Classes
class	EpsilonPolicy

class	Experience
	This class computes averages and counts for a multi-agent cooperative Bandit problem. More...

class	FlattenedModel
	This class flattens a factored bandit model. More...

class	LLRPolicy
	This class represents the Learning with Linear Rewards algorithm. More...

class	LocalSearch
	This class approximately finds the best joint action using Local Search. More...

struct	MakeGraph
	This class is the public interface for initializing the graph in generic code that uses the maximizers. More...

struct	MakeGraph< MaxPlus >

struct	MakeGraph< ReusingIterativeLocalSearch >

struct	MakeGraphImpl
	This class clumps all implementations that create graphs for data for certain Maximizers. More...

struct	MakeGraphImpl< LocalSearch, Iterable >

struct	MakeGraphImpl< LocalSearch, QFunction >

struct	MakeGraphImpl< VariableElimination, Data >

class	MAUCEPolicy
	This class represents the Multi-Agent Upper Confidence Exploration algorithm. More...

class	MaxPlus
	This class represents the Max-Plus optimization algorithm for loopy FactorGraphs. More...

class	MiningBandit
	This class represents the mining bandit problem. More...

class	Model
	This class represents a factored multi-armed bandit. More...

struct	MOQFunctionRule
	This struct represents a single action/values pair. More...

class	MultiObjectiveVariableElimination
	This class represents the Multi Objective Variable Elimination process. More...

class	PolicyInterface
	Simple typedef for most of a normal Bandit's policy needs. More...

struct	QFunctionRule
	This struct represents a single action/value pair. More...

class	QGreedyPolicy
	This class implements a greedy policy through a QFunction. More...

class	RandomPolicy
	This class represents a random policy. More...

class	ReusingIterativeLocalSearch
	This class approximately finds the best joint action with Reusing Iterative Local Search. More...

class	SingleActionPolicy
	This class represents a policy always picking the same action. More...

class	ThompsonSamplingPolicy
	This class implements a Thompson sampling policy. More...

class	UCVE
	This class represents the UCVE process. More...

struct	UpdateGraph
	This class is the public interface for updating the input graph with the input data in generic code that uses the maximizers. More...

struct	UpdateGraph< MaxPlus >

struct	UpdateGraph< ReusingIterativeLocalSearch >

struct	UpdateGraphImpl
	This class clumps all implementations that update graphs with data for certain Maximizers. More...

struct	UpdateGraphImpl< LocalSearch, Iterable >

struct	UpdateGraphImpl< LocalSearch, QFunction >

struct	UpdateGraphImpl< VariableElimination, Iterable >

struct	UpdateGraphImpl< VariableElimination, QFunction >

class	VariableElimination
	This class represents the Variable Elimination algorithm. More...

Typedefs
using	QFunction = FactoredVector
	This represents a factored QFunction. More...

Functions
std::tuple< Action, std::vector< unsigned >, std::vector< double > >	makeMiningParameters (unsigned seed)
	This function generates the parameters for a random MiningBandit. More...

Variables
template<typename QR >
concept	IsQFunctionRule
	This concept models the interface for a QFunctionRule. More...

template<typename T >
concept	QFRuleRange = std::ranges::range<T> && IsQFunctionRule<std::ranges::range_value_t<T>>
	This concept represents a range of QFunctionRules. More...

Typedef Documentation

◆ QFunction

using AIToolbox::Factored::Bandit::QFunction = typedef FactoredVector

This represents a factored QFunction.

Function Documentation

◆ makeMiningParameters()

std::tuple<Action, std::vector<unsigned>, std::vector<double> > AIToolbox::Factored::Bandit::makeMiningParameters ( unsigned seed )

This function generates the parameters for a random MiningBandit.

The parameters are uniformly sampled from the following ranges:

villages: [5, 15]
mines: villages + 3
workers per village: [1, 5]
mines per village: [2, 4]
productivity per mine: [0, 0.5]

We return a tuple containing the parameters for a MiningBandit. In particular:

The action space.
How many workers there are in each village.
The productivity factor for each mine.

Parameters

seed	A random seed for a random generator to sample the parameters.

Returns: A tuple containing parameters for a MiningBandit, in the order of its constructor.

Variable Documentation

◆ IsQFunctionRule

template<typename QR >

concept AIToolbox::Factored::Bandit::IsQFunctionRule

Initial value:

= requires (const QR qr) {
        { qr.action } -> std::convertible_to<PartialAction>;
        { qr.value } -> std::convertible_to<double>;
    }

This concept models the interface for a QFunctionRule.

This is needed so we can consider MDP QFunctionRules as Bandit QFunctionRules, without having to use inheritance to connect them.

◆ QFRuleRange

template<typename T >

concept AIToolbox::Factored::Bandit::QFRuleRange = std::ranges::range<T> && IsQFunctionRule<std::ranges::range_value_t<T>>

This concept represents a range of QFunctionRules.

Classes

Typedefs

Functions