#include <sourcesinkdata.h>

Inheritance diagram for SourceSinkData:

Collaboration diagram for SourceSinkData:

Public Member Functions
	SourceSinkData ()
	Default constructor.

	SourceSinkData (const SourceSinkData &other)
	Copy constructor.

SourceSinkData &	operator= (const SourceSinkData &other)
	Assignment operator.

void	Clear ()
	Clear all data from the object.

Elemental_Profile_Set *	AppendSampleSet (const string &name, const Elemental_Profile_Set &elemental_profile_set=Elemental_Profile_Set())
	Add a new sample set (source or target group) to the dataset.

vector< string >	SourceGroupNames () const
	Retrieves the names of all source groups (excluding target)

bool	SetSelectedTargetSample (const string &sample_name)
	Sets the currently selected target sample for analysis.

Elemental_Profile_Set *	GetSampleSet (const string &name)
	Get a sample set (source or target group) by name.

bool	SetTargetGroup (const string &targroup)
	Sets the target/sink group designation.

string	GetTargetGroup () const
	Retrieves the name of the target group.

SourceSinkData	CreateCorrectedDataset (const string &target, bool omnsizecorrect, map< string, element_information > *elementinfo)
	Create a corrected copy of the dataset for a specific target sample.

SourceSinkData	CreateCorrectedAndFilteredDataset (bool exclude_samples, bool exclude_elements, bool omnsizecorrect, const string &target="") const
	Create a corrected and filtered copy of the dataset.

SourceSinkData	ExtractChemicalElements (bool isotopes) const
	Extract only chemical elements (and optionally isotopes)

SourceSinkData	BoxCoxTransformed (bool calculate_optimal_lambda=false)
	Applies Box-Cox transformation to all source groups for normalization.

map< string, element_information > *	GetElementInformation ()
	Retrieves pointer to the element information map.

void	IncludeExcludeElementsBasedOn (const vector< string > &elements)
	Sets element inclusion based on a specified list.

Distribution *	GetFittedDistribution (const string &element_name)
	Get the fitted distribution for a specific element at dataset level.

bool	PerformRegressionVsOMAndSize (const string &om, const string &particle_size, regression_form form, const double &p_value_threshold=0.05)
	Performs multiple linear regression of elements vs OM and particle size.

vector< string >	OMandSizeConstituents ()
	Retrieves the names of OM and particle size constituents.

vector< ResultItem >	GetMLRResults ()
	Retrieves multiple linear regression results for all sample groups.

string	FirstOMConstituent ()
	Retrieves the name of the first organic matter constituent.

string	FirstSizeConstituent ()
	Retrieves the name of the first particle size constituent.

void	SetOMandSizeConstituents (const string &_omconstituent, const string &_sizeconsituent)
	Sets the names of OM and particle size constituents.

void	SetOMandSizeConstituents (const vector< string > &_omsizeconstituents)
	Sets the names of OM and particle size constituents from a vector.

void	OutlierAnalysisForAll (const double &lower_threshold=-3, const double &upper_threshold=3)
	Performs outlier detection on all source groups.

vector< string >	NegativeValueCheck ()
	Checks for zero or negative concentration values across all sources.

CMBVector	BracketTest (const string &target_sample, bool correct_based_on_om_n_size)
	Performs bracket test to check if target concentrations fall within source ranges.

CMBMatrix	BracketTest (bool correct_based_on_om_n_size, bool exclude_elements, bool exclude_samples)
	Performs bracket test on all target samples.

DFA_result	DiscriminantFunctionAnalysis ()
	Performs discriminant function analysis for all source groups.

DFA_result	DiscriminantFunctionAnalysis (const string &source1, const string &source2)
	Performs discriminant function analysis between two specific sources.

DFA_result	DiscriminantFunctionAnalysis (const string &source1)
	Performs discriminant function analysis for one source vs all others.

vector< CMBVector >	StepwiseDiscriminantFunctionAnalysis (const string &source1, const string &source2)
	Performs stepwise discriminant analysis between two specific sources.

vector< CMBVector >	StepwiseDiscriminantFunctionAnalysis ()
	Performs stepwise discriminant analysis across all source groups.

vector< CMBVector >	StepwiseDiscriminantFunctionAnalysis (const string &source1)
	Performs stepwise discriminant analysis for one source vs all others.

Elemental_Profile_Set	DifferentiationPower (bool use_log, bool include_target)
	Computes differentiation power for all source pairs.

Elemental_Profile_Set	DifferentiationPower_Percentage (bool include_target)
	Computes rank-based differentiation percentage for all source pairs.

Elemental_Profile_Set	DifferentiationPower_P_value (bool include_target)
	Computes t-test p-values for all source pairs.

Elemental_Profile	DifferentiationPower_Percentage (const string &source1, const string &source2)
	Computes rank-based differentiation percentage between two sources.

Elemental_Profile	DifferentiationPower (const string &source1, const string &source2, bool use_log)
	Computes differentiation power metric between two sources.

Elemental_Profile	t_TestPValue (const string &source1, const string &source2, bool use_log)
	Computes t-test p-values for element-wise differences between two sources.

CMBVector	ANOVA (bool use_log)
	Performs one-way ANOVA for all elements across source groups.

ANOVA_info	ANOVA (const string &element, bool use_log)
	Performs one-way ANOVA for a single element across source groups.

string	GetParameterName (int index) const
	Get the name of a parameter by its index.

bool	InitializeParametersAndObservations (const string &targetsamplename, estimation_mode est_mode=estimation_mode::elemental_profile_and_contribution)
	Initialize parameters and observations for MCMC optimization.

void	SetParameterEstimationMode (estimation_mode est_mode)
	Sets the estimation mode for parameter optimization.

bool	SetParameterValue (size_t index, double value)
	Sets a parameter value and updates corresponding model components.

void	SetProgressWindow (ProgressWindow *_rtw)
	Sets the progress window for displaying optimization progress.

bool	SolveLevenberg_Marquardt (transformation trans=transformation::linear)
	Solves for optimal source contributions using the Levenberg-Marquardt algorithm.

