Constant data structure for MH random walk tasks with a value histogram stats collector. More...

#include <tomographer/mhrw_valuehist_tools.h>

Inheritance diagram for Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >:

Collaboration diagram for Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >:

Public Types
typedef MHRWTasks::CDataBase< MHWalkerParams_, IterCountIntType_, RngSeedType_ >	Base
	The MHRWTasks::CDataBase base class.

typedef Base::MHWalkerParams	MHWalkerParams
	The MHWalkerParams required for our MHWalker (for instance a MHWalkerParamsStepSize)

typedef Base::RngSeedType	RngSeedType
	Type of the seed for the pseudo-random number generator.

typedef Base::IterCountIntType	IterCountIntType
	The integer type which serves to count the number of iterations (see MHRWParams)

typedef HistCountIntType_	HistCountIntType
	The integer counting type in our underlying raw histogram type.

typedef ValueCalculator_	ValueCalculator
	The class which calculates the value we are collecting a histogram of (ValueCalculator Interface compliant)

typedef CountRealType_	CountRealType
	The real type which serves to average histogram counts (typically `double`)

typedef tomo_internal::valuehist_types< CDataBase, UseBinningAnalysis >::ValueStatsCollectorResultType	ValueStatsCollectorResultType
	The result type of our stats collector. More...

typedef tomo_internal::valuehist_types< CDataBase, UseBinningAnalysis >::MHRWStatsResultsBaseType	MHRWStatsResultsBaseType
	Stores result of the stats collector. May serve as base class for your own MHRWStatsResults class. More...

typedef tomo_internal::valuehist_types< CDataBase, UseBinningAnalysis >::HistogramType	HistogramType
	The histogram type reported by the task. More...

typedef tomo_internal::valuehist_types< CDataBase, UseBinningAnalysis >::HistogramParams	HistogramParams
	The appropriate parameters type for the histogram reported by the task.

typedef MHRWParams< MHWalkerParams, IterCountIntType >	MHRWParamsType
	Type for the parameters of the random walk.

typedef tomo_internal::valuehist_types< CDataBase, UseBinningAnalysis >::AggregatedHistogramType	AggregatedHistogramType

Public Types inherited from Tomographer::MHRWTasks::CDataBase< MHWalkerParams_, IterCountIntType_, RngSeedType_ >
typedef IterCountIntType_	IterCountIntType
	Type used to count the number of iterations.

typedef MHWalkerParams_	MHWalkerParams
	Type used to specify the step size.

typedef RngSeedType_	RngSeedType
	Type used to specify the seed of the random number generator. More...

typedef MHRWParams< MHWalkerParams, IterCountIntType >	MHRWParamsType
	Type to store the parameters of the Metropolis-Hastings random walk (number of runs, sweep size, etc.) More...

Public Member Functions
template<ENABLED_IF( !UseBinningAnalysis) >
	CDataBase (const ValueCalculator &valcalc_, HistogramParams histogram_params_, MHRWParamsType p, RngSeedType base_seed=0)
	Constructor (use only without binning analysis)

template<ENABLED_IF( !UseBinningAnalysis) >
	CDataBase (const ValueCalculator &valcalc_, HistogramParams histogram_params_, MHRWParamsType p, std::vector< RngSeedType > seeds)
	Constructor (use only without binning analysis), with full list of rng seeds.

template<ENABLED_IF( UseBinningAnalysis) >
	CDataBase (const ValueCalculator &valcalc_, HistogramParams histogram_params_, int binning_num_levels_, MHRWParamsType p, RngSeedType base_seed=0)
	Constructor (use only with binning analysis)

template<ENABLED_IF( UseBinningAnalysis) >
	CDataBase (const ValueCalculator &valcalc_, HistogramParams histogram_params_, int binning_num_levels_, MHRWParamsType p, std::vector< RngSeedType > seeds)
	Constructor (use only with binning analysis), with full list of rng seeds.

template<ENABLED_IF( std::is_default_constructible< ValueCalculator >::value) >
	CDataBase ()
	Construct an invalid object – ONLY for use with Boost.serialization.

template<typename LoggerType , ENABLED_IF( !UseBinningAnalysis) >
ValueHistogramMHRWStatsCollector< ValueCalculator, LoggerType, HistogramType >	createValueStatsCollector (LoggerType &logger) const
	Create the stats collector (without binning analysis) More...

template<typename LoggerType , ENABLED_IF( UseBinningAnalysis) >
ValueHistogramWithBinningMHRWStatsCollector< typename tomo_internal::valuehist_types< CDataBase, true >::BinningMHRWStatsCollectorParams, LoggerType >	createValueStatsCollector (LoggerType &logger) const
	Create the stats collector (with binning analysis) More...

template<typename TaskResultType >
AggregatedHistogramType	aggregateResultHistograms (const std::vector< TaskResultType *> &task_result_list)
	Convenience function for aggregating histograms resulting from value-histogram tasks. More...

