gtsam
Loading...
Searching...
No Matches
gtsam::Ordering Class Reference
Inheritance diagram for gtsam::Ordering:

Public Member Functions

 Ordering ()
 Create an empty ordering.
template<typename KEYS>
 Ordering (const KEYS &keys)
 Create from a container.
This & operator+= (Key key)
 Add new variables to the ordering as ordering += key1, key2, ....
This & operator, (Key key)
 Overloading the comma operator allows for chaining appends.
This & operator+= (KeyVector &keys)
 Append new keys to the ordering as ordering += keys.
bool contains (const Key &key) const
 Check if key exists in ordering.
FastMap< Key, size_t > invert () const
 Invert (not reverse) the ordering - returns a map from key to order position.
Testable
void print (const std::string &str="", const KeyFormatter &keyFormatter=DefaultKeyFormatter) const
bool equals (const Ordering &other, double tol=1e-9) const

Static Public Member Functions

Fill-reducing Orderings
template<class FACTOR_GRAPH>
static Ordering Colamd (const FACTOR_GRAPH &graph)
 Compute a fill-reducing ordering using COLAMD from a factor graph (see details for note on performance).
static Ordering Colamd (const VariableIndex &variableIndex)
 Compute a fill-reducing ordering using COLAMD from a VariableIndex.
template<class FACTOR_GRAPH>
static Ordering ColamdConstrainedLast (const FACTOR_GRAPH &graph, const KeyVector &constrainLast, bool forceOrder=false)
 Compute a fill-reducing ordering using constrained COLAMD from a factor graph (see details for note on performance).
static Ordering ColamdConstrainedLast (const VariableIndex &variableIndex, const KeyVector &constrainLast, bool forceOrder=false)
 Compute a fill-reducing ordering using constrained COLAMD from a VariableIndex.
template<class FACTOR_GRAPH>
static Ordering ColamdConstrainedFirst (const FACTOR_GRAPH &graph, const KeyVector &constrainFirst, bool forceOrder=false)
 Compute a fill-reducing ordering using constrained COLAMD from a factor graph (see details for note on performance).
static Ordering ColamdConstrainedFirst (const VariableIndex &variableIndex, const KeyVector &constrainFirst, bool forceOrder=false)
 Compute a fill-reducing ordering using constrained COLAMD from a VariableIndex.
template<class FACTOR_GRAPH>
static Ordering ColamdConstrained (const FACTOR_GRAPH &graph, const FastMap< Key, int > &groups)
 Compute a fill-reducing ordering using constrained COLAMD from a factor graph (see details for note on performance).
static Ordering ColamdConstrained (const VariableIndex &variableIndex, const FastMap< Key, int > &groups)
 Compute a fill-reducing ordering using constrained COLAMD from a VariableIndex.
template<class FACTOR_GRAPH>
static Ordering Natural (const FACTOR_GRAPH &fg)
 Return a natural Ordering. Typically used by iterative solvers.
template<class FACTOR_GRAPH>
static void CSRFormat (std::vector< int > &xadj, std::vector< int > &adj, const FACTOR_GRAPH &graph)
 METIS Formatting function.
static Ordering Metis (const MetisIndex &met, int seed=4321)
 Compute an ordering determined by METIS from a VariableIndex.
template<class FACTOR_GRAPH>
static Ordering Metis (const FACTOR_GRAPH &graph, int seed=4321)
Named Constructors
template<class FACTOR_GRAPH>
static Ordering Create (OrderingType orderingType, const FACTOR_GRAPH &graph)

Public Types

enum  OrderingType { COLAMD , METIS , NATURAL , CUSTOM }
 Type of ordering to use.
typedef Ordering This
 Typedef to this class.
typedef std::shared_ptr< This > shared_ptr
 shared_ptr to this class

Protected Types

typedef KeyVector Base

Member Function Documentation

◆ Colamd()

template<class FACTOR_GRAPH>
Ordering gtsam::Ordering::Colamd ( const FACTOR_GRAPH & graph)
inlinestatic

Compute a fill-reducing ordering using COLAMD from a factor graph (see details for note on performance).

This internally builds a VariableIndex so if you already have a VariableIndex, it is faster to use COLAMD(const VariableIndex&)

◆ ColamdConstrained() [1/2]

template<class FACTOR_GRAPH>
Ordering gtsam::Ordering::ColamdConstrained ( const FACTOR_GRAPH & graph,
const FastMap< Key, int > & groups )
inlinestatic

Compute a fill-reducing ordering using constrained COLAMD from a factor graph (see details for note on performance).

This internally builds a VariableIndex so if you already have a VariableIndex, it is faster to use COLAMD(const VariableIndex&). In this function, a group for each variable should be specified in groups, and each group of variables will appear in the ordering in group index order. groups should be a map from Key to group index. The group indices used should be consecutive starting at 0, but may appear in groups in arbitrary order. Any variables not present in groups will be assigned to group 0. This function simply fills the cmember argument to CCOLAMD with the supplied indices, see the CCOLAMD documentation for more information.