CMBTimeSeriesSet	LM_Batch (transformation transform, bool apply_om_size_correction, map< string, vector< string > > &negative_elements)
	Solves CMB model for all target samples using Levenberg-Marquardt.

CVector	GetPredictedValues ()
	Retrieves predicted values for all observations.

CVector	GradientUpdate (estimation_mode estmode=estimation_mode::elemental_profile_and_contribution)
	Performs one gradient ascent step with adaptive step size.

Results	MCMC (const string &target_sample, map< string, string > arguments, CMCMC< SourceSinkData > mcmc, ProgressWindow progress_window, const string &working_folder)
	Performs Markov Chain Monte Carlo analysis for Bayesian source apportionment.

CMBMatrix	MCMC_Batch (map< string, string > arguments, CMCMC< SourceSinkData > mcmc, ProgressWindow progress_window, const string &working_folder)
	Performs batch MCMC analysis on all target samples.

double	GetObjectiveFunctionValue ()
	Returns the objective function value for optimization algorithms.

double	LogLikelihood (estimation_mode est_mode=estimation_mode::elemental_profile_and_contribution)
	Calculates the total log-likelihood for Bayesian source apportionment.

CMBTimeSeriesSet	BootStrap (const double &percentage, unsigned int num_iterations, string target_sample, bool use_softmax)
	Performs bootstrap uncertainty analysis on source contributions.

bool	BootStrap (Results *results, const double &percentage, unsigned int num_iterations, string target_sample, bool use_softmax)
	Performs bootstrap analysis and generates comprehensive uncertainty results.

CMBTimeSeriesSet	VerifySource (const string &source_group, bool use_softmax, bool apply_om_size_correction)
	Performs leave-one-out validation on a source group.

ResultItem	GetContribution ()
	Packages source contributions into a ResultItem for output.

ResultItem	GetObservedElementalProfile ()
	Retrieves the observed elemental concentrations for the selected target sample.

ResultItem	GetObservedElementalProfile_Isotope ()
	Retrieves the observed isotope delta values for the selected target sample.

ResultItem	GetPredictedElementalProfile (parameter_mode param_mode=parameter_mode::based_on_fitted_distribution)
	Generates predicted elemental concentrations for the target sample.

ResultItem	GetPredictedElementalProfile_Isotope (parameter_mode param_mode=parameter_mode::based_on_fitted_distribution)
	Generates predicted isotope delta values for the target sample.

ResultItem	GetObservedvsModeledElementalProfile (parameter_mode param_mode=parameter_mode::based_on_fitted_distribution)
	Creates a comparison of observed vs modeled elemental profiles.

ResultItem	GetObservedvsModeledElementalProfile_Isotope (parameter_mode param_mode=parameter_mode::based_on_fitted_distribution)
	Creates a comparison of observed vs modeled isotope delta values.

ResultItem	GetCalculatedElementMeans ()
	Computes estimated mean concentrations for all source elements.

ResultItem	GetCalculatedElementSigma ()
	Computes estimated standard deviations for all source elements.

ResultItem	GetCalculatedElementMu ()
	Computes μ parameters from fitted log-normal distributions for source elements.

ResultItem	GetEstimatedElementMu ()
	Retrieves estimated μ parameters from Bayesian inference for source elements.

ResultItem	GetEstimatedElementMean ()
	Computes actual mean concentrations from estimated log-normal parameters.

ResultItem	GetEstimatedElementSigma ()
	Retrieves estimated σ parameters from Bayesian inference for source elements.

vector< ResultItem >	GetSourceProfiles ()
	Retrieves elemental profiles for all source groups.

void	AddtoToolsUsed (const string &tool)
	Adds a tool name to the list of tools used in analysis.

vector< Parameter > &	Parameters ()
	Retrieves reference to the parameters vector.

size_t	ParametersCount ()
	Returns the number of parameters.

size_t	ObservationsCount ()
	Returns the number of observations.

Parameter *	parameter (size_t i)
	Retrieves pointer to a parameter by index.

const Parameter *	parameter (size_t i) const
	Retrieves const pointer to a parameter by index.

Observation *	observation (size_t i)
	Retrieves pointer to an observation by index.

estimation_mode	ParameterEstimationMode ()
	Retrieves the current estimation mode.

ConcentrationSet *	GetElementDistribution (const string &element_name)
	Retrieves pointer to element distribution at dataset level.

ConcentrationSet *	GetElementDistribution (const string &element_name, const string &sample_group)
	Retrieves pointer to element distribution for a specific group.

element_information *	GetElementInformation (const string &element_name)
	Retrieves pointer to element information metadata by name.

void	PopulateElementDistributions ()
	Populate element distributions from all groups.

void	PopulateElementInformation (const map< string, element_information > *ElementInfo=nullptr)
	Populate element information metadata.

void	AssignAllDistributions ()
	Assign distributions to all elements at both dataset and group levels.

int	CountElements (bool exclude_elements) const
	Count the number of elements in the dataset.

vector< string >	GetSampleNames (const string &group_name) const
	Get all sample names within a specific group.

vector< string >	GetGroupNames () const
	Get all group names in the dataset.

vector< string >	GetElementNames () const
	Get all element names in the dataset.

void	IncludeExcludeAllElements (bool include_in_analysis)
	Sets inclusion flag for all elements.

Elemental_Profile	Sample (const string &sample_name) const
	Retrieves an elemental profile by sample name.

profiles_data	ExtractConcentrationData (const vector< vector< string > > &indicators) const
	Extract concentration data for specified samples.

Elemental_Profile_Set	ExtractSamplesAsProfileSet (const vector< vector< string > > &indicators) const
	Extract samples as an Elemental_Profile_Set.

element_data	ExtractElementConcentrations (const string &element, const string &group) const
	Extract concentration data for a specific element from a group.

map< string, vector< double > >	ExtractElementDataByGroup (const string &element) const
	Extract concentration data for a specific element from all groups.