Public Member Functions inherited from Tomographer::MHRWTasks::CDataBase< MHWalkerParams_, IterCountIntType_, RngSeedType_ >
template<typename MHRWParamsType >
	CDataBase (MHRWParamsType &&p, RngSeedType base_seed_)
	Constructor. More...

template<typename MHRWParamsType >
	CDataBase (MHRWParamsType &&p, std::vector< RngSeedType > task_seeds_)
	Constructor. More...

	CDataBase ()
	Construct an invalid object – ONLY for use with Boost.serialization.

template<typename TaskNoCountIntType >
RngSeedType	getTaskInput (TaskNoCountIntType k) const
	Returns a random seed to seed the random number generator with for run number k. More...

void	printBasicCDataMHRWInfo (std::ostream &str) const
	Get some human-readable info about the random walk as a string. More...

std::string	getBasicCDataMHRWInfo () const
	Get some human-readable info about the random walk as a string, see printBasicCDataMHRWInfo()

Public Attributes
ValueCalculator	valcalc
	The value calculator instance. More...

HistogramParams	histogram_params
	The parameters of the histogram that we are collecting. More...

Tools::StoreIfEnabled< int, UseBinningAnalysis >	binningNumLevels
	The number of levels in the binning analysis (only if we are using a binning analysis) More...

Public Attributes inherited from Tomographer::MHRWTasks::CDataBase< MHWalkerParams_, IterCountIntType_, RngSeedType_ >
MHRWParamsType	mhrw_params
	Parameters of the random walk. More...

RngSeedType	base_seed
	A base random seed from which each run seed will be derived. More...

std::vector< RngSeedType >	task_seeds
	A list of random seeds to use for each task. More...

Static Public Attributes
static constexpr bool	UseBinningAnalysis = UseBinningAnalysis_
	Whether or not we are to use a binning analysis for calculating the error bars.

Detailed Description

template<typename ValueCalculator_, bool UseBinningAnalysis_ = true, typename MHWalkerParams_ = MHWalkerParamsStepSize<double>, typename RngSeedType_ = std::mt19937::result_type, typename IterCountIntType_ = int, typename CountRealType_ = double, typename HistCountIntType_ = IterCountIntType_>
struct Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >

Constant data structure for MH random walk tasks with a value histogram stats collector.

You should define your TaskCData class to inherit this class, and be compliant with the MHRandomWalkTaskCData Interface. This class already inherits MHRWTasks::CDataBase, so you only have to inherit this class and provide a createMHWalker() method (see MHRandomWalkTaskCData Interface). (Note that your class will automatically be compliant with the TaskCData Interface in this way.)

Look at the file test/minimal_tomorun.cxx for an example of you to define your cdata class.

We provide some useful typedefs, as well as a createValueStatsCollector() which should be used by subclasses in the setupRandomWalkAndRun() method of their MHRandomWalkTaskCData Interface implementation.

Since: Changed in Tomographer 5.0: Removed StepRealType template parameter, added MHWalkerParams, beware the new order!

Template Parameters

ValueCalculator	the value calculator type you wish to use; defining which value you are interested in collecting a histogram of during the random walk;
UseBinningAnalysis	whether or not to use a binning analysis to obtain reliable error bars during the random walk.
MHWalkerParams	The MHWalkerParams required for our MHWalker (for instance a MHWalkerParamsStepSize)
RngSeedType	The type used to store the seed of the pseudo random number generator. See MHRWTasks::CDataBase.
IterCountIntType	the integer type to use for counting iterations.
CountRealType	the real type to use when calculating the scaled histogram with error bars.
HistCountIntType	the integer type to record the histogram counts in the underlying raw histogram.

