Main Gaia namespace, which contains all the library functions. More...

Namespaces
	convert
	Utility namespace used for various types of conversion.

Classes
class	AddField
	The AddField analyzer allows you to add new fields in a layout of an already existing DataSet. More...

class	AddFieldApplier
	AddFieldApplier applier class. More...

class	Analyzer
	the Analyzer abstract base class. More...

class	Applier
	The Applier abstract base class. More...

class	Array

class	BaseQueryOptimizer
	The QueryOptimizer class tries to optimize a query by reducing the SearchSpace on which it is supposed to operate, mostly by the means of the associated filter. More...

class	BaseResultSet
	This class serves as a ref-counted wrapper for a SearchSpace, which is much more convenient to deal with than having to remember who owns the results or who's responsible for deleting it. More...

class	BaseSearchSpace
	A SearchSpace is a structure dedicated to the task of storing pointers to Point with an associated distance and is the structure on which the nearest-neighbours queries are performed. More...

class	BaseSearchSpacePool
	The SearchSpacePool accomplishes 2 main functions: More...

class	BaseView
	The View class contains a specific view on a dataset. More...

class	BPFDistance
	This class composes the breakpoint function specified by its pairs of (split point, value) with the given distance. More...

class	Center
	The Center analyzer centers the points around the mean of each descriptor. More...

class	Cleaner
	The Cleaner analyzer checks for values that may cause problems when doing mathematical analysis (such as normalize, pca, ...), and removes them. More...

class	CosineAngleDistance
	This class computes the angle in radians between 2 points and divides it by Pi (so that it's normalized between 0 and 1). More...

class	CosineDistance

class	CosineSimilarity
	This class computes the cosine similarity of a given Region. More...

class	CrossSegmentDistance
	This class calculates a distance between segments from two points. More...

class	DataSet
	This class represents a dataset and all related information. More...

class	DescCompare
	Compare the descriptor names using the following criteria: More...

class	DescriptorTree

class	DistAggr

class	DistanceFunction

class	Distribute
	Distribute applier class. More...

class	DotProduct
	DotProduct applier class. More...

class	Enumerate
	The Enumerate analyzer allows you to specify a list of string descriptors which you want to be stored as enumerations. More...

class	EnumerateApplier
	EnumerateApplier applier class. More...

class	EuclideanDistance
	This class computes the euclidean distance of a given Region. More...

class	ExponentialCompressDistance
	This class compresses a distance from [0; inf) into [0; 1) using the following formula: dist = 1 - exp(-alpha*dist), alpha being a compression factor. More...

class	Extract
	The Extract analyzer allows you to extract only specific dimensions from any given number of descriptors. More...

class	ExtractApplier
	Extract applier class. More...

class	Factory
	A basic implementation of a generic factory class. More...

class	Filter
	The Filter class allows to check whether a predicate is true for a given Point. More...

class	FixLength
	The FixLength analyzer allows you to specify a list of descriptors for which you want to fix the length definitely. More...

class	FixLengthApplier
	FixLengthApplier applier class. More...

class	ForceIdentityDistance
	This metric forces the identity condition to be valid on the composed distance by checking first if the 2 points are equal and returning 0 if that is the case. More...

class	FrozenCosineAngleDistance

class	FrozenDataSet
	A FrozenDataSet is a dataset that has been flagged as immutable. More...

class	FrozenDistance

class	FrozenEuclideanDistance

class	FrozenExponentialCompressDistance

class	FrozenLinearCombinationDistance

class	FrozenSearchPoint

class	GaiaException
	Exception class that can take up to 3 arguments of any type, which will be serialized into a QString for the error message. More...

class	GaiaMap
	Map class (derived from QMap) but that throws a custom Exception instead of returning a default constructed value when the key isn't in there. More...

class	Gaussianize
	The Gaussianize analyzer fits each descriptor into a gaussian distribution. More...

class	InnerDim

class	InverseProjection
	The inverse projection Analyzer computes the inverse the previous transformation, given it was a linear projection (such as PCA or RCA). More...

class	Key

class	KeyDistance
	This class computes the harmonic distance between keys. More...

class	KullbackLeiblerDistance
	This class computes the symmetric Kullback-Leibler distance of a given Region. More...

class	LayoutAwareFactory
	A tuned implementation of a factory. More...

class	LinearCombinationDistance
	This class computes a distance that is the linear combination of any number of other DistanceFunctions. More...