SourceSinkData	ExtractSpecificElements (const vector< string > &element_list) const
	Extract specific elements from source groups.

map< string, ConcentrationSet >	ExtractConcentrationSet ()
	Extracts concentration distributions for all elements across sources.

vector< string >	ElementsToBeUsedInCMB ()
	Identifies chemical elements to be used in CMB analysis.

vector< string >	IsotopesToBeUsedInCMB ()
	Identifies isotopes to be used in CMB analysis.

void	PopulateConstituentOrders ()
	Populates all element ordering vectors used throughout CMB analysis.

string	SelectedTargetSample () const
	Retrieves the name of the currently selected target sample.

bool	ToolsUsed (const string &tool_name)
	Checks if a specific analysis tool has been used.

string	GetOutputPath () const
	Get the output directory path.

QMap< QString, double > *	GetOptions ()
	Retrieves pointer to the options map.

bool	SetOutputPath (const string &output_path)
	Set the output directory path.

QJsonObject	ElementInformationToJsonObject () const
	Exports element information metadata to a JSON object.

QJsonArray	ToolsUsedToJsonObject () const
	Exports the list of analysis tools used to a JSON array.

QJsonObject	OptionsToJsonObject () const
	Exports analysis options/settings to a JSON object.

bool	ReadToolsUsedFromJsonObject (const QJsonArray &jsonarray)
	Deserializes the list of analysis tools from a JSON array.

bool	ReadElementInformationfromJsonObject (const QJsonObject &jsonobject)
	Deserializes element information metadata from a JSON object.

bool	ReadElementDatafromJsonObject (const QJsonObject &jsonobject)
	Deserializes elemental profile data from a JSON object.

bool	ReadOptionsfromJsonObject (const QJsonObject &jsonobject)
	Deserializes analysis options from a JSON object.

QJsonObject	ElementDataToJsonObject () const
	Exports all elemental profile data to a JSON object.

bool	WriteDataToFile (QFile *file)
	Writes elemental profile data to a text file.

bool	WriteToFile (QFile *file)
	Writes dataset to a text file.

bool	ReadFromFile (QFile *fil)
	Loads complete dataset from a JSON file.

Private Member Functions
Elemental_Profile *	GetElementalProfile (const string &sample_name)
	Finds and retrieves an elemental profile by sample name.

QString	Role (const element_information::role &role) const
	Converts element role enum to string representation.

element_information::role	Role (const QString &role_string) const
	Converts string representation to element role enum.

double	GetParameterValue (size_t index) const
	Retrieves a single parameter value by index.

bool	SetParameterValue (const CVector &values)
	Sets multiple parameter values from a vector.

CVector	GetParameterValue () const
	Retrieves all current parameter values as a vector.

bool	InitializeContributionsRandomly ()
	Initialize source contributions randomly (linear constraint)

bool	InitializeContributionsRandomlySoftmax ()
	Initialize source contributions randomly (softmax transformation)

CVector	GetSourceContributions ()
	Retrieves the contributions from all sources.

CVector	GetContributionVector (bool include_all=true)
	Retrieves the current source contribution fractions.

CVector	GetContributionVectorSoftmax ()
	Retrieves the softmax parameters for source contributions.

void	SetContribution (size_t source_index, double contribution_value)
	Sets a single source contribution value.

void	SetContributionSoftmax (size_t source_index, double softmax_value)
	Sets a single softmax parameter value.

void	SetContribution (const CVector &contributions)
	Sets all source contributions from a vector.

void	SetContributionSoftmax (const CVector &softmax_params)
	Sets all source contributions using softmax transformation.

Parameter *	GetElementDistributionMuParameter (size_t element_index, size_t source_index)
	Retrieves pointer to the μ (mean) parameter for an element distribution.

Parameter *	GetElementDistributionSigmaParameter (size_t element_index, size_t source_index)
	Retrieves pointer to the σ (std dev) parameter for an element distribution.

double	GetElementDistributionMuValue (size_t element_index, size_t source_index)
	Retrieves the current value of the μ parameter for an element distribution.

double	GetElementDistributionSigmaValue (size_t element_index, size_t source_index)
	Retrieves the current value of the σ parameter for an element distribution.

CVector	ObservedDataforSelectedSample (const string &SelectedTargetSample="")
	Retrieves the observed elemental data for a selected target sample.

CVector	ObservedDataforSelectedSample_Isotope (const string &SelectedTargetSample="")
	Retrieves the observed isotopic data for a selected target sample.

CVector	ObservedDataforSelectedSample_Isotope_delta (const string &SelectedTargetSample="")
	Retrieves the observed isotopic data in delta notation.

CVector	PredictTarget (parameter_mode param_mode=parameter_mode::direct)
	Predicts target sample elemental concentrations based on source contributions.

CVector	PredictTarget_Isotope (parameter_mode param_mode=parameter_mode::direct)
	Predicts target sample isotopic compositions based on source contributions.

CVector	PredictTarget_Isotope_delta (parameter_mode param_mode=parameter_mode::based_on_fitted_distribution)
	Predicts target sample isotopic compositions in delta notation.

CMatrix	BuildSourceMeanMatrix (parameter_mode param_mode=parameter_mode::based_on_fitted_distribution)
	Builds the source mean concentration matrix for chemical elements.

CMatrix	BuildSourceMeanMatrix_Isotopes (parameter_mode param_mode=parameter_mode::based_on_fitted_distribution)
	Builds the source mean concentration matrix for isotopes.

double	LogPriorContributions ()
	Calculates the log prior probability for source contributions.

double	LogLikelihoodSourceElementalDistributions ()
	Calculates the log-likelihood of source elemental distributions.

double	LogLikelihoodModelvsMeasured (estimation_mode est_mode=estimation_mode::elemental_profile_and_contribution)
	Calculates the log-likelihood of the model prediction versus measured data.

double	LogLikelihoodModelvsMeasured_Isotope (estimation_mode est_mode=estimation_mode::elemental_profile_and_contribution)
	Calculates the log-likelihood of model versus measured isotopic data.