Note: if your subclass also takes a template parameter to allow or not the use of a binning analysis, then you'll have to define two conditionally enabled constructors to cover both cases. For instance (see also the code in tomorun/tomorun_dispatch.h for another example):

struct MyCData
  : public Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ... >
{
  typedef Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ... > Base;
  TOMOGRAPHER_ENABLED_IF(!BinningAnalysisEnabled)
  MyCData(ValueCalculator valcalc, RngSeedType base_seed, ...)
    : Base(valcalc,
           typename Base::HistogramParams(...),
           typename Base::MHRWParamsType(...),
           base_seed)
  { ... }
  TOMOGRAPHER_ENABLED_IF(BinningAnalysisEnabled)
  MyCData(ValueCalculator valcalc, RngSeedType base_seed, int binning_analysis_num_levels, ...)
    : Base(valcalc,
           typename Base::HistogramParams(...),
           binning_analysis_num_levels,
           typename Base::MHRWParamsType(...),
           base_seed)
  { ... }
  ...
};

Since: Since Tomographer 5.3, this class can be serialized with Boost.Serialization.

Definition at line 240 of file mhrw_valuehist_tools.h.

Member Typedef Documentation

§ HistogramType

template<typename ValueCalculator_ , bool UseBinningAnalysis_ = true, typename MHWalkerParams_ = MHWalkerParamsStepSize<double>, typename RngSeedType_ = std::mt19937::result_type, typename IterCountIntType_ = int, typename CountRealType_ = double, typename HistCountIntType_ = IterCountIntType_>

typedef tomo_internal::valuehist_types<CDataBase,UseBinningAnalysis>::HistogramType Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >::HistogramType

The histogram type reported by the task.

Depending on whether we are using a binning analysis, this histogram type will have error bars.

It is either:

(no binning analysis): a Histogram of appropriate parameters type
(with binning analysis): a HistogramWithErrorBars of appropriate parameters types. This is in fact ValueHistogramWithBinningMHRWStatsCollectorParams::HistogramType.

Definition at line 309 of file mhrw_valuehist_tools.h.

§ MHRWStatsResultsBaseType

template<typename ValueCalculator_ , bool UseBinningAnalysis_ = true, typename MHWalkerParams_ = MHWalkerParamsStepSize<double>, typename RngSeedType_ = std::mt19937::result_type, typename IterCountIntType_ = int, typename CountRealType_ = double, typename HistCountIntType_ = IterCountIntType_>

typedef tomo_internal::valuehist_types<CDataBase,UseBinningAnalysis>::MHRWStatsResultsBaseType Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >::MHRWStatsResultsBaseType

Stores result of the stats collector. May serve as base class for your own MHRWStatsResults class.

Depending on UseBinningAnalysis, this is either MHRWStatsResultsBaseSimple, or directly the corresponding ValueHistogramWithBinningMHRWStatsCollectorResult type.

Remember, this class takes a single argument in the constructor which is a rvalue-reference to the result of the value stats collector.

Definition at line 293 of file mhrw_valuehist_tools.h.

§ ValueStatsCollectorResultType

template<typename ValueCalculator_ , bool UseBinningAnalysis_ = true, typename MHWalkerParams_ = MHWalkerParamsStepSize<double>, typename RngSeedType_ = std::mt19937::result_type, typename IterCountIntType_ = int, typename CountRealType_ = double, typename HistCountIntType_ = IterCountIntType_>

typedef tomo_internal::valuehist_types<CDataBase,UseBinningAnalysis>::ValueStatsCollectorResultType Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >::ValueStatsCollectorResultType

The result type of our stats collector.

This is either:

(without binning analysis): a simple histogram type, without error bars
(with binning analysis): a structure containing a histogram with error bars, along with a convergence analysis of the errors. It is a ValueHistogramWithBinningMHRWStatsCollectorResult type.

Definition at line 280 of file mhrw_valuehist_tools.h.

Member Function Documentation

§ aggregateResultHistograms()

template<typename ValueCalculator_ , bool UseBinningAnalysis_ = true, typename MHWalkerParams_ = MHWalkerParamsStepSize<double>, typename RngSeedType_ = std::mt19937::result_type, typename IterCountIntType_ = int, typename CountRealType_ = double, typename HistCountIntType_ = IterCountIntType_>

template<typename TaskResultType >

AggregatedHistogramType Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >::aggregateResultHistograms ( const std::vector< TaskResultType *> & task_result_list )

inline

Convenience function for aggregating histograms resulting from value-histogram tasks.

The TaskResultType is expected to be a subclass of MHRWStatsResultsBaseSimple. This will automatically be the case if you're using our CDataBase type, and if you have defined your custom MHRWStatsResults type (see MHRandomWalkTaskCData Interface) using CDataBase::MHRWStatsResultsBaseType as base class.

Definition at line 441 of file mhrw_valuehist_tools.h.