class	Logger
	Asynchronous thread-safe logger object. More...

class	ManhattanDistance
	This class computes the Manhattan distance of a given Region. More...

class	MatMult
	MatMult class. More...

class	Merge
	Merge applier class. More...

class	MergeRegion
	The MergeRegion analyzer allows you to merge multiple descriptors into a single one. More...

class	MergeRegionApplier
	MergeRegion applier class. More...

class	MinkowskiDistance
	This class computes the Minkowski distance of a Region. More...

class	Normalize
	The Normalize analyzer normalizes real descriptors. More...

class	NullDistance
	This class computes a distance which is always equal to 0. More...

class	ParamCheckingFactory

class	Parameter

class	ParameterMap

class	PCA
	The PCA Analyzer computes the Principal Components Analysis of the given DataSet using the specified descriptors. More...

class	Point

class	PointArray
	A PointArray is a simple subclass of a QVector<Point*> that owns the memory to the points (hence will delete them when it is itself deleted). More...

class	PointLayout
	This class describes the layout of a point. More...

class	PointLayoutData

class	RandomDistance
	This class computes a distance which returns a random number between 0 and the specified range. More...

class	RCA
	The RCA Analyzer computes the Relevant Components Analysis of the given DataSet using the specified descriptors. More...

class	Region
	A region is a physical location in the point layout which consists in a list of segments. More...

class	Remove
	The Remove analyzer allows you to discard specific descriptors. More...

class	RemoveDesc
	RemoveDesc applier class. More...

class	RemoveVL
	The RemoveVL analyzer checks for descriptors which are variable-length descriptors, such as the list of beats, ticks, chords, etc. and removes them. More...

class	Rename
	The Rename analyzer allows you to rename an existing field. More...

class	RenameApplier
	Rename applier class. More...

class	ResistorAverageDistance
	This class computes the symmetric resistor-average distance of a given Region. More...

class	RhythmDistance
	This class computes the rhythm distance of a given descriptor. More...

class	Scope

class	ScopedData

class	SearchPoint

class	SearchSpaceWrapper

class	Segment
	A segment is a physical contiguous location of descriptors of the same type in the point layout. More...

class	Select
	The Select analyzer allows you to keep only specific descriptors. More...

class	SelectDesc
	SelectDesc applier class. More...

class	SemanticClassifierDistance
	: desc is wrong, still the wpearson one... More...

class	SingleValueMorphable

class	SortNode

class	SortOn

class	Stringifier

class	SummedSegmentDistance
	This class calculates a distance between segments from two points and sums the results. More...

class	SVMPredict
	SVMPredict applier class. More...

class	SVMTrain
	The SVMTrain analyzer creates a SVM model for the given dataset using libsvm. More...

class	TextProgress
	This class is a very simple class which allows you to print a progress percentage on the standard output. More...

class	Timer
	This class allows you to measure a certain amount of time, for instance if you want to know how long the execution of a given function takes. More...

class	TransfoChain

class	Transformation
	Class containing all info pertaining to a specific transformation, ie: the name of the analyzer class, the input parameters, the results of this analysis (= the applier parameters), and the name of the applier class. More...

class	UnscopedData

class	WeightedEuclideanDistance
	This class computes the euclidean distance of a given Region, applying a weight on each descriptor. More...

class	WeightedPearsonDistance
	This class computes the weighted Pearson correlation between 2 points. More...

Typedefs
typedef ParamCheckingFactory< QString, Analyzer, ParameterMap >	AnalyzerFactory

typedef Factory< QString, Applier, Transformation >	ApplierFactory

typedef QMap< Segment, QPair< RealDescriptor, RealDescriptor > >	CoeffMap

typedef QMap< QString, QVariant >	VMap

typedef BaseView< DataSet, Point, SearchPoint, DistanceFunction >	View

typedef SingleValueMorphable< QString >	StringDescriptor

typedef SingleValueMorphable< Real >	RealDescriptor

typedef SingleValueMorphable< Enum >	EnumDescriptor

typedef BaseView< FrozenDataSet, FrozenPoint, FrozenSearchPoint, FrozenDistance >	FrozenView

typedef LayoutAwareFactory< QString, DistanceFunction, PointLayout, ParameterMap >	MetricFactory