CVector	Gradient (const CVector &parameters, estimation_mode est_mode)
	Computes the normalized gradient of the log-likelihood function.

CVector	ResidualVector ()
	Calculates the combined residual vector for elemental and isotopic predictions.

CVector_arma	ResidualVector_arma ()
	Calculates the combined residual vector using Armadillo vector format.

CMatrix_arma	ResidualJacobian_arma ()
	Calculates the Jacobian matrix of residuals with respect to contributions (Armadillo)

CMatrix	ResidualJacobian ()
	Calculates the Jacobian matrix of residuals with respect to contributions.

CMatrix	ResidualJacobian_softmax ()
	Calculates the Jacobian using softmax parameterization of contributions.

CVector	OneStepLevenberg_Marquardt (double lambda)
	Performs one iteration of the Levenberg-Marquardt optimization algorithm.

CVector	OneStepLevenberg_Marquardt_softmax (double lambda)
	Performs one iteration of Levenberg-Marquardt using softmax parameterization.

vector< string >	GetSourceOrder () const
	Retrieves the ordering of source groups.

vector< string >	SamplesetsOrder ()
	Retrieves the ordering of sample sets.

vector< string >	ConstituentOrder ()
	Retrieves the ordering of all constituents.

vector< string >	ElementOrder ()
	Retrieves the ordering of chemical elements.

vector< string >	IsotopeOrder ()
	Retrieves the ordering of isotopes.

vector< string >	SizeOMOrder ()
	Retrieves the ordering of size and OM constituents.

int	TotalNumberofSourceSamples () const
	Counts the total number of source samples across all source groups.

double	GrandMean (const string &element, bool use_log)
	Computes grand mean concentration for an element across all sources.

Elemental_Profile_Set	LumpAllProfileSets ()
	Combines all source samples into a single profile set.

CMBVector	OptimalBoxCoxParameters ()
	Computes optimal Box-Cox transformation parameters for all elements.

SourceSinkData	RandomlyEliminateSourceSamples (const double &percentage)
	Creates dataset with randomly excluded source samples for validation.

vector< string >	AllSourceSampleNames () const
	Retrieves names of all samples across all source groups.

vector< string >	RandomlypickSamples (const double &percentage) const
	Randomly selects a subset of source samples.

SourceSinkData	ReplaceSourceAsTarget (const string &source_sample_name) const
	Creates a new dataset with a source sample designated as the target.

CMatrix	WithinGroupCovarianceMatrix ()
	Computes pooled within-group covariance matrix.

CMatrix	BetweenGroupCovarianceMatrix ()
	Computes between-group covariance matrix.

CMatrix	TotalScatterMatrix ()
	Computes total scatter matrix.

double	WilksLambda ()
	Computes Wilks' Lambda statistic for multivariate group separation.

double	DFA_P_Value ()
	Computes p-value for discriminant function analysis.

CMBVectorSet	DFA_Projected ()
	Projects all groups onto the discriminant function axis.

CMBVectorSet	DFA_Projected (const string &source1, const string &source2)
	Projects two specific groups onto the discriminant function axis.

CMBVectorSet	DFA_Projected (const string &source1, SourceSinkData *original)
	Projects samples from original dataset onto discriminant function.

CMBVector	DFA_eigvector ()
	Computes the primary discriminant function eigenvector.

CMBVector	DFA_weight_vector (const string &source1, const string &source2)
	Computes discriminant weight vector for two-group comparison.

CMBVector	DFATransformed (const CMBVector &eigenvector, const string &source_group)
	Projects samples onto a discriminant function axis.

Elemental_Profile_Set	TheRest (const string &excluded_source)
	Collects all source samples except those from a specified source group.

CMBVector	MeanElementalContent (const string &group_name)
	Computes mean elemental concentrations for a specific group.

CMBVector	MeanElementalContent ()
	Computes weighted mean elemental concentrations across all sources.

Private Attributes
map< string, element_information >	element_information_

map< string, ConcentrationSet >	element_distributions_

int	numberofconstituents_

int	numberofisotopes_

int	numberofsourcesamplesets_

vector< Observation >	observations_

vector< Parameter >	parameters_

string	outputpath_

string	target_group_

string	selected_target_sample_

vector< string >	samplesetsorder_

vector< string >	constituent_order_

vector< string >	element_order_

vector< string >	isotope_order_

vector< string >	size_om_order_

estimation_mode	parameter_estimation_mode_

string	omconstituent_

string	sizeconsituent_

double	regression_p_value_threshold_

double	distance_coeff_

double	error_stdev_

double	error_stdev_isotope_

double	epsilon_

ProgressWindow *	rtw_ = nullptr

list< string >	tools_used_

QMap< QString, double >	options_

Detailed Description

Definition at line 66 of file sourcesinkdata.h.

Constructor & Destructor Documentation

◆ SourceSinkData() [1/2]

SourceSinkData::SourceSinkData ( )

Default constructor.

Initializes an empty SourceSinkData object with default options. Sets outlier deviation threshold to 3.0 by default.

Definition at line 17 of file sourcesinkdata.cpp.

References options_, rtw_, and string.

◆ SourceSinkData() [2/2]

SourceSinkData::SourceSinkData ( const SourceSinkData & other )

Copy constructor.

Creates a deep copy of another SourceSinkData object, including all elemental profile sets, element information, distributions, parameters, observations, and analysis settings.

Parameters

other The SourceSinkData object to copy from

Definition at line 46 of file sourcesinkdata.cpp.

References rtw_, and string.

Member Function Documentation

◆ AddtoToolsUsed()

void SourceSinkData::AddtoToolsUsed ( const string & tool )

Adds a tool name to the list of tools used in analysis.

Records that a particular analysis tool was applied to this dataset. Prevents duplicate entries.

Parameters

tool	Name of the tool used (e.g., "MCMC", "Bootstrap")

Definition at line 2606 of file sourcesinkdata.cpp.

References tools_used_, and ToolsUsed().

Referenced by Conductor::Execute(), and ReadToolsUsedFromJsonObject().

