doc/graphs.dox
author deba
Thu, 02 Sep 2004 10:07:30 +0000
changeset 782 df2e45e09652
parent 666 410a1419e86b
child 808 9cabbdd73375
permissions -rw-r--r--
--This line, and those below, will be ignored--

A hugo/sym_map_factory.h
M hugo/list_graph.h
A hugo/array_map_factory.h
A hugo/map_registry.h
M hugo/smart_graph.h
A hugo/map_defines.h
A hugo/extended_pair.h
M hugo/full_graph.h
A hugo/vector_map_factory.h
ladanyi@666
     1
/*!
ladanyi@666
     2
ladanyi@666
     3
\page graphs How to use graphs
ladanyi@666
     4
alpar@756
     5
The primary data structures of HugoLib are the graph classes. They all
alpar@756
     6
provide a node list - edge list interface, i.e. they have
alpar@756
     7
functionalities to list the nodes and the edges of the graph as well
alpar@756
     8
as in incoming and outgoing edges of a given node. 
alpar@756
     9
alpar@756
    10
alpar@756
    11
Each graph should meet the \ref ConstGraph concept. This concept does
alpar@756
    12
makes it possible to change the graph (i.e. it is not possible to add
alpar@756
    13
or delete edges or nodes). Most of the graph algorithms will run on
alpar@756
    14
these graphs.
alpar@756
    15
alpar@756
    16
The graphs meeting the \ref ExtendableGraph concept allow node and
alpar@756
    17
edge addition. You can also "clear" (i.e. erase all edges and nodes)
alpar@756
    18
such a graph.
alpar@756
    19
alpar@756
    20
In case of graphs meeting the full feature \ref ErasableGraph concept
alpar@756
    21
you can also erase individual edges and node in arbitrary order.
alpar@756
    22
alpar@756
    23
The implemented graph structures are the following.
alpar@756
    24
\li \ref hugo::ListGraph "ListGraph" is the most versatile graph class. It meets
alpar@756
    25
the ErasableGraph concept and it also have some convenience features.
alpar@756
    26
\li \ref hugo::SmartGraph "SmartGraph" is a more memory
alpar@756
    27
efficient version of \ref hugo::ListGraph "ListGraph". The
alpar@756
    28
price of it is that it only meets the \ref ExtendableGraph concept,
alpar@756
    29
so you cannot delete individual edges or nodes.
alpar@756
    30
\li \ref hugo::SymListGraph "SymListGraph" and
alpar@756
    31
\ref hugo::SymSmartGraph "SymSmartGraph" classes are very similar to
alpar@756
    32
\ref hugo::ListGraph "ListGraph" and \ref hugo::SmartGraph "SmartGraph".
alpar@756
    33
The difference is that whenever you add a
alpar@756
    34
new edge to the graph, it actually adds a pair of oppositely directed edges.
alpar@756
    35
They are linked together so it is possible to access the counterpart of an
alpar@756
    36
edge. An even more important feature is that using these classes you can also
alpar@756
    37
attach data to the edges in such a way that the stored data
alpar@756
    38
are shared by the edge pairs. 
alpar@756
    39
\li \ref hugo::FullGraph "FullGraph"
alpar@756
    40
implements a full graph. It is a \ref ConstGraph, so you cannot
alpar@756
    41
change the number of nodes once it is constructed. It is extremely memory
alpar@756
    42
efficient: it uses constant amount of memory independently from the number of
alpar@756
    43
the nodes of the graph. Of course, the size of the \ref maps "NodeMap"'s and
alpar@756
    44
\ref maps "EdgeMap"'s will depend on the number of nodes.
alpar@756
    45
alpar@756
    46
\li \ref hugo::NodeSet "NodeSet" implements a graph with no edges. This class
alpar@756
    47
can be used as a base class of \ref hugo::EdgeSet "EdgeSet".
alpar@756
    48
\li \ref hugo::EdgeSet "EdgeSet" can be used to create a new graph on
alpar@756
    49
the edge set of another graph. The base graph can be an arbitrary graph and it
alpar@756
    50
is possible to attach several \ref hugo::EdgeSet "EdgeSet"'s to a base graph.
alpar@756
    51
alpar@756
    52
\todo Don't we need SmartNodeSet and SmartEdgeSet?
alpar@756
    53
\todo Some cross-refs are wrong.
