Interface for Nearest Neighbor Interchanges algorithms. More...

#include <Bpp/Phyl/Tree/NNISearchable.h>

Inheritance diagram for bpp::NNISearchable:

Collaboration diagram for bpp::NNISearchable:

Public Member Functions
	NNISearchable ()

virtual	~NNISearchable ()

virtual NNISearchable *	clone () const =0

virtual double	testNNI (int nodeId) const =0
	Send the score of a NNI movement, without performing it. More...

virtual void	doNNI (int nodeId)=0
	Perform a NNI movement. More...

virtual const Tree &	topology () const =0
	Get the tree associated to this NNISearchable object. More...

virtual double	getTopologyValue () const =0
	Get the current score of this NNISearchable object. More...

virtual void	topologyChangePerformed (const TopologyChangeEvent &event)
	Notify a topology change event. More...

virtual void	topologyChangeTested (const TopologyChangeEvent &event)=0
	Notify a topology change event. More...

virtual void	topologyChangeSuccessful (const TopologyChangeEvent &event)=0
	Tell that a topology change is definitive. More...

Detailed Description

Interface for Nearest Neighbor Interchanges algorithms.

This interface defines the methods to work with NNI algorithms.

NNISearchable objects are supposed to work with TreeTemplate objects. NNI are defined as follow:

------------->
    +------- C
    |
D --+ X  +-- B
    |    |
    +----+ F
         |
         +-- A

Where: -F is the focus (parent) node, -A and B are the son of F -X is the parent of F and so on. Two NNI's are possible for branch (XF):

swapping B and C, which is the same as D and A
swapping A and C, which is the same as D and B Because of the rooted representation, we'll consider B $\leftrightarrow$ C and A $\leftrightarrow$ C, which are simpler to perform.

For unrooted tree, we have at the 'root' node:

------------->
  +------- D
  |
X +------- C
  |
  |    +-- B
  |    |
  +----+ F
       |
       +-- A

In this case, we swap A or B with one of C or D. Which one of C or D depends on the implementation, but will always be the same, so that swapping A or B involve 2 distinct NNI.

Definition at line 60 of file NNISearchable.h.

Constructor & Destructor Documentation

◆ NNISearchable()

bpp::NNISearchable::NNISearchable ( )

inline

Definition at line 65 of file NNISearchable.h.

◆ ~NNISearchable()

virtual bpp::NNISearchable::~NNISearchable ( )

inlinevirtual

Definition at line 66 of file NNISearchable.h.

Member Function Documentation

◆ clone()

virtual NNISearchable* bpp::NNISearchable::clone ( ) const

pure virtual

Implements bpp::Clonable.

Implemented in bpp::DRTreeParsimonyScore, and bpp::NNIHomogeneousTreeLikelihood.

Referenced by bpp::NNITopologySearch::searchPhyML().

◆ doNNI()

virtual void bpp::NNISearchable::doNNI ( int nodeId )

pure virtual

Perform a NNI movement.

Parameters

nodeId The id of the node defining the NNI movement.

Exceptions

NodeException If the node does not define a valid NNI.

Implemented in bpp::DRTreeParsimonyScore, and bpp::NNIHomogeneousTreeLikelihood.

◆ getTopologyValue()

virtual double bpp::NNISearchable::getTopologyValue ( ) const

pure virtual

Get the current score of this NNISearchable object.

Returns: The current score of this instance.

Implemented in bpp::DRTreeParsimonyScore, and bpp::NNIHomogeneousTreeLikelihood.

◆ testNNI()

virtual double bpp::NNISearchable::testNNI ( int nodeId ) const

pure virtual

Send the score of a NNI movement, without performing it.

This methods sends the score variation. This variation must be negative if the new point is better, i.e. the object is to be used with a minimizing optimization (for consistence with Optimizer objects).

Parameters

nodeId The id of the node defining the NNI movement.