◆ AllSourceSampleNames()

vector< string > SourceSinkData::AllSourceSampleNames ( ) const

private

Retrieves names of all samples across all source groups.

Collects sample names from all source groups (excluding target) into a single flat list.

Returns: Vector of all source sample names

Note: Target group samples are excluded

Definition at line 4000 of file sourcesinkdata.cpp.

References target_group_.

Referenced by RandomlypickSamples().

◆ ANOVA() [1/2]

CMBVector SourceSinkData::ANOVA ( bool use_log )

Performs one-way ANOVA for all elements across source groups.

Conducts analysis of variance to test whether mean concentrations differ significantly among source groups for each element.

Null hypothesis: μ₁ = μ₂ = ... = μₖ (all source means equal)

Parameters

use_log If true, perform ANOVA on log-transformed concentrations; if false, use linear-space concentrations

Returns: CMBVector of p-values for each element, labeled with element names

Note: Lower p-values indicate significant differences among sources; Target group is excluded from analysis

Definition at line 3490 of file sourcesinkdata.cpp.

References ANOVA(), GetElementNames(), ANOVA_info::p_value, and CMBVector::SetLabels().

Referenced by ANOVA(), and Conductor::ExecuteANOVA().

◆ ANOVA() [2/2]

ANOVA_info SourceSinkData::ANOVA	(	const string &	element,
		bool	use_log
	)

Performs one-way ANOVA for a single element across source groups.

Conducts detailed analysis of variance decomposing total variance into between-group and within-group components.

ANOVA decomposition:

SST (Total) = SSB + SSW
SSB (Between-group) = Σ n_i(μ_i - μ_grand)²
SSW (Within-group) = Σ Σ (x_ij - μ_i)²
F-statistic = MSB / MSW

Parameters

element	Name of the element to analyze
use_log	If true, perform ANOVA on log-transformed concentrations

Returns: ANOVA_info structure containing SST, SSB, SSW, MSB, MSW, F, and p-value

Definition at line 3506 of file sourcesinkdata.cpp.

References ConcentrationSet::CalculateMean(), ConcentrationSet::CalculateMeanLog(), ConcentrationSet::CalculateSSE(), ConcentrationSet::CalculateSSELog(), ConcentrationSet::CalculateStdDevLog(), ANOVA_info::F, Elemental_Profile_Set::GetElementDistribution(), LumpAllProfileSets(), ANOVA_info::MSB, ANOVA_info::MSW, ANOVA_info::p_value, ANOVA_info::SSB, ANOVA_info::SST, ANOVA_info::SSW, and target_group_.

◆ AppendSampleSet()

Elemental_Profile_Set * SourceSinkData::AppendSampleSet	(	const string &	name,
		const Elemental_Profile_Set &	elemental_profile_set = `Elemental_Profile_Set()`
	)

Add a new sample set (source or target group) to the dataset.

Adds a collection of elemental profiles under a group name. Typical group names are source identifiers (e.g., "Agricultural soil", "Road dust") or the target group name (e.g., "Stream sediment").

Parameters

name	Group name for this sample set
elemental_profile_set	Collection of elemental profiles for this group (default: empty)

Returns: Pointer to the added profile set, or nullptr if name already exists

Definition at line 368 of file sourcesinkdata.cpp.

References elemental_profile_set, and name.

Referenced by DiscriminantFunctionAnalysis(), DiscriminantFunctionAnalysis(), and MainWindow::ReadExcel().

◆ AssignAllDistributions()

void SourceSinkData::AssignAllDistributions ( )

Assign distributions to all elements at both dataset and group levels.

For each element:

At dataset level: Selects best-fitting distribution (normal vs lognormal) and estimates parameters from all data
At group level: Assigns same distribution type and estimates parameters from group-specific data

Special handling for isotopes: Always uses normal distribution regardless of best-fit results.

Note: Should be called after PopulateElementDistributions()

Definition at line 557 of file sourcesinkdata.cpp.

References Distribution::distribution, element_distributions_, element_information_, ConcentrationSet::EstimateDistributionParameters(), GetElementNames(), ConcentrationSet::GetFittedDistribution(), element_information::isotope, normal, Distribution::parameters, and Role().

Referenced by BoxCoxTransformed(), CreateCorrectedAndFilteredDataset(), CreateCorrectedDataset(), ExtractChemicalElements(), ExtractSpecificElements(), MainWindow::LoadModel(), RandomlyEliminateSourceSamples(), MainWindow::ReadExcel(), and ReplaceSourceAsTarget().

◆ BetweenGroupCovarianceMatrix()

CMatrix SourceSinkData::BetweenGroupCovarianceMatrix ( )

private

Computes between-group covariance matrix.

Calculates covariance matrix based on deviations of group means from the overall mean. Represents the "signal" or between-group scatter.

Formula: S_B = Σ[n_i × (μ_i - μ_overall)(μ_i - μ_overall)ᵀ] / N

Returns: Between-group covariance matrix (num_elements × num_elements)

Note: Target group excluded

Definition at line 4474 of file sourcesinkdata.cpp.

References GetElementNames(), MeanElementalContent(), CMBVector::size(), and target_group_.

Referenced by DFA_eigvector(), and WilksLambda().

◆ BootStrap() [1/2]

CMBTimeSeriesSet SourceSinkData::BootStrap	(	const double &	percentage,
		unsigned int	num_iterations,
		string	target_sample,
		bool	use_softmax
	)

Performs bootstrap uncertainty analysis on source contributions.

Conducts bootstrap resampling to estimate uncertainty in source contribution estimates. Each bootstrap iteration randomly excludes a percentage of source samples, re-solves the CMB model, and records the resulting contributions.

Parameters

percentage	Percentage of source samples to exclude per iteration (0-100)
num_iterations	Number of bootstrap iterations
target_sample	Name of target sample to apportion
use_softmax	If true, use softmax transformation for contributions

Returns: CMBTimeSeriesSet containing contribution estimates from all iterations

Note: Progress updates sent to rtw_ if available

Definition at line 3699 of file sourcesinkdata.cpp.

