#include <BaseGrammarHypothesis.h>

Inheritance diagram for BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >:

Collaboration diagram for BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >:

Public Types
using	HYP = _HYP

using	Predict_t = _Predict_t

using	LL_t = std::unordered_map< typename datum_t::data_t *, std::vector< Vector > >

Public Types inherited from Bayesable< Args... >
typedef Args...	datum_t

typedef std::vector< Args... >	data_t

Public Member Functions
	BaseGrammarHypothesis ()

	BaseGrammarHypothesis (std::vector< HYP > &hypotheses, const data_t *human_data)

void	set_decay (const ExponentialVariable &ev)

void	set_decay_untransformed (double v)

void	set_alpha (const UniformVariable &a)

void	set_alpha_untransformed (double v)

void	copy_parameters (const BaseGrammarHypothesis &h)

float	get_alpha () const

float	get_decay () const

size_t	ndata () const

virtual void	set_hypotheses_and_data (std::vector< HYP > &hypotheses, const data_t &human_data)
	This is the primary function for setting hypothese and data on construction. More...

virtual void	set_can_propose (size_t i, bool b)
	Set whether I can propose to a value in logA – this is handled by VectorNormalHypothesis Here, though, we warn if the value is not 1.0. More...

void	set_flat_prior (bool fp)

virtual size_t	nhypotheses () const
	A convenient function that uses C to say how many hypotheses. More...

virtual void	recompute_C (const std::vector< HYP > &hypotheses)
	Computes our matrix C[h,r] of hypotheses (rows) by counts of each grammar rule. (requires that each hypothesis use the same grammar) More...

virtual void	recompute_LL (std::vector< HYP > &hypotheses, const data_t &human_data)
	Recompute LL[h,di] a hypothesis from each hypothesis and data point to a vector of prior responses. (We need the vector instead of just the sum to implement memory decay. More...

virtual void	recompute_decayedLikelihood (const data_t &human_data)
	Recomputes the decayed likelihood (e.g. at the given decay level, the total ll for each data point. NOTE: This does create a new decayed likelihood each time we compute. More...

virtual void	recompute_P (std::vector< HYP > &hypotheses, const data_t &human_data)=0
	Recompute the predictions for the hypotheses and data. More...

virtual std::map< typename HYP::output_t, double >	compute_model_predictions (const data_t &human_data, const size_t i, const Matrix &hposterior) const =0
	This uses hposterior (computed via this->compute_normalized_posterior()) to compute the model predictions on a the i'th human data point. To do this, we marginalize over hypotheses, computing the weighted sum of outputs. More...

virtual const HYP &	computeMAP (const size_t di, const Matrix &hposterior) const

virtual double	human_chance_lp (const typename datum_t::output_t &r, const datum_t &hd) const
	Get the chance probability of response r in hd (i.e. of a human response). This may typically be pretty boring (like just hd.chance) but w we need to be able to overwrite it. More...

virtual double	compute_prior () override

virtual Matrix	compute_normalized_posterior () const
	This returns a matrix hposterior[h,di] giving the posterior on the h'th element. NOTE: not output is NOT logged. More...

virtual double	compute_likelihood (const data_t &human_data, const double breakout=-infinity) override
	This computes the likelihood of the human data. More...

virtual this_t	restart () const override

virtual std::optional< std::pair< this_t, double > >	propose () const override
	Propose to the hypothesis. The sometimes does grammar parameters and sometimes other parameters, which are all stored as VectorHypotheses. It is sure to call the recompute functions when necessary. More...

virtual Vector	hypothesis_prior () const
	Compute a vector of the prior (one for each hypothesis) using the given counts matrix (hypotheses x rules), AT the specified temperature. More...

virtual bool	operator== (const this_t &h) const override

virtual std::string	string (std::string prefix="") const override
	This returns a string for this hypothesis. Defaulty, now, just in tidy format with all the parameters, one on each row. Note these parameters are shown after transformation (e.g. the prior parameters are NOT logged) More...

virtual void	show (std::string prefix="") override
	Need to override print since it will print in a different format. More...