alpar@756
    54
alpar@756
    55
alpar@756
    56
The graph structures itself can not store data attached
alpar@756
    57
to the edges and nodes. However they all provide
alpar@756
    58
\ref maps "map classes"
alpar@756
    59
to dynamically attach data the to graph components.
alpar@756
    60
alpar@756
    61
alpar@756
    62
alpar@756
    63
ladanyi@666
    64
The following program demonstrates the basic features of HugoLib's graph
ladanyi@666
    65
structures.
ladanyi@666
    66
ladanyi@666
    67
\code
ladanyi@666
    68
#include <iostream>
ladanyi@666
    69
#include <hugo/list_graph.h>
ladanyi@666
    70
ladanyi@666
    71
using namespace hugo;
ladanyi@666
    72
ladanyi@666
    73
int main()
ladanyi@666
    74
{
ladanyi@666
    75
  typedef ListGraph Graph;
ladanyi@666
    76
\endcode
ladanyi@666
    77
ladanyi@666
    78
ListGraph is one of HugoLib's graph classes. It is based on linked lists,
ladanyi@666
    79
therefore iterating throuh its edges and nodes is fast.
ladanyi@666
    80
ladanyi@666
    81
\code
ladanyi@666
    82
  typedef Graph::Edge Edge;
ladanyi@666
    83
  typedef Graph::InEdgeIt InEdgeIt;
ladanyi@666
    84
  typedef Graph::OutEdgeIt OutEdgeIt;
ladanyi@666
    85
  typedef Graph::EdgeIt EdgeIt;
ladanyi@666
    86
  typedef Graph::Node Node;
ladanyi@666
    87
  typedef Graph::NodeIt NodeIt;
ladanyi@666
    88
ladanyi@666
    89
  Graph g;
ladanyi@666
    90
  
ladanyi@666
    91
  for (int i = 0; i < 3; i++)
ladanyi@666
    92
    g.addNode();
ladanyi@666
    93
  
ladanyi@666
    94
  for (NodeIt i(g); g.valid(i); g.next(i))
ladanyi@666
    95
    for (NodeIt j(g); g.valid(j); g.next(j))
ladanyi@666
    96
      if (i != j) g.addEdge(i, j);
ladanyi@666
    97
\endcode
ladanyi@666
    98
ladanyi@666
    99
After some convenience typedefs we create a graph and add three nodes to it.
ladanyi@666
   100
Then we add edges to it to form a full graph.
ladanyi@666
   101
ladanyi@666
   102
\code
ladanyi@666
   103
  std::cout << "Nodes:";
ladanyi@666
   104
  for (NodeIt i(g); g.valid(i); g.next(i))
ladanyi@666
   105
    std::cout << " " << g.id(i);
ladanyi@666
   106
  std::cout << std::endl;
ladanyi@666
   107
\endcode
ladanyi@666
   108
ladanyi@666
   109
Here we iterate through all nodes of the graph. We use a constructor of the
ladanyi@666
   110
node iterator to initialize it to the first node. The next member function is
ladanyi@666
   111
used to step to the next node, and valid is used to check if we have passed the
ladanyi@666
   112
last one.
ladanyi@666
   113
ladanyi@666
   114
\code
ladanyi@666
   115
  std::cout << "Nodes:";
ladanyi@666
   116
  NodeIt n;
ladanyi@666
   117
  for (g.first(n); n != INVALID; g.next(n))
ladanyi@666
   118
    std::cout << " " << g.id(n);
ladanyi@666
   119
  std::cout << std::endl;
ladanyi@666
   120
\endcode
ladanyi@666
   121
ladanyi@666
   122
Here you can see an alternative way to iterate through all nodes. Here we use a
ladanyi@666
   123
member function of the graph to initialize the node iterator to the first node
ladanyi@666
   124
of the graph. Using next on the iterator pointing to the last node invalidates
ladanyi@666
   125
the iterator i.e. sets its value to INVALID. Checking for this value is
ladanyi@666
   126
equivalent to using the valid member function.