References GetContributionVector(), InitializeParametersAndObservations(), linear, numberofsourcesamplesets_, RandomlyEliminateSourceSamples(), rtw_, samplesetsorder_, ProgressWindow::SetProgress(), softmax, and SolveLevenberg_Marquardt().

Referenced by Conductor::ExecuteErrorAnalysis().

◆ BootStrap() [2/2]

bool SourceSinkData::BootStrap	(	Results *	results,
		const double &	percentage,
		unsigned int	num_iterations,
		string	target_sample,
		bool	use_softmax
	)

Performs bootstrap analysis and generates comprehensive uncertainty results.

Conducts bootstrap resampling with full statistical analysis including:

Time series of contribution estimates
Posterior distributions of contributions
95% credible intervals (2.5th to 97.5th percentiles)
Mean and median contribution estimates

Parameters

results	Pointer to Results object to populate with analysis outputs
percentage	Percentage of source samples to exclude per iteration (0-100)
num_iterations	Number of bootstrap iterations
target_sample	Name of target sample to apportion
use_softmax	If true, use softmax transformation

Returns: true if analysis completed successfully

Note: Generates stacked bar chart for contributions (if iterations ≤ 100); Computes posterior distributions with 100 bins

Definition at line 3744 of file sourcesinkdata.cpp.

References Results::Append(), counter, distribution, GetContributionVector(), GetSourceOrder(), high, InitializeParametersAndObservations(), linear, low, normal, numberofsourcesamplesets_, RandomlyEliminateSourceSamples(), rangeset, rtw_, samplesetsorder_, Range::Set(), Range::SetMean(), Range::SetMedian(), ResultItem::SetName(), Results::SetName(), ProgressWindow::SetProgress(), ResultItem::SetResult(), ResultItem::SetShowAsString(), ResultItem::SetShowGraph(), ResultItem::SetShowTable(), ResultItem::SetType(), ResultItem::SetXAxisMode(), ResultItem::setXAxisTitle(), ResultItem::SetYAxisMode(), ResultItem::setYAxisTitle(), ResultItem::SetYLimit(), softmax, SolveLevenberg_Marquardt(), and stacked_bar_chart.

◆ BoxCoxTransformed()

SourceSinkData SourceSinkData::BoxCoxTransformed ( bool calculate_optimal_lambda = false )

Applies Box-Cox transformation to all source groups for normalization.

Transforms elemental concentrations using the Box-Cox power transformation to improve normality of distributions. This is useful for statistical analyses that assume normally distributed data (e.g., t-tests, ANOVA, discriminant analysis).

Box-Cox transformation: y = (x^λ - 1) / λ (or ln(x) if λ = 0)

The optimal λ parameter for each element is determined by maximum likelihood estimation to best achieve normality.

Parameters

calculate_optimal_lambda If true, compute optimal λ for each element; if false, use default transformation

Returns: New SourceSinkData object with transformed concentrations

Note: Target group is not transformed (only sources); Element distributions are recalculated after transformation; Original data is preserved (returns new object)

See also: OptimalBoxCoxParameters() for λ parameter computation

Definition at line 3117 of file sourcesinkdata.cpp.

References AssignAllDistributions(), OptimalBoxCoxParameters(), PopulateElementDistributions(), and target_group_.

Referenced by Conductor::ExecuteANOVA(), Conductor::ExecuteDFA(), Conductor::ExecuteDFAM(), Conductor::ExecuteDFAOnevsRest(), Conductor::ExecuteDistributionFitting(), Conductor::ExecuteEDP(), Conductor::ExecuteEDPM(), Conductor::ExecuteSDFA(), Conductor::ExecuteSDFAM(), and Conductor::ExecuteSDFAOnevsRest().

◆ BracketTest() [1/2]

CMBMatrix SourceSinkData::BracketTest	(	bool	correct_based_on_om_n_size,
		bool	exclude_elements,
		bool	exclude_samples
	)

Performs bracket test on all target samples.

Applies the bracket test to every sample in the target group, producing a matrix showing which elements in which samples fall outside source concentration ranges.

Matrix structure:

Rows: Elements
Columns: Target samples
Values: 0 = pass (within range), 1 = fail (outside range)

Parameters

correct_based_on_om_n_size	If true, apply OM/size corrections before testing
exclude_elements	If true, filter out elements marked for exclusion
exclude_samples	If true, filter out samples marked for exclusion

Returns: CMBMatrix of bracket test results for all target samples

Note: Each target sample is tested independently; Matrix dimensions: (num_elements × num_target_samples)

Definition at line 3067 of file sourcesinkdata.cpp.

References BracketTest(), CountElements(), CreateCorrectedAndFilteredDataset(), GetElementNames(), CMBMatrix::SetColumnLabel(), CMBMatrix::SetRowLabel(), CMBVector::size(), target_group_, and CMBVector::valueAt().

◆ BracketTest() [2/2]

CMBVector SourceSinkData::BracketTest	(	const string &	target_sample,
		bool	correct_based_on_om_n_size
	)

Performs bracket test to check if target concentrations fall within source ranges.

Tests whether each element's concentration in the target sample falls within the range (min to max) observed across all source samples. Elements outside source ranges indicate potential issues: missing sources, measurement errors, or non-conservative behavior.

Test criteria for each element:

Pass (0): Target concentration within [min_sources, max_sources]
Fail (1): Target concentration outside source ranges

Parameters

target_sample	Name of the target sample to test
correct_based_on_om_n_size	If true, apply OM/size corrections before testing

Returns: CMBVector of pass/fail flags (0 = pass, 1 = fail) for each element

Note: Flags are labeled with element names; Failed elements are documented in target sample notes; Test assumes conservative mixing (no gains/losses)

Definition at line 2994 of file sourcesinkdata.cpp.

References CreateCorrectedAndFilteredDataset(), GetElementDistribution(), GetElementNames(), ConcentrationSet::GetMaximum(), CMBVector::SetLabel(), and target_group_.

