...one of the most highly
regarded and expertly designed C++ library projects in the
world.

— Herb Sutter and Andrei
Alexandrescu, C++
Coding Standards

boost::kamada_kawai_spring_layout — Kamada-Kawai spring layout for connected, undirected graphs.

template<typenameGraph,typenamePositionMap,typenameWeightMap,typenameT,boolEdgeOrSideLength,typenameDone,typenameVertexIndexMap,typenameDistanceMatrix,typenameSpringStrengthMatrix,typenamePartialDerivativeMap>boolkamada_kawai_spring_layout(constGraph & g, PositionMap position, WeightMap weight,unspecifiededge_or_side_length, Done done,typenameproperty_traits< WeightMap >::value_type spring_constant, VertexIndexMap index, DistanceMatrix distance, SpringStrengthMatrix spring_strength, PartialDerivativeMap partial_derivatives);template<typenameGraph,typenamePositionMap,typenameWeightMap,typenameT,boolEdgeOrSideLength,typenameDone,typenameVertexIndexMap>boolkamada_kawai_spring_layout(constGraph & g, PositionMap position, WeightMap weight,unspecifiededge_or_side_length, Done done,typenameproperty_traits< WeightMap >::value_type spring_constant, VertexIndexMap index);template<typenameGraph,typenamePositionMap,typenameWeightMap,typenameT,boolEdgeOrSideLength,typenameDone>boolkamada_kawai_spring_layout(constGraph & g, PositionMap position, WeightMap weight,unspecifiededge_or_side_length, Done done,typenameproperty_traits< WeightMap >::value_type spring_constant = typename property_traits< WeightMap >::value_type(1));template<typenameGraph,typenamePositionMap,typenameWeightMap,typenameT,boolEdgeOrSideLength>boolkamada_kawai_spring_layout(constGraph & g, PositionMap position, WeightMap weight,unspecifiededge_or_side_length);

**Parameters**

- distance
(UTIL/OUT) will be used to store the distance from every vertex to every other vertex, which is computed in the first stages of the algorithm. This value's type must be a model of BasicMatrix with value type equal to the value type of the weight map. The default is a a vector of vectors.

- done
(IN) is a 4-argument function object that is passed the current value of delta_p (i.e., the energy of vertex

`p`), the vertex`p`, the graph`g`, and a boolean flag indicating whether`delta_p`is the maximum energy in the system (when`true`) or the energy of the vertex being moved. Defaults to`layout_tolerance`instantiated over the value type of the weight map.- edge_or_side_length
(IN) provides either the unit length

`e`of an edge in the layout or the length of a side`s`of the display area, and must be either`boost::edge_length(e)`or`boost::side_length(s)`, respectively.- g
(IN) must be a model of Vertex List Graph, Edge List Graph, and Incidence Graph and must be connected and undirected.

- index
(IN) is a mapping from vertices to index values between 0 and

`num_vertices(g)`. The default is`get(vertex_index,g)`.- partial_derivatives
(UTIL) will be used to store the partial derivates of each vertex with respect to the

`x`and`y`coordinates. This must be a Read/Write Property Map whose value type is a pair with both types equivalent to the value type of the weight map. The default is an iterator property map.- position
(OUT) must be a model of Lvalue Property Map, where the value type is a class containing fields

`x`and`y`that will be set to the`x`and`y`coordinates of each vertex.- spring_constant
(IN) is the constant multiplied by each spring's strength. Larger values create systems with more energy that can take longer to stabilize; smaller values create systems with less energy that stabilize quickly but do not necessarily result in pleasing layouts. The default value is 1.

- spring_strength
(UTIL/OUT) will be used to store the strength of the spring between every pair of vertices. This value's type must be a model of BasicMatrix with value type equal to the value type of the weight map. The default is a a vector of vectors.

- weight
(IN) must be a model of Readable Property Map, which provides the weight of each edge in the graph

`g`.

This algorithm [57] performs graph layout (in two dimensions) for connected, undirected graphs. It operates by relating the layout of graphs to a dynamic spring system and minimizing the energy within that system. The strength of a spring between two vertices is inversely proportional to the square of the shortest distance (in graph terms) between those two vertices. Essentially, vertices that are closer in the graph-theoretic sense (i.e., by following edges) will have stronger springs and will therefore be placed closer together.

Prior to invoking this algorithm, it is recommended that the vertices be placed along the vertices of a regular n-sided polygon via `circle_layout`.

**Returns**:
`true` if layout was successful or `false` if a negative weight cycle was detected or the graph is disconnected.

Copyright © 2004 |
Douglas Gregor, Indiana University (dgregor -at cs.indiana.edu) Andrew Lumsdaine, Indiana University (lums@osl.iu.edu) |