BKSolver< T > Class Template Reference

MaxFlowSolver backed by the Boykov–Kolmogorov max-flow algorithm. More...

#include <BKSolver.hpp>

Inheritance diagram for BKSolver< T >:

Collaboration diagram for BKSolver< T >:

Public Member Functions
	BKSolver (int var_num_max, int edge_num_max)
	Constructs the solver with pre-allocated capacity.
	~BKSolver () override
MaxFlowSolver< T >::Var	add_variable () override
	Introduces a new binary variable and returns its handle.
void	add_constant (T E) override
	Adds a constant term to the energy function.
void	add_term1 (typename MaxFlowSolver< T >::Var x, T E0, T E1) override
void	add_term2 (typename MaxFlowSolver< T >::Var x, typename MaxFlowSolver< T >::Var y, T E00, T E01, T E10, T E11) override
T	minimize () override
	Minimizes the energy and returns the minimum value.
int	get_var (typename MaxFlowSolver< T >::Var x) override
Public Member Functions inherited from MaxFlowSolver< T >
virtual	~MaxFlowSolver ()=default
virtual void	add_term1 (Var x, T E0, T E1)=0
	Adds a unary term E(x) where x ∈ {0, 1}.
virtual void	add_term2 (Var x, Var y, T E00, T E01, T E10, T E11)=0
	Adds a pairwise term E(x, y) where x, y ∈ {0, 1}.
virtual int	get_var (Var x)=0
	Returns the optimal value of variable x (0 or 1) after minimize().

Additional Inherited Members
Public Types inherited from MaxFlowSolver< T >
typedef int	Var
	Integer handle identifying a binary variable.

Detailed Description

template<typename T>
class BKSolver< T >

MaxFlowSolver backed by the Boykov–Kolmogorov max-flow algorithm.

This solver wraps the original BK implementation by Yuri Boykov and Vladimir Kolmogorov. It is the default solver and is available in all builds.

License note: The underlying BK implementation is restricted to non-commercial, research use only. See the project README for details.

Template Parameters

T	Numeric cost type (int32_t, float, or double).

Constructor & Destructor Documentation

BKSolver()

template<typename T >

BKSolver< T >::BKSolver ( int  var_num_max, int  edge_num_max  )

inline

Constructs the solver with pre-allocated capacity.

Parameters

var_num_max	Maximum number of binary variables expected.
edge_num_max	Maximum number of edges expected (used for memory pre-allocation).

~BKSolver()

template<typename T >

BKSolver< T >::~BKSolver ( )

inlineoverride

Member Function Documentation

add_constant()

template<typename T >

void BKSolver< T >::add_constant ( T  E )

inlineoverridevirtual

Adds a constant term to the energy function.

Parameters

E	Constant to add.

Implements MaxFlowSolver< T >.

add_term1()

template<typename T >

void BKSolver< T >::add_term1 ( typename MaxFlowSolver< T >::Var  x, T  E0, T  E1  )

inlineoverride

add_term2()

template<typename T >

void BKSolver< T >::add_term2 ( typename MaxFlowSolver< T >::Var  x, typename MaxFlowSolver< T >::Var  y, T  E00, T  E01, T  E10, T  E11  )

inlineoverride

add_variable()

template<typename T >

MaxFlowSolver< T >::Var BKSolver< T >::add_variable ( )

inlineoverridevirtual

Introduces a new binary variable and returns its handle.

Implements MaxFlowSolver< T >.

get_var()

template<typename T >

int BKSolver< T >::get_var ( typename MaxFlowSolver< T >::Var  x )

inlineoverride

minimize()

template<typename T >

T BKSolver< T >::minimize ( )

inlineoverridevirtual

Minimizes the energy and returns the minimum value.

Implements MaxFlowSolver< T >.

---

The documentation for this class was generated from the following file:

• src/solvers/BKSolver.hpp