Referenced by BracketTest(), Conductor::ExecuteBracketingAnalysis(), and MainWindow::onIncludeExcludeSample().

◆ BuildSourceMeanMatrix()

CMatrix SourceSinkData::BuildSourceMeanMatrix ( parameter_mode param_mode = parameter_mode::based_on_fitted_distribution )

private

Builds the source mean concentration matrix for chemical elements.

Constructs a matrix where each entry [i,j] represents the mean concentration of element i in source j. Used in the mixing model equation: C_target = S × f

Parameters

param_mode Controls whether to use parametric or empirical means

Returns: Matrix of size (num_elements × num_sources) containing mean concentrations

Note: Elements are ordered according to element_order_ vector; Sources are ordered according to samplesetsorder_ vector

Definition at line 1492 of file sourcesinkdata.cpp.

References based_on_fitted_distribution, Distribution::DataMean(), element_order_, Elemental_Profile_Set::GetEstimatedDistribution(), GetSampleSet(), Distribution::Mean(), numberofsourcesamplesets_, and samplesetsorder_.

Referenced by PredictTarget(), and PredictTarget_Isotope_delta().

◆ BuildSourceMeanMatrix_Isotopes()

CMatrix SourceSinkData::BuildSourceMeanMatrix_Isotopes ( parameter_mode param_mode = parameter_mode::based_on_fitted_distribution )

private

Builds the source mean concentration matrix for isotopes.

Constructs a matrix where each entry [i,j] represents the mean absolute concentration of isotope i in source j. Converts from delta notation to absolute concentrations.

Conversion formula: [isotope] = (δ/1000 + 1) × R_standard × [base_element]

Parameters

param_mode Controls whether to use parametric or empirical means

Returns: Matrix of size (num_isotopes × num_sources)

Note: Mixing occurs in concentration space, not delta space

Definition at line 1530 of file sourcesinkdata.cpp.

References element_information::base_element, based_on_fitted_distribution, Distribution::DataMean(), element_information_, Elemental_Profile_Set::GetEstimatedDistribution(), GetSampleSet(), isotope_order_, Distribution::Mean(), numberofsourcesamplesets_, samplesetsorder_, and element_information::standard_ratio.

Referenced by PredictTarget_Isotope(), and PredictTarget_Isotope_delta().

◆ Clear()

void SourceSinkData::Clear ( )

Clear all data from the object.

Resets the SourceSinkData object to its initial empty state by:

Clearing all profile sets (sources and targets)
Clearing element information and distributions
Resetting all counters to zero
Clearing parameters and observations
Clearing all ordering vectors
Resetting file paths and identifiers to empty strings
Clearing tools usage history

Definition at line 332 of file sourcesinkdata.cpp.

References constituent_order_, element_distributions_, element_information_, element_order_, isotope_order_, numberofconstituents_, numberofisotopes_, numberofsourcesamplesets_, observations_, outputpath_, parameters_, samplesetsorder_, selected_target_sample_, size_om_order_, target_group_, and tools_used_.

Referenced by MainWindow::ReadExcel(), and ReadFromFile().

◆ ConstituentOrder()

vector< string > SourceSinkData::ConstituentOrder ( )

private

Retrieves the ordering of all constituents.

Returns: Vector of all constituent names

Definition at line 5090 of file sourcesinkdata.cpp.

References constituent_order_.

◆ CountElements()

int SourceSinkData::CountElements ( bool exclude_elements ) const

Count the number of elements in the dataset.

Counts elements based on their inclusion status and role. Can count either:

All elements (if exclude_elements = false)
Only elements marked for analysis, excluding OM, particle size, and do_not_include roles

Parameters

exclude_elements If true, only count elements marked for analysis with valid roles

Returns: Number of elements meeting the criteria

Definition at line 191 of file sourcesinkdata.cpp.

References element_information::do_not_include, element_information_, element_information::organic_carbon, and element_information::particle_size.

Referenced by BracketTest().

◆ CreateCorrectedAndFilteredDataset()

SourceSinkData SourceSinkData::CreateCorrectedAndFilteredDataset	(	bool	exclude_samples,
		bool	exclude_elements,
		bool	omnsizecorrect,
		const string &	target = `""`
	)		const

Create a corrected and filtered copy of the dataset.

Creates a new SourceSinkData object with optional filtering and corrections:

Can exclude samples not marked for analysis
Can exclude elements not marked for analysis
Can apply organic matter and particle size corrections
Target group is never corrected (only source groups)

This is the main method for preparing data before analysis, allowing fine control over which samples/elements to include and whether to apply normalizations.

Parameters

exclude_samples	If true, exclude samples not marked for analysis
exclude_elements	If true, exclude elements not marked for analysis
omnsizecorrect	If true, apply OM and particle size corrections to sources
target	Target sample name (empty to use current selected_target_sample_)

Returns: New corrected and filtered SourceSinkData object

Definition at line 213 of file sourcesinkdata.cpp.

References AssignAllDistributions(), element_information_, omconstituent_, options_, PopulateElementDistributions(), PopulateElementInformation(), regression_p_value_threshold_, selected_target_sample_, sizeconsituent_, and target_group_.

◆ CreateCorrectedDataset()

SourceSinkData SourceSinkData::CreateCorrectedDataset	(	const string &	target,
		bool	omnsizecorrect,
		map< string, element_information > *	elementinfo
	)

Create a corrected copy of the dataset for a specific target sample.

Creates a new SourceSinkData object with:

Only samples marked for inclusion in analysis
Optional organic matter and particle size corrections applied
Element information filtered based on analysis flags
All distributions populated and assigned

This method is typically used to prepare data for source apportionment of a specific target sample.

Parameters

target	Name of the target sample to analyze
omnsizecorrect	If true, apply OM and size corrections to source profiles
elementinfo	Element metadata for filtering (nullptr to use internal metadata)

Returns: New SourceSinkData object ready for analysis

Definition at line 111 of file sourcesinkdata.cpp.