virtual size_t	hash () const override

virtual std::string	serialize () const override

Public Member Functions inherited from MCMCable< this_t, datum_t, data_t >
	MCMCable ()

virtual bool	operator!= (const this_t &h) const

Public Member Functions inherited from Bayesable< Args... >
	Bayesable ()

virtual double	compute_single_likelihood (const datum_t &datum)
	Compute the likelihood of a single data point. More...

virtual void	clear_bayes ()

virtual double	compute_tempered_likelihood (const data_t &data, int ladder_rank, const double breakout=-infinity)

virtual double	compute_posterior (const data_t &data, const std::pair< double, double > breakoutpair=std::make_pair(-infinity, 1.0))
	Compute the posterior, by calling prior and likelihood. This includes only a little bit of fanciness, which is that if our prior is -inf, then we don't both computing the likelihood. More...

virtual double	at_temperature (double t) const

auto	operator (const Bayesable< datum_t, data_t > &other) const

Static Public Member Functions
static this_t	sample (std::vector< HYP > &hypotheses, const data_t *human_data)

static this_t	deserialize (const std::string &s)

Static Public Member Functions inherited from MCMCable< this_t, datum_t, data_t >
static this_t	sample ()
	Static function for making a hypothesis. Be careful using this with references because they may not foward right (for reasons that are unclear to me) More...

Static Public Member Functions inherited from Serializable< this_t >
static this_t	deserialize (const std::string &)

Public Attributes
HYP::Grammar_t *	grammar

bool	flat_prior

VectorHalfNormalHypothesis	logA

std::shared_ptr< Matrix >	C

std::shared_ptr< LL_t >	LL

std::shared_ptr< Predict_t >	P

std::shared_ptr< Matrix >	decayedLikelihood

const data_t *	which_data

std::vector< HYP > *	which_hypotheses

Public Attributes inherited from Bayesable< Args... >
double	prior

double	likelihood

double	posterior

uintmax_t	born

size_t	born_chain_idx

Protected Attributes
UniformVariable	alpha

ExponentialVariable	decay

Detailed Description

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>
class BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >

Author: piantado

Date: 02/08/20

Member Typedef Documentation

◆ HYP

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

using BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::HYP = _HYP

◆ LL_t

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

using BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::LL_t = std::unordered_map<typename datum_t::data_t*, std::vector<Vector> >

◆ Predict_t

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

using BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::Predict_t = _Predict_t

Constructor & Destructor Documentation

◆ BaseGrammarHypothesis() [1/2]

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::BaseGrammarHypothesis ( )

inline

◆ BaseGrammarHypothesis() [2/2]

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::BaseGrammarHypothesis	(	std::vector< HYP > &	hypotheses,
		const data_t *	human_data
	)

inline

Member Function Documentation

◆ compute_likelihood()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual double BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::compute_likelihood	(	const data_t &	human_data,
		const double	breakout = `-infinity`
	)

inlineoverridevirtual

This computes the likelihood of the human data.

Parameters

data
breakout

Returns

Reimplemented from Bayesable< Args... >.

◆ compute_model_predictions()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual std::map<typename HYP::output_t, double> BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::compute_model_predictions	(	const data_t &	human_data,
		const size_t	i,
		const Matrix &	hposterior
	)		const

pure virtual

This uses hposterior (computed via this->compute_normalized_posterior()) to compute the model predictions on a the i'th human data point. To do this, we marginalize over hypotheses, computing the weighted sum of outputs.

Parameters

hd
hposterior

Implemented in ThunkGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >, ThunkGrammarHypothesis< MyGrammarHypothesis, MyHypothesis, MyHumanDatum, std::vector< MyHumanDatum >, Vector2D< DiscreteDistribution< S > > >, FullGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >, DeterministicGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >, DeterministicGrammarHypothesis< MyGrammarHypothesis, MyHypothesis, MyHumanDatum >, and DeterministicGrammarHypothesis< MyGrammarHypothesis, MyHypothesis >.

◆ compute_normalized_posterior()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual Matrix BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::compute_normalized_posterior ( ) const

inlinevirtual

This returns a matrix hposterior[h,di] giving the posterior on the h'th element. NOTE: not output is NOT logged.

Returns

◆ compute_prior()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual double BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::compute_prior ( )

inlineoverridevirtual

Implements Bayesable< Args... >.