§ createValueStatsCollector() [1/2]

template<typename ValueCalculator_ , bool UseBinningAnalysis_ = true, typename MHWalkerParams_ = MHWalkerParamsStepSize<double>, typename RngSeedType_ = std::mt19937::result_type, typename IterCountIntType_ = int, typename CountRealType_ = double, typename HistCountIntType_ = IterCountIntType_>

template<typename LoggerType , ENABLED_IF( !UseBinningAnalysis) >

ValueHistogramMHRWStatsCollector<ValueCalculator,LoggerType,HistogramType> Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >::createValueStatsCollector ( LoggerType & logger ) const

inline

Create the stats collector (without binning analysis)

This method is provided so that user-provided CData random walk description classes can easily create the necessary value histogram stats collector for collecting a histogram of values during a MHRandomWalkTask. This function should typically be called from within the user's setupRandomWalkAndRun() function.

Definition at line 393 of file mhrw_valuehist_tools.h.

§ createValueStatsCollector() [2/2]

template<typename ValueCalculator_ , bool UseBinningAnalysis_ = true, typename MHWalkerParams_ = MHWalkerParamsStepSize<double>, typename RngSeedType_ = std::mt19937::result_type, typename IterCountIntType_ = int, typename CountRealType_ = double, typename HistCountIntType_ = IterCountIntType_>

template<typename LoggerType , ENABLED_IF( UseBinningAnalysis) >

ValueHistogramWithBinningMHRWStatsCollector< typename tomo_internal::valuehist_types<CDataBase, true>::BinningMHRWStatsCollectorParams, LoggerType> Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >::createValueStatsCollector ( LoggerType & logger ) const

inline

Create the stats collector (with binning analysis)

This method is provided so that user-provided CData random walk description classes can easily create the necessary value histogram stats collector for collecting a histogram of values during a MHRandomWalkTask. This function should typically be called from within the user's setupRandomWalkAndRun() function.

Definition at line 414 of file mhrw_valuehist_tools.h.

Member Data Documentation

§ binningNumLevels

template<typename ValueCalculator_ , bool UseBinningAnalysis_ = true, typename MHWalkerParams_ = MHWalkerParamsStepSize<double>, typename RngSeedType_ = std::mt19937::result_type, typename IterCountIntType_ = int, typename CountRealType_ = double, typename HistCountIntType_ = IterCountIntType_>

Tools::StoreIfEnabled<int, UseBinningAnalysis> Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >::binningNumLevels

The number of levels in the binning analysis (only if we are using a binning analysis)

Since: In Tomographer 5.3: Removed the const attribute (to make serialization easier; anyway in multiprocessing implementations a const pointer to this class is kept ensuring const-ness already)

Definition at line 380 of file mhrw_valuehist_tools.h.

§ histogram_params

template<typename ValueCalculator_ , bool UseBinningAnalysis_ = true, typename MHWalkerParams_ = MHWalkerParamsStepSize<double>, typename RngSeedType_ = std::mt19937::result_type, typename IterCountIntType_ = int, typename CountRealType_ = double, typename HistCountIntType_ = IterCountIntType_>

HistogramParams Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >::histogram_params

The parameters of the histogram that we are collecting.

Since: In Tomographer 5.3: Removed the const attribute (to make serialization easier; anyway in multiprocessing implementations a const pointer to this class is kept ensuring const-ness already)

Definition at line 372 of file mhrw_valuehist_tools.h.

§ valcalc

template<typename ValueCalculator_ , bool UseBinningAnalysis_ = true, typename MHWalkerParams_ = MHWalkerParamsStepSize<double>, typename RngSeedType_ = std::mt19937::result_type, typename IterCountIntType_ = int, typename CountRealType_ = double, typename HistCountIntType_ = IterCountIntType_>

ValueCalculator Tomographer::MHRWTasks::ValueHistogramTools::CDataBase< ValueCalculator_, UseBinningAnalysis_, MHWalkerParams_, RngSeedType_, IterCountIntType_, CountRealType_, HistCountIntType_ >::valcalc

The value calculator instance.

Since: In Tomographer 5.3: Removed the const attribute (to make serialization easier; anyway in multiprocessing implementations a const pointer to this class is kept ensuring const-ness already)

Definition at line 365 of file mhrw_valuehist_tools.h.

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

tomographer/mhrw_valuehist_tools.h (v5.4)

Public Types

Public Member Functions

Public Attributes

Static Public Attributes

Detailed Description

Member Typedef Documentation

§ HistogramType

§ MHRWStatsResultsBaseType

§ ValueStatsCollectorResultType

Member Function Documentation

§ aggregateResultHistograms()

§ createValueStatsCollector() [1/2]

§ createValueStatsCollector() [2/2]

Member Data Documentation

§ binningNumLevels

§ histogram_params

§ valcalc