typedef MetricFactory	DistanceFunctionFactory

typedef QPair< int, Real >	DimWeight

typedef LayoutAwareFactory< QString, FrozenDistance, FrozenDataSet, ParameterMap >	FrozenMetricFactory

typedef FrozenMetricFactory	FrozenDistanceFunctionFactory

typedef NullDistance	LabelDistance

typedef QMap< Enum, QString >	EnumMap

typedef QMap< QString, Enum >	ReverseEnumMap

typedef BaseQueryOptimizer< SearchPoint, DataSet >	QueryOptimizer

typedef QPair< QString, Real >	Result

typedef QList< Result >	SearchResults

typedef BaseSearchSpace< SearchPoint, DataSet >	SearchSpace

typedef BaseSearchSpace< FrozenSearchPoint, FrozenDataSet >	FrozenSearchSpace

typedef BaseResultSet< SearchPoint, DataSet >	ResultSet

typedef ResultSet	InputSpace

typedef BaseSearchSpacePool< SearchPoint, DataSet >	SearchSpacePool

typedef BaseSearchSpacePool< FrozenSearchPoint, FrozenDataSet >	FrozenSearchSpacePool

typedef std::pair< float, int >	fspoint

typedef float	Real

typedef int	Enum

typedef Eigen::Matrix< float, Eigen::Dynamic, Eigen::Dynamic, Eigen::RowMajor >	MatrixRXf

typedef Eigen::RowVectorXf	FrozenPoint

typedef QList< QPair< Segment, Segment > >	IndexMap

typedef BaseResultSet< FrozenSearchPoint, FrozenDataSet >	FrozenResultSet

Enumerations
enum	DebuggingModule { GMemory = 1 << 0, GAlgorithms = 1 << 1, GMath = 1 << 2, GParser = 1 << 3, GIO = 1 << 4, GPerf = 1 << 5, GTools = 1 << 6, GDataSet = 1 << 10, GDescriptorTree = 1 << 11, GRegion = 1 << 12, GPoint = 1 << 13, GView = 1 << 14, GSearchSpace = 1 << 15, GNone = 0, GAll = (1 << 30) -1 }

enum	DescriptorType { UndefinedType, RealType, StringType, EnumType }
	The possible types of descriptors accepted.

enum	DescriptorLengthType { FixedLength, VariableLength }
	Either fixed-length or variable-length.

enum	UpdateType { UpdateOnUnit, UpdateOnPercent }

enum	BinaryFileFormat { Gaia_2_0 = 100, Gaia_2_1 = 101, Gaia_2_3 = 102 }

Functions
QPair< Point, Point >	getRegionMeanVar (const DataSet &dataset, const Region &region)
	Gets the mean and variance of the descriptors in the given region. More...

QPair< Point, Point >	getRegionMinMax (const DataSet &dataset, const Region &region)
	Gets the mean and variance of the descriptors in the given region. More...

QStringList	findVariableLengthDescriptors (const DataSet *dataset)
	For all variable-length descriptors in this DataSet, find all the ones which actually are of different lengths, ie: those such that there exists 2 points p1 and p2 such as len(p1.desc) != len(p2.desc). More...

void	registerAnalyzers ()
	Registers all the analyzers available inside the AnalyzerFactory. More...

void	registerAppliers ()

Real	totalVariancePercentage (const VectorXd &eigenValues, int targetDimension)

const char *	debugModuleDescription (DebuggingModule module)

void	setDebugLevel (int levels)

void	unsetDebugLevel (int levels)

template<typename T >
SingleValueMorphable< T >	operator- (const T &x, const SingleValueMorphable< T > &y)

template<typename T >
SingleValueMorphable< T >	operator/ (const T &x, const SingleValueMorphable< T > &y)

template<typename T >
SingleValueMorphable< T >	operator+ (const T &x, const SingleValueMorphable< T > &y)

template<typename T >
SingleValueMorphable< T >	operator* (const T &x, const SingleValueMorphable< T > &y)

template<typename T >
Stringifier &	operator<< (Stringifier &out, const SingleValueMorphable< T > &array)

template<>
RealDescriptor	sqrt (const RealDescriptor &x)

void	init ()
	Init function that makes sure that all the factories have correctly registered all the types they can create. More...

void	shutdown ()
	Shutdown function that mostly frees up resources (statically) allocated for having better performance in gaia.