◆ computeMAP()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual const HYP& BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::computeMAP	(	const size_t	di,
		const Matrix &	hposterior
	)		const

inlinevirtual

◆ copy_parameters()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::copy_parameters ( const BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t > & h )

inline

◆ deserialize()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

static this_t BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::deserialize ( const std::string & s )

inlinestatic

◆ get_alpha()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

float BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::get_alpha ( ) const

inline

◆ get_decay()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

float BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::get_decay ( ) const

inline

◆ hash()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual size_t BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::hash ( ) const

inlineoverridevirtual

Implements Bayesable< Args... >.

◆ human_chance_lp()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual double BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::human_chance_lp	(	const typename datum_t::output_t &	r,
		const datum_t &	hd
	)		const

inlinevirtual

Get the chance probability of response r in hd (i.e. of a human response). This may typically be pretty boring (like just hd.chance) but w we need to be able to overwrite it.

Parameters

hd

Returns

◆ hypothesis_prior()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual Vector BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::hypothesis_prior ( ) const

inlinevirtual

Compute a vector of the prior (one for each hypothesis) using the given counts matrix (hypotheses x rules), AT the specified temperature.

Parameters

◆ ndata()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

size_t BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::ndata ( ) const

inline

◆ nhypotheses()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual size_t BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::nhypotheses ( ) const

inlinevirtual

A convenient function that uses C to say how many hypotheses.

Returns

◆ operator==()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual bool BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::operator== ( const this_t & h ) const

inlineoverridevirtual

Implements MCMCable< this_t, datum_t, data_t >.

◆ propose()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual std::optional<std::pair<this_t,double> > BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::propose ( ) const

inlineoverridevirtual

Propose to the hypothesis. The sometimes does grammar parameters and sometimes other parameters, which are all stored as VectorHypotheses. It is sure to call the recompute functions when necessary.

Parameters

data
breakout

Returns

Implements MCMCable< this_t, datum_t, data_t >.

◆ recompute_C()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::recompute_C ( const std::vector< HYP > & hypotheses )

inlinevirtual

Computes our matrix C[h,r] of hypotheses (rows) by counts of each grammar rule. (requires that each hypothesis use the same grammar)

Parameters

hypotheses

◆ recompute_decayedLikelihood()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::recompute_decayedLikelihood ( const data_t & human_data )

inlinevirtual

Recomputes the decayed likelihood (e.g. at the given decay level, the total ll for each data point. NOTE: This does create a new decayed likelihood each time we compute.

Parameters

hypotheses
human_data

Returns

◆ recompute_LL()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::recompute_LL	(	std::vector< HYP > &	hypotheses,
		const data_t &	human_data
	)

inlinevirtual

Recompute LL[h,di] a hypothesis from each hypothesis and data point to a vector of prior responses. (We need the vector instead of just the sum to implement memory decay.

Parameters

hypotheses
human_data

Returns

Reimplemented in ThunkGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >, and ThunkGrammarHypothesis< MyGrammarHypothesis, MyHypothesis, MyHumanDatum, std::vector< MyHumanDatum >, Vector2D< DiscreteDistribution< S > > >.

◆ recompute_P()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::recompute_P	(	std::vector< HYP > &	hypotheses,
		const data_t &	human_data
	)

pure virtual

Recompute the predictions for the hypotheses and data.

Parameters

hypotheses
human_data

Returns

Implemented in ThunkGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >, ThunkGrammarHypothesis< MyGrammarHypothesis, MyHypothesis, MyHumanDatum, std::vector< MyHumanDatum >, Vector2D< DiscreteDistribution< S > > >, FullGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >, DeterministicGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >, DeterministicGrammarHypothesis< MyGrammarHypothesis, MyHypothesis, MyHumanDatum >, and DeterministicGrammarHypothesis< MyGrammarHypothesis, MyHypothesis >.

◆ restart()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual this_t BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::restart ( ) const

inlineoverridevirtual

Implements MCMCable< this_t, datum_t, data_t >.

◆ sample()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

static this_t BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::sample	(	std::vector< HYP > &	hypotheses,
		const data_t *	human_data
	)

inlinestatic

◆ serialize()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual std::string BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::serialize ( ) const

inlineoverridevirtual

Implements Serializable< this_t >.