References AssignAllDistributions(), constituent_order_, element_information::do_not_include, element_distributions_, element_information_, element_order_, isotope_order_, numberofconstituents_, numberofisotopes_, numberofsourcesamplesets_, observations_, omconstituent_, options_, element_information::organic_carbon, outputpath_, parameter_estimation_mode_, element_information::particle_size, PopulateElementDistributions(), regression_p_value_threshold_, samplesetsorder_, selected_target_sample_, size_om_order_, sizeconsituent_, and target_group_.

◆ DFA_eigvector()

CMBVector SourceSinkData::DFA_eigvector ( )

private

Computes the primary discriminant function eigenvector.

Calculates the eigenvector corresponding to the largest eigenvalue of S_W⁻¹ × S_B, which defines the discriminant function axis.

Returns: CMBVector containing discriminant function coefficients

Note: Returns empty vector if S_W is singular; Uses largest eigenvalue by absolute value

Definition at line 4664 of file sourcesinkdata.cpp.

References BetweenGroupCovarianceMatrix(), GetElementNames(), CMBVector::SetLabels(), and WithinGroupCovarianceMatrix().

Referenced by DFA_Projected(), DFA_Projected(), DFA_Projected(), DiscriminantFunctionAnalysis(), and DiscriminantFunctionAnalysis().

◆ DFA_P_Value()

double SourceSinkData::DFA_P_Value ( )

private

Computes p-value for discriminant function analysis.

Tests the null hypothesis that all source groups have the same mean elemental composition using Wilks' Lambda transformed to chi-squared.

Chi-squared transformation: χ² = -[N - 1 - (p + k)/2] × ln(Λ)

Returns: P-value for test of group equality

Note: P-value < 0.05 indicates significant differences

Definition at line 4568 of file sourcesinkdata.cpp.

References GetElementNames(), log, numberofsourcesamplesets_, target_group_, TotalNumberofSourceSamples(), and WilksLambda().

Referenced by DiscriminantFunctionAnalysis(), DiscriminantFunctionAnalysis(), and StepwiseDiscriminantFunctionAnalysis().

◆ DFA_Projected() [1/3]

CMBVectorSet SourceSinkData::DFA_Projected ( )

private

Projects all groups onto the discriminant function axis.

Computes discriminant scores for all samples in all groups by projecting elemental profiles onto the primary discriminant eigenvector.

Returns: CMBVectorSet with discriminant scores for each group

Definition at line 4599 of file sourcesinkdata.cpp.

References CMBVectorSet::Append(), and DFA_eigvector().

Referenced by DiscriminantFunctionAnalysis(), DiscriminantFunctionAnalysis(), and StepwiseDiscriminantFunctionAnalysis().

◆ DFA_Projected() [2/3]

CMBVectorSet SourceSinkData::DFA_Projected	(	const string &	source1,
		const string &	source2
	)

private

Projects two specific groups onto the discriminant function axis.

Parameters

source1	Name of first source group
source2	Name of second source group

Returns: CMBVectorSet with discriminant scores

Note: Current implementation projects all groups

Definition at line 4619 of file sourcesinkdata.cpp.

References CMBVectorSet::Append(), and DFA_eigvector().

◆ DFA_Projected() [3/3]

CMBVectorSet SourceSinkData::DFA_Projected	(	const string &	source1,
		SourceSinkData *	original
	)

private

Projects samples from original dataset onto discriminant function.

Uses discriminant function computed from current dataset to project samples from the original dataset. Useful for validating on held-out data.

Parameters

source1	Name of source used in discriminant function
original	Pointer to original dataset

Returns: CMBVectorSet with discriminant scores for original samples

Note: Target group from original dataset is excluded

Definition at line 4640 of file sourcesinkdata.cpp.

References CMBVectorSet::Append(), DFA_eigvector(), and target_group_.

◆ DFA_weight_vector()

CMBVector SourceSinkData::DFA_weight_vector	(	const string &	source1,
		const string &	source2
	)

private

Computes discriminant weight vector for two-group comparison.

Calculates the linear discriminant weights for separating two specific source groups using Fisher's linear discriminant analysis.

Formula: w = (μ₂ - μ₁) / (Σ₁ + Σ₂)

Parameters

source1	Name of first source group
source2	Name of second source group

Returns: Normalized CMBVector of discriminant weights (unit length)

Definition at line 4702 of file sourcesinkdata.cpp.

References GetElementNames(), and CMBVector::SetLabels().

◆ DFATransformed()

CMBVector SourceSinkData::DFATransformed	(	const CMBVector &	eigenvector,
		const string &	source_group
	)

private

Projects samples onto a discriminant function axis.

Transforms elemental profiles from a source group by computing their projections (dot products) with a discriminant eigenvector.

Projection formula: score_i = Σ(x_ij × w_j)

Parameters

eigenvector	Discriminant function coefficients
source_group	Name of the source group to project

Returns: CMBVector containing discriminant scores, labeled with sample names

Definition at line 2947 of file sourcesinkdata.cpp.

References CMBVector::SetLabel().

◆ DifferentiationPower() [1/2]

Elemental_Profile_Set SourceSinkData::DifferentiationPower	(	bool	use_log,
		bool	include_target
	)

Computes differentiation power for all source pairs.

Performs pairwise differentiation power analysis for all combinations of source groups, generating a comprehensive matrix of source separability.

Formula: D = 2 × |μ₁ - μ₂| / (σ₁ + σ₂)

Interpretation:

D > 2: Strong differentiation
D ≈ 1: Moderate differentiation
D < 0.5: Weak differentiation

Parameters

use_log	If true, compute using log-space statistics; if false, use linear-space statistics
include_target	If true, include target group in comparisons

Returns: Elemental_Profile_Set with differentiation profiles for each source pair

Definition at line 3296 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::AppendProfile(), DifferentiationPower(), and target_group_.

Referenced by DifferentiationPower(), Conductor::ExecuteEDP(), and Conductor::ExecuteEDPM().

◆ DifferentiationPower() [2/2]

Elemental_Profile SourceSinkData::DifferentiationPower	(	const