QTextStream &	operator<< (QTextStream &out, const QPair< int, int > &p)

QTextStream &	operator<< (QTextStream &out, const QStringList &slist)

	GAIA_DEFINE_EXCEPTION (FixedLengthException)

double	erfinv (double P)
	Computes the inverse error function (http://en.wikipedia.org/wiki/Error_function).

double	norminv (double P, double mu=0, double sigma=1)
	Computes the inverse of the normal cdf with parameter mu and sigma.

double	chi2inv (double P, int v)
	Computes the inverse of the chi-square cdf with v degrees of freedom.

template<typename T >
bool	isnan (T x)

template<typename T >
bool	isinf (T x)

template<typename T >
T	mean (const T *array, uint n)
	Computes the mean of an array.

template<typename T >
T	mean (const std::vector< T > &v)

template<typename T >
T	variance (const T *array, uint n)
	Computes the variance of an array.

template<typename T >
T	variance (const std::vector< T > &v)

template<typename T >
T	skewness (const T *array, uint n)
	Computes the skewness of an array. More...

template<typename T >
T	skewness (const std::vector< T > &v)

template<typename T >
T	kurtosis (const T *array, uint n)
	Computes the kurtosis of an array. More...

template<typename T >
T	kurtosis (const std::vector< T > &v)

template<typename RandomAccessIterator >
void	sort (RandomAccessIterator first, RandomAccessIterator last)

template<typename RandomAccessIterator , typename StrictWeakOrdering >
void	sort (RandomAccessIterator first, RandomAccessIterator last, StrictWeakOrdering comp)

template<typename Container >
void	sort (Container &container)

template<typename T >
void	hist (const T array, uint n, int n_array, T *x_array, uint n_bins)
	Given a set of values, computes the associated histogram. More...

template<typename T >
T	sgn (T x)

template<typename T >
T	intpow (T x, int n)
	Computes the integral power n of x.

template<template< typename > class Container, typename T >
int	binarySearch (const Container< T > &v, T value)
	Iterative function for binary search (more efficient than recursive one). More...

template<typename T >
T	sqrt (const T &x)

Real	clip (Real value, Real minv, Real maxv)
	Returns the given value clipped inside the specified interval.

template<typename T >
Eigen::Matrix< T, Eigen::Dynamic, Eigen::Dynamic >	covarianceMatrix_tmpl (const PointArray &v, const Region &region)

Eigen::MatrixXf	covarianceMatrixSinglePrecision (const PointArray &v, const Region &region)

Eigen::MatrixXd	covarianceMatrix (const PointArray &v, const Region &region)
	Computes the covariance matrix given a set of points. More...

template<typename T , typename Compare >
void	sortEigenVectors (Eigen::Matrix< T, Eigen::Dynamic, 1 > &values, Eigen::Matrix< T, Eigen::Dynamic, Eigen::Dynamic > &vectors, Compare cmp)
	Sort the eigen vectors and their corresponding values according to the given compare function.

template<typename T >
void	sortEigenVectors (Eigen::Matrix< T, Eigen::Dynamic, 1 > &values, Eigen::Matrix< T, Eigen::Dynamic, Eigen::Dynamic > &vectors)
	Sort the eigen vectors and their corresponding values in ascending order.

void	registerMetrics ()
	Registers all the metrics available inside the MetricFactory. More...

void	registerFrozenMetrics ()

Real	traceMatrixProduct (const Real m1, const Real m2, int size)
	Compute the trace of the matrix product of m1 and m2. More...

Real	traceMatrixProductMean (const Real m1, const Real m2, const Real v1, const Real v2, int size)
	Compute the trace of the product of 2 matrices and vectors using the formula: trace((m1+m2)(v1-v2)(v1-v2)') More...

ParameterMap	toParameterMap (const yaml::Node &node)

bool	isAlwaysTrue (parser::Predicate *pred)
	Returns true if the given predicate is always true (ie: it is a boolean constant which value is True).

bool	isAlwaysFalse (parser::Predicate *pred)
	Returns true if the given predicate is always false (ie: it is a boolean constant which value is False).

QDataStream &	operator<< (QDataStream &out, const ScopedData &s)

QDataStream &	operator>> (QDataStream &in, ScopedData &s)

QString	typeToString (DescriptorType type)

QString	lengthTypeToString (DescriptorLengthType ltype)

yaml::Node	toYaml (const SearchResults &results)

template<typename SearchPointType >
bool	pointerOrderCompare (const SearchPointType &p1, const SearchPointType &p2)

template<>
bool	pointerOrderCompare (const SearchPoint &p1, const SearchPoint &p2)

template<>
bool	pointerOrderCompare (const FrozenSearchPoint &p1, const FrozenSearchPoint &p2)

template<typename SearchPointType , typename DataSetType >
void	copySearchPoints (BaseSearchSpace< SearchPointType, DataSetType > dest, const BaseSearchSpace< SearchPointType, DataSetType > src, int start, int end, int destStart)

template<typename SearchPointType , typename DataSetType >
void	copySearchPoints (BaseSearchSpace< SearchPointType, DataSetType > dest, const BaseSearchSpace< SearchPointType, DataSetType > src, int start, int end)

template<typename SearchPointType , typename DataSetType >
void	copySearchPoints (BaseSearchSpace< SearchPointType, DataSetType > dest, const BaseSearchSpace< SearchPointType, DataSetType > src)

template<typename SearchPointType >
const Point *	refPoint (const SearchPointType &p, const DataSet *refDataSet)

template<>
const Point *	refPoint (const SearchPoint &p, const DataSet *refDataSet)

template<>
const Point *	refPoint (const FrozenSearchPoint &p, const DataSet *refDataSet)

template<typename SearchPointType , typename DataSetType >
const Point *	refPoint (const SearchPointType &p, const BaseSearchSpace< SearchPointType, DataSetType > &sspace)

template<>
const Point *	refPoint (const SearchPoint &p, const SearchSpace &sspace)

template<>
const Point *	refPoint (const FrozenSearchPoint &p, const FrozenSearchSpace &sspace)

template<typename SearchSpaceType >
int	binarySearch (const SearchSpaceType &sspace, int idx, Real value, int start, int end)

QString	RTIME (double t)

void	normalSearch (DataSet &dataset, const DistanceFunction *dist, const Point &query, int N)

template<typename DistanceFunction >
void	frozenSearch (const DataSet &dataset, const float frozenDataSet, int dimension, const float queryPoint, int N, const DistanceFunction &dist)

void	deepFreeze (FrozenDataSet &frozenDS, const FrozenDistance &dist, const Eigen::RowVectorXf &queryPoint, int N)

void	deepFoldedFreeze (FrozenDataSet &frozenDS, const FrozenDistance &dist, const Eigen::RowVectorXf &queryPoint, int N)

void	deepMultiFreeze (FrozenDataSet &frozenDS, const QList< FrozenDistance * > &dists, const QList< Eigen::RowVectorXf > queries, int N)

QDataStream &	operator<< (QDataStream &out, const Transformation &transfo)

QDataStream &	operator>> (QDataStream &in, Transformation &transfo)

QDataStream &	operator<< (QDataStream &out, const TransfoChain &transfo)

QDataStream &	operator>> (QDataStream &in, TransfoChain &transfo)

DataSet *	transform (DataSet *ds, const QString &name, const ParameterMap &params)

DataSet *	applyTransfoChain (DataSet *ds, const QString &yamlTransfoChain)

DataSet *	applyTransfoChain (DataSet *ds, const std::string &yamlTransfoChain)

template<typename T >
QDataStream &	operator<< (QDataStream &out, const Array< T > &array)

template<typename T >
QDataStream &	operator>> (QDataStream &in, Array< T > &array)

template<typename Key , typename Value , typename ExceptionType >
Stringifier &	operator<< (Stringifier &out, const GaiaMap< Key, Value, ExceptionType > &gmap)

QPair< QString, int >	parseDimension (const QString &dimension)
	Parse a string representing a descriptor + optional dimension as a QPair<QString, int>. More...

void	checkIsomorphDataSets (const DataSet dataset1, const DataSet dataset2, bool checkOriginalLayout=true)
	Returns normally if both dataset are isomorph, meaning that they can be linked together, merged together, etc... More...

QStringList	selectDescriptors (const PointLayout &layout, DescriptorType type=UndefinedType, const QStringList &select=QStringList("*"), const QStringList &exclude=QStringList(), bool failOnUnmatched=true)
	This function returns the region of descriptors from a layout that match the patterns given by the select argument but that do not match the ones given by the exclude argument.

void	createIndexMappings (const PointLayout &oldLayout, PointLayout &newLayout, const Region &region, IndexMap &realMap, IndexMap &stringMap, IndexMap &enumMap)
	Creates mappings from indices in the NEW layout to indices in the OLD layout.

void	transferPointData (const Point p, Point result, const IndexMap &realMap, const IndexMap &stringMap, const IndexMap &enumMap)
	Given a set of mappings, transfers the data from a Point to a new one.

Point *	mapPoint (const Point *p, const PointLayout &newLayout, const IndexMap &realMap, const IndexMap &stringMap, const IndexMap &enumMap)
	Creates a new Point that is a mapping of the given one, copying only those descriptors that are in newLayout. More...

RealDescriptor	mergeDescriptors (const Point *p, int nseg, const Region &region)
	Merges the descriptors contained into the given region into a single one.

PointLayout	mergeLayouts (const PointLayout &layout1, const PointLayout &layout2)
	Merges two PointLayouts together, provided that they don't overlap.

Point *	mergePoints (const Point ds1, const Point ds2)
	Merges two points together, provided that their layout don't overlap, and return the resulting point.

DataSet *	mergeDataSets (const DataSet ds1, const DataSet ds2)
	Merges two datasets together, provided that their layout don't overlap, and return the resulting dataset.

QMap< int, QString >	mapRegionIndicesToNames (const Region &region, const DataSet *dataset)
	Given a DataSet (for reference) and a Region, returns the mapping between the region indices and the names of the corresponding descriptors. More...

QList< QPair< double, QString > >	dimensionParticipation (const Eigen::VectorXd &values, const QMap< int, QString > &dnames)
	Given a list of values and a map from their indices to their respective names, returns a list of pairs containing the participation of each original dimension in the target space, expressed as a percentage of the total (meaning the sum of the participation is equal to 100).

QList< QPair< double, QString > >	dimensionBoost (const Eigen::MatrixXf &featureVector, const Region &region, const DataSet *dataset)
	Dimension boost after applying the rotation/scaling defined by the given base vectors. More...

QString	formatDimensionParticipation (const QList< QPair< double, QString > > &participation, bool ratio=false)
	Formats the results of the dimensionParticipation function into a nicely printable string. More...

QMap< QString, QList< int > >	dimensionListToMapping (const QStringList &dims)
	Converts a list of dimensions names in the format `descriptorName`[dim] to a map of descriptor name to list of integers representing the dimensions selected for each one.

int	gaiaVersion (const QDataStream &stream)
	Returns the Gaia version of this stream.

void	setDataStreamVersion (QDataStream &out)
	Sets the version of the QDataStream to the current version of Gaia. More...

void	checkValidDataStream (QDataStream &in)
	Checks that the input QDataStream is a valid one (ie: supported by our version of Gaia) and sets its corresponding type so that it's ready to be read.

int	memoryUsage (QString mtype="VmSize")
	Returns the memory currently used by this process, in KBs. More...

QString	mergeName (const QStringList &name)
	Utility function that merges a sequence of node names into a single fully-qualified name.

QStringList	splitName (const QString &name, bool includeRootNode=false)
	Splits a fully-qualified name into its consecutive node names.

Variables
int	activatedDebugLevels = 0

Logger	loggerInstance

bool	verbose
	Flag that controls whether gaia should be silent or not, in general. More...

const char *	version = GAIA_VERSION

const char *	version_git_sha = GAIA_GIT_SHA

const ParameterMap	noParams

Detailed Description

Main Gaia namespace, which contains all the library functions.

Function Documentation

template<template< typename > class Container, typename T >

int gaia2::binarySearch	(	const Container< T > &	v,
		T	value
	)

Iterative function for binary search (more efficient than recursive one).

If n is the size of the vector, the returned value will be between 0 and n (inclusive) and correspond to the interval between numbers. Ex: r = 0 means it is lower than the lowest value in v, r = n means it is greater than the greatest value in v.

Referenced by gaia2::Distribute::mapPoint(), and gaia2::DataSet::pointIndex().

void gaia2::checkIsomorphDataSets	(	const DataSet *	dataset1,
		const DataSet *	dataset2,
		bool	checkOriginalLayout = `true`
	)

Returns normally if both dataset are isomorph, meaning that they can be linked together, merged together, etc...

In particular, they need to have exactly the same points and the same original layout (we don't care about the transformation history). In the case they aren't, throw an exception with an error message. In case any of the 2 datasets is the null pointer, this doesn't throw an exception.

References gaia2::Point::name(), and gaia2::DataSet::originalLayout().

Referenced by mergeDataSets(), gaia2::DataSet::setReferenceDataSet(), and splitName().

Eigen::MatrixXd gaia2::covarianceMatrix	(	const PointArray &	v,
		const Region &	region
	)

Computes the covariance matrix given a set of points.

Parameters

v	the set of points for which to compute the covariance matrix.
region	the Region containing the descriptors to be included in the covariance matrix computation.

Referenced by gaia2::RCA::analyze(), and gaia2::RCA::computeCovarianceMatrix().

QList< QPair< double, QString > > gaia2::dimensionBoost	(	const Eigen::MatrixXf &	featureVector,
		const Region &	region,
		const DataSet *	dataset
	)

Dimension boost after applying the rotation/scaling defined by the given base vectors.

Useful in particular in RCA (where the vectors are not an orthonormal base).

References mapRegionIndicesToNames().

Referenced by gaia2::RCA::analyze(), and splitName().

QStringList gaia2::findVariableLengthDescriptors ( const DataSet * dataset )

For all variable-length descriptors in this DataSet, find all the ones which actually are of different lengths, ie: those such that there exists 2 points p1 and p2 such as len(p1.desc) != len(p2.desc).

If there is a descriptor which has len == 0 for all points, then this descriptor is considered variable-length as well. The rationale is the following: if they're all empty, it only makes sense if they're actually lists of values which (by chance) happen to be all empty, but it can in no case be a fixed-length descriptor, because that would be completely stupid to have such an empty descriptor.

References gaia2::PointLayout::descriptorName(), gaia2::DataSet::layout(), and gaia2::PointArray::samplePoint().

QString gaia2::formatDimensionParticipation	(	const QList< QPair< double, QString > > &	participation,
		bool	ratio = `false`
	)

Formats the results of the dimensionParticipation function into a nicely printable string.

If ratio == true, prints "x 1.23" instead of "1.23%".

Referenced by gaia2::RCA::analyze(), and splitName().

QPair< Point, Point > gaia2::getRegionMeanVar	(	const DataSet &	dataset,
		const Region &	region
	)

Gets the mean and variance of the descriptors in the given region.

As those concepts only make sense for real numbers, the string descriptors contained in this region will be silently ignored. This function requires all descriptors in the region to be fixed-length, otherwise it throws an exception.

References gaia2::Region::checkLengthTypeOnlyFrom(), gaia2::PointLayout::copy(), gaia2::DataSet::layout(), gaia2::Region::select(), gaia2::Point::setLayout(), and gaia2::PointArray::totalSegments().

QPair< Point, Point > gaia2::getRegionMinMax	(	const DataSet &	dataset,
		const Region &	region
	)

Gets the mean and variance of the descriptors in the given region.

As those concepts only make sense for real numbers, the string descriptors contained in this region will be silently ignored. This function requires all descriptors in the region to be fixed-length, otherwise it throws an exception.

References gaia2::Region::checkLengthTypeOnlyFrom(), gaia2::PointLayout::copy(), gaia2::DataSet::layout(), gaia2::PointArray::samplePoint(), gaia2::Region::select(), and gaia2::Point::setLayout().

template<typename T >

void gaia2::hist	(	const T *	array,
		uint	n,
		int *	n_array,
		T *	x_array,
		uint	n_bins
	)

Given a set of values, computes the associated histogram.

This method is designed to work the same way as in MatLab/Octave. It is based on the algorithms used in Octave rather than Matlab. The result structures (n_arr, x_arr) have to be allocated before calling this function.

Parameters

array	the input array, containing the data values
n	the number of elements of this array
n_array	the array where the distribution will be written
x_array	the array that will contain the centers of each bin
n_bins	the number of desired bins for the distribution

void gaia2::init ( )

Init function that makes sure that all the factories have correctly registered all the types they can create.

You need to call this function once in your program, as soon as you can.

References registerAnalyzers(), and registerMetrics().

template<typename T >

T gaia2::kurtosis	(	const T *	array,
		uint	n
	)

Computes the kurtosis of an array.

It is mathematically defined as: kurtosis = mu_4 / mu_2**{2}, where mu_i is the i_th central moment.

References mean(), and variance().

Point * gaia2::mapPoint	(	const Point *	p,
		const PointLayout &	newLayout,
		const IndexMap &	realMap,
		const IndexMap &	stringMap,
		const IndexMap &	enumMap
	)

Creates a new Point that is a mapping of the given one, copying only those descriptors that are in newLayout.