◆ set_alpha()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::set_alpha ( const UniformVariable & a )

inline

◆ set_alpha_untransformed()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::set_alpha_untransformed ( double v )

inline

◆ set_can_propose()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::set_can_propose	(	size_t	i,
		bool	b
	)

inlinevirtual

Set whether I can propose to a value in logA – this is handled by VectorNormalHypothesis Here, though, we warn if the value is not 1.0.

Parameters

i
b

◆ set_decay()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::set_decay ( const ExponentialVariable & ev )

inline

◆ set_decay_untransformed()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::set_decay_untransformed ( double v )

inline

◆ set_flat_prior()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::set_flat_prior ( bool fp )

inline

◆ set_hypotheses_and_data()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::set_hypotheses_and_data	(	std::vector< HYP > &	hypotheses,
		const data_t &	human_data
	)

inlinevirtual

This is the primary function for setting hypothese and data on construction.

Parameters

hypotheses
human_data

◆ show()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual void BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::show ( std::string prefix = "" )

inlineoverridevirtual

Need to override print since it will print in a different format.

Returns

Reimplemented from Bayesable< Args... >.

◆ string()

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

virtual std::string BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::string ( std::string prefix = "" ) const

inlineoverridevirtual

This returns a string for this hypothesis. Defaulty, now, just in tidy format with all the parameters, one on each row. Note these parameters are shown after transformation (e.g. the prior parameters are NOT logged)

Returns

Implements Bayesable< Args... >.

Member Data Documentation

◆ alpha

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

UniformVariable BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::alpha

protected

◆ C

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

std::shared_ptr<Matrix> BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::C

◆ decay

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

ExponentialVariable BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::decay

protected

◆ decayedLikelihood

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

std::shared_ptr<Matrix> BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::decayedLikelihood

◆ flat_prior

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

bool BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::flat_prior

◆ grammar

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

HYP::Grammar_t* BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::grammar

◆ LL

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

std::shared_ptr<LL_t> BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::LL

◆ logA

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

VectorHalfNormalHypothesis BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::logA

◆ P

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

std::shared_ptr<Predict_t> BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::P

◆ which_data

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

const data_t* BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::which_data

◆ which_hypotheses

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>

std::vector<HYP>* BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >::which_hypotheses

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

src/Hypotheses/GrammarHypotheses/BaseGrammarHypothesis.h

Public Types

Public Member Functions

Static Public Member Functions

Public Attributes

Protected Attributes

Detailed Description

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>> class BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >

Member Typedef Documentation

◆ HYP

◆ LL_t

◆ Predict_t

Constructor & Destructor Documentation

◆ BaseGrammarHypothesis() [1/2]

◆ BaseGrammarHypothesis() [2/2]

Member Function Documentation

◆ compute_likelihood()

◆ compute_model_predictions()

◆ compute_normalized_posterior()

◆ compute_prior()

◆ computeMAP()

◆ copy_parameters()

◆ deserialize()

◆ get_alpha()

◆ get_decay()

◆ hash()

◆ human_chance_lp()

◆ hypothesis_prior()

◆ ndata()

◆ nhypotheses()

◆ operator==()

◆ propose()

◆ recompute_C()

◆ recompute_decayedLikelihood()

◆ recompute_LL()

◆ recompute_P()

◆ restart()

◆ sample()

◆ serialize()

◆ set_alpha()

◆ set_alpha_untransformed()

◆ set_can_propose()

◆ set_decay()

◆ set_decay_untransformed()

◆ set_flat_prior()

◆ set_hypotheses_and_data()

◆ show()

◆ string()

Member Data Documentation

◆ alpha

◆ C

◆ decay

◆ decayedLikelihood

◆ flat_prior

◆ grammar

◆ LL

◆ logA

◆ P

◆ which_data

◆ which_hypotheses

template<typename this_t, typename _HYP, typename datum_t = HumanDatum<_HYP>, typename data_t = std::vector<datum_t>, typename _Predict_t = Vector2D<DiscreteDistribution<typename _HYP::output_t>>>
class BaseGrammarHypothesis< this_t, _HYP, datum_t, data_t, _Predict_t >