ladanyi@666
   127
ladanyi@666
   128
Both of the previous code fragments print out the same:
ladanyi@666
   129
ladanyi@666
   130
\code
ladanyi@666
   131
Nodes: 2 1 0
ladanyi@666
   132
\endcode
ladanyi@666
   133
ladanyi@666
   134
\code
ladanyi@666
   135
  std::cout << "Edges:";
ladanyi@666
   136
  for (EdgeIt i(g); g.valid(i); g.next(i))
ladanyi@666
   137
    std::cout << " (" << g.id(g.tail(i)) << "," << g.id(g.head(i)) << ")";
ladanyi@666
   138
  std::cout << std::endl;
ladanyi@666
   139
\endcode
ladanyi@666
   140
ladanyi@666
   141
\code
ladanyi@666
   142
Edges: (0,2) (1,2) (0,1) (2,1) (1,0) (2,0)
ladanyi@666
   143
\endcode
ladanyi@666
   144
ladanyi@666
   145
We can also iterate through all edges of the graph very similarly. The head and
ladanyi@666
   146
tail member functions can be used to access the endpoints of an edge.
ladanyi@666
   147
ladanyi@666
   148
\code
ladanyi@666
   149
  NodeIt first_node(g);
ladanyi@666
   150
ladanyi@666
   151
  std::cout << "Out-edges of node " << g.id(first_node) << ":";
ladanyi@666
   152
  for (OutEdgeIt i(g, first_node); g.valid(i); g.next(i))
ladanyi@666
   153
    std::cout << " (" << g.id(g.tail(i)) << "," << g.id(g.head(i)) << ")"; 
ladanyi@666
   154
  std::cout << std::endl;
ladanyi@666
   155
ladanyi@666
   156
  std::cout << "In-edges of node " << g.id(first_node) << ":";
ladanyi@666
   157
  for (InEdgeIt i(g, first_node); g.valid(i); g.next(i))
ladanyi@666
   158
    std::cout << " (" << g.id(g.tail(i)) << "," << g.id(g.head(i)) << ")"; 
ladanyi@666
   159
  std::cout << std::endl;
ladanyi@666
   160
\endcode
ladanyi@666
   161
ladanyi@666
   162
\code
ladanyi@666
   163
Out-edges of node 2: (2,0) (2,1)
ladanyi@666
   164
In-edges of node 2: (0,2) (1,2)
ladanyi@666
   165
\endcode
ladanyi@666
   166
ladanyi@666
   167
We can also iterate through the in and out-edges of a node. In the above
ladanyi@666
   168
example we print out the in and out-edges of the first node of the graph.
ladanyi@666
   169
ladanyi@666
   170
\code
ladanyi@666
   171
  Graph::EdgeMap<int> m(g);
ladanyi@666
   172
ladanyi@666
   173
  for (EdgeIt e(g); g.valid(e); g.next(e))
ladanyi@666
   174
    m.set(e, 10 - g.id(e));
ladanyi@666
   175
  
ladanyi@666
   176
  std::cout << "Id Edge  Value" << std::endl;
ladanyi@666
   177
  for (EdgeIt e(g); g.valid(e); g.next(e))
ladanyi@666
   178
    std::cout << g.id(e) << "  (" << g.id(g.tail(e)) << "," << g.id(g.head(e))
ladanyi@666
   179
      << ") " << m[e] << std::endl;
ladanyi@666
   180
\endcode
ladanyi@666
   181
ladanyi@666
   182
\code
ladanyi@666
   183
Id Edge  Value
ladanyi@666
   184
4  (0,2) 6
ladanyi@666
   185
2  (1,2) 8
ladanyi@666
   186
5  (0,1) 5
ladanyi@666
   187
0  (2,1) 10
ladanyi@666
   188
3  (1,0) 7
ladanyi@666
   189
1  (2,0) 9
ladanyi@666
   190
\endcode
ladanyi@666
   191
ladanyi@666
   192
In generic graph optimization programming graphs are not containers rather
ladanyi@666
   193
incidence structures which are iterable in many ways. HugoLib introduces
ladanyi@666
   194
concepts that allow us to attach containers to graphs. These containers are
ladanyi@666
   195
called maps.
ladanyi@666
   196
ladanyi@666
   197
In the example above we create an EdgeMap which assigns an int value to all
ladanyi@666
   198
edges of the graph. We use the set member function of the map to write values
ladanyi@666
   199
into the map and the operator[] to retrieve them.
ladanyi@666
   200
ladanyi@666
   201
Here we used the maps provided by the ListGraph class, but you can also write
ladanyi@666
   202
your own maps. You can read more about using maps \ref maps "here".
ladanyi@666
   203
ladanyi@666
   204
*/