    typedef typename Graph::Matrix Matrix; //actually a vector<
    typedef typename Matrix::const_reference const_reference;


    MatrixAccessor(const Graph* g)
        : m_g(g)
    {
        static_assert(boost::is_same<size_t, typename Graph::vertex_descriptor>::value, "Vertex descriptor should be of integer type");
    }

    const_reference operator()(size_t u, size_t v) const
    {
        return m_g->get_edge(u, v);
    }

    const Graph* m_g;
};

void use_matrix(const MatrixGraph & mg)
{
    MatrixAccessor<MatrixGraph> matr(&mg);
    assert(matr(0, 1) == 1);
    assert(matr(0, 2) == 0);
}

在情况下，你adjacency_matrix有一些边缘捆绑的属性，您可能需要MatrixAccessor修改运算符（）。

In case your adjacency_matrix has some edge-bundled properties, you might need to modify the operator() in MatrixAccessor.

根据你多少的差别的使用，可以进一步细化MatrixAccessor。例如， out_edge_iterator 的MatrixGraph的定顶点实际上是在矩阵列的迭代器; vertex_iterator可以被视为迭代以上矩阵的行，等等。

Depending on how much uBLAS you use, you can refine MatrixAccessor further. For example, out_edge_iterator for a given vertex of a MatrixGraph is actually an iterator over matrix column; vertex_iterator can be treated as iterator over matrix rows, etc.

当然，图矩阵是不可改变的，因此应谨慎使用。

Of course, graph matrix is immutable and as such should be used with care.

这篇关于从BGL的图提取邻接矩阵的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！