LinearSolver Class Reference

External linear solver wrapper. More...

#include <LinearSolver.h>

Public Member Functions
	LinearSolver (AlternativesPerformance &ap, Config &conf, float delta=0.000001, std::string solver="GLOP")
void	initializeSolver ()
void	updateConstraints (std::vector< std::vector< std::vector< bool >>> x_matrix, std::vector< std::vector< std::vector< bool >>> y_matrix)
std::pair< float, std::vector< float > >	solve (std::vector< std::vector< std::vector< bool >>> x_matrix, std::vector< std::vector< std::vector< bool >>> y_matrix)
AlternativesPerformance &	getAlternativesPerformance () const
operations_research::MPSolver *	getSolver () const
std::vector< operations_research::MPVariable * >	getWeights () const
std::vector< operations_research::MPVariable * >	getXa () const
std::vector< operations_research::MPVariable * >	getXap () const
std::vector< operations_research::MPVariable * >	getYa () const
std::vector< operations_research::MPVariable * >	getYap () const
Config	getConf () const

Detailed Description

External linear solver wrapper.

The linear solver class is a wrapper that makes an abstraction between the weight updater and the resolution of the linear problem. Currently this wrapper uses ortools interface and only the ortools solver (GLOP) is implemented. Other solver could be used by following the instruction on how to install them for ortools and changing the initialisation of the solver to use the right one, however the implementation of the resolution shouldn't have to be changed.

The abstraction was made by receiving the constraint matrices according to the equations described in the thesis and returning the solution found for the new weights and lambda.

Link to ortools (google) : https://github.com/google/or-tools

Constructor & Destructor Documentation

LinearSolver()

LinearSolver::LinearSolver	(	AlternativesPerformance &	ap,
		Config &	conf,
		float	delta = 0.000001,
		std::string	solver = "GLOP"
	)

LinearSolver standard constructor.

Parameters

ap	AlternativesPerformance objet that represents the dataset of the problem
config	config setup from the app
solver	solver type to use. Currently only "GLOP" is implemented

Member Function Documentation

getAlternativesPerformance()

AlternativesPerformance& LinearSolver::getAlternativesPerformance

(

)

const

getAlternativesPerformance getter of the alternative performance

Returns

ap

getConf()

Config LinearSolver::getConf

(

)

const

getConf getter of the conf

Returns

conf

getSolver()

operations_research::MPSolver* LinearSolver::getSolver

(

)

const

getSolver getter of the solver

Returns

solver

getWeights()

std::vector<operations_research::MPVariable *> LinearSolver::getWeights

(

)

const

getWeights getter of the weights

Returns

weights

getXa()

std::vector<operations_research::MPVariable *> LinearSolver::getXa

(

)

const

getXa getter of the x_a variables

Returns

x_a

getXap()

std::vector<operations_research::MPVariable *> LinearSolver::getXap

(

)

const

getXap getter of the x_ap variables

Returns

x_ap

getYa()

std::vector<operations_research::MPVariable *> LinearSolver::getYa

(

)

const

getYa getter of the y_a variables

Returns

y_a

getYap()

std::vector<operations_research::MPVariable *> LinearSolver::getYap

(

)

const

getYap getter of the y_ap variables

Returns

y_ap

initializeSolver()

void LinearSolver::initializeSolver

(

)

initializeSolver Initialise the solver given the alternative performance (dataset): add variables and constraints that are not changing given a particular model.

solve()

std::pair<float, std::vector<float> > LinearSolver::solve	(	std::vector< std::vector< std::vector< bool >>>	x_matrix,
		std::vector< std::vector< std::vector< bool >>>	y_matrix
	)

solve Solve the linear problem given the constraint matrix.

Parameters

x_matrix	matrix recapitulating the constraints to add to the linear problem for the x variables
y_matrix	matrix recapitulating the constraints to add to the linear problem for the y variables

Returns

results contained in a pair of (lambda, vector of weights)

updateConstraints()

void LinearSolver::updateConstraints	(	std::vector< std::vector< std::vector< bool >>>	x_matrix,
		std::vector< std::vector< std::vector< bool >>>	y_matrix
	)

updateConstraints reset the previous constraints and add the new ones given by the matrixes

Parameters

x_matrix	matrix recapitulating the constraints to add to the linear problem for the x variables
y_matrix	matrix recapitulating the constraints to add to the linear problem for the y variables

---

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

• /Users/ThomasVindard 1/Desktop/3A/Projet/fastPL/include/learning/LinearSolver.h