◆ ColamdConstrained() [2/2]

Ordering gtsam::Ordering::ColamdConstrained ( const VariableIndex & variableIndex,
const FastMap< Key, int > & groups )
static

Compute a fill-reducing ordering using constrained COLAMD from a VariableIndex.

In this function, a group for each variable should be specified in groups, and each group of variables will appear in the ordering in group index order. groups should be a map from Key to group index. The group indices used should be consecutive starting at 0, but may appear in groups in arbitrary order. Any variables not present in groups will be assigned to group 0. This function simply fills the cmember argument to CCOLAMD with the supplied indices, see the CCOLAMD documentation for more information.

◆ ColamdConstrainedFirst() [1/2]

template<class FACTOR_GRAPH>
Ordering gtsam::Ordering::ColamdConstrainedFirst ( const FACTOR_GRAPH & graph,
const KeyVector & constrainFirst,
bool forceOrder = false )
inlinestatic

Compute a fill-reducing ordering using constrained COLAMD from a factor graph (see details for note on performance).

This internally builds a VariableIndex so if you already have a VariableIndex, it is faster to use COLAMD(const VariableIndex&). This function constrains the variables in constrainLast to the end of the ordering, and orders all other variables before in a fill-reducing ordering. If forceOrder is true, the variables in constrainFirst will be ordered in the same order specified in the KeyVector constrainFirst. If forceOrder is false, the variables in constrainFirst will be ordered before all the others, but will be rearranged by CCOLAMD to reduce fill-in as well.

◆ ColamdConstrainedFirst() [2/2]

Ordering gtsam::Ordering::ColamdConstrainedFirst ( const VariableIndex & variableIndex,
const KeyVector & constrainFirst,
bool forceOrder = false )
static

Compute a fill-reducing ordering using constrained COLAMD from a VariableIndex.

This function constrains the variables in constrainFirst to the front of the ordering, and orders all other variables after in a fill-reducing ordering. If forceOrder is true, the variables in constrainFirst will be ordered in the same order specified in the KeyVector constrainFirst. If forceOrder is false, the variables in constrainFirst will be ordered before all the others, but will be rearranged by CCOLAMD to reduce fill-in as well.

◆ ColamdConstrainedLast() [1/2]

template<class FACTOR_GRAPH>
Ordering gtsam::Ordering::ColamdConstrainedLast ( const FACTOR_GRAPH & graph,
const KeyVector & constrainLast,
bool forceOrder = false )
inlinestatic

Compute a fill-reducing ordering using constrained COLAMD from a factor graph (see details for note on performance).

This internally builds a VariableIndex so if you already have a VariableIndex, it is faster to use COLAMD(const VariableIndex&). This function constrains the variables in constrainLast to the end of the ordering, and orders all other variables before in a fill-reducing ordering. If forceOrder is true, the variables in constrainLast will be ordered in the same order specified in the KeyVector constrainLast. If forceOrder is false, the variables in constrainLast will be ordered after all the others, but will be rearranged by CCOLAMD to reduce fill-in as well.

◆ ColamdConstrainedLast() [2/2]

Ordering gtsam::Ordering::ColamdConstrainedLast ( const VariableIndex & variableIndex,
const KeyVector & constrainLast,
bool forceOrder = false )
static

Compute a fill-reducing ordering using constrained COLAMD from a VariableIndex.

This function constrains the variables in constrainLast to the end of the ordering, and orders all other variables before in a fill-reducing ordering. If forceOrder is true, the variables in constrainLast will be ordered in the same order specified in the KeyVector constrainLast. If forceOrder is false, the variables in constrainLast will be ordered after all the others, but will be rearranged by CCOLAMD to reduce fill-in as well.

◆ invert()

FastMap< Key, size_t > gtsam::Ordering::invert ( ) const

Invert (not reverse) the ordering - returns a map from key to order position.

Returns
FastMap<Key, size_t>

◆ Metis()

Ordering gtsam::Ordering::Metis ( const MetisIndex & met,
int seed = 4321 )
static

Compute an ordering determined by METIS from a VariableIndex.

METIS's nested dissection is randomized: seed initializes METIS's internal generator, which drives its initial bisections, matching order, and refinement tie-breaks. The default 4321 is the value METIS itself uses when no seed is supplied, so it reproduces the historical ordering exactly; other seeds draw different orderings, and keeping the best of k seeds (by fill or maximum clique) is a cheap quality boost.

◆ operator+=()

Ordering::This & gtsam::Ordering::operator+= ( KeyVector & keys)

Append new keys to the ordering as ordering += keys.

Parameters
keysThe key vector to append to this ordering.
Returns
The ordering variable with appended keys.
The documentation for this class was generated from the following files:
  • /tmp/gtsam-4.3.0-doxygen.rsXPUS/source/gtsam/inference/Ordering.h
  • /tmp/gtsam-4.3.0-doxygen.rsXPUS/source/gtsam/inference/Ordering.cpp