The Index maps should be precalculated with the createIndexMappings function.

References gaia2::Point::name(), gaia2::Point::setLayout(), gaia2::Point::setName(), and transferPointData().

Referenced by gaia2::SVMPredict::mapDataSet(), gaia2::EnumerateApplier::mapDataSet(), gaia2::AddFieldApplier::mapDataSet(), gaia2::RenameApplier::mapPoint(), gaia2::MatMult::mapPoint(), gaia2::SVMPredict::mapPoint(), gaia2::RemoveDesc::mapPoint(), gaia2::EnumerateApplier::mapPoint(), gaia2::SelectDesc::mapPoint(), gaia2::FixLengthApplier::mapPoint(), gaia2::AddFieldApplier::mapPoint(), gaia2::ExtractApplier::mapPoint(), and splitName().

QMap< int, QString > gaia2::mapRegionIndicesToNames	(	const Region &	region,
		const DataSet *	dataset
	)

Given a DataSet (for reference) and a Region, returns the mapping between the region indices and the names of the corresponding descriptors.

This function will throw if the Region is not composed of a single type of descriptors.

References gaia2::PointLayout::descriptorName(), gaia2::DataSet::layout(), gaia2::PointArray::samplePoint(), gaia2::Region::select(), and gaia2::Region::type().

Referenced by dimensionBoost(), and splitName().

int gaia2::memoryUsage ( QString mtype = "VmSize" )

Returns the memory currently used by this process, in KBs.

mtype can be either one of:

VmSize: total virtual memory
VmRSS: resident segment size

NB: Only works on linux at the moment, will return 0 on other platforms.

Referenced by splitName().

QPair< QString, int > gaia2::parseDimension ( const QString & dimension )

Parse a string representing a descriptor + optional dimension as a QPair<QString, int>.

In case the dimension is not specified, 0 will be returned.

Referenced by gaia2::Distribute::mapPoint(), and splitName().

void gaia2::registerAnalyzers ( )

Registers all the analyzers available inside the AnalyzerFactory.

NB: For the python docstrings to work correctly, the name with which the analyzer is registered needs to be the exact same name as the class name (case-sensitive).

Referenced by init().

void gaia2::registerMetrics ( )

Registers all the metrics available inside the MetricFactory.

NB: For the python docstrings to work correctly, the name with which the distance is registered needs to be the exact same name as the class name, with the trailing "Distance" removed (case-sensitive).

References gaia2::PointLayout::descriptorLocation(), and selectDescriptors().

Referenced by init().

void gaia2::setDataStreamVersion ( QDataStream & out )

Sets the version of the QDataStream to the current version of Gaia.

This function must be called before serializing anything in the stream.

Referenced by gaia2::DataSet::mergeFiles(), gaia2::TransfoChain::retransformDataSet(), gaia2::FrozenDataSet::save(), splitName(), gaia2::Point::toBase64(), and gaia2::DataSet::toBase64().

template<typename T >

T gaia2::skewness	(	const T *	array,
		uint	n
	)

Computes the skewness of an array.

It is mathematically defined as: skewness = mu_3 / mu_2**{3/2}, where mu_i is the i_th central moment.

References mean(), and variance().

Real gaia2::traceMatrixProduct	(	const Real *	m1,
		const Real *	m2,
		int	size
	)

inline

Compute the trace of the matrix product of m1 and m2.

Both matrices need to be squared of dimension size.

Referenced by gaia2::KullbackLeiblerDistance::operator()().

Real gaia2::traceMatrixProductMean	(	const Real *	m1,
		const Real *	m2,
		const Real *	v1,
		const Real *	v2,
		int	size
	)

inline

Compute the trace of the product of 2 matrices and vectors using the formula: trace((m1+m2)*(v1-v2)*(v1-v2)')

m1 and m2 are square matrices of dimension size, v1 and v2 are vectors of dimension size.

Referenced by gaia2::KullbackLeiblerDistance::operator()().

Variable Documentation

bool gaia2::verbose

Flag that controls whether gaia should be silent or not, in general.

That could include information such as transformation progress, time needed for doing queries, etc...

Namespaces

Classes

Typedefs

Enumerations

Functions

Variables

Detailed Description

Function Documentation

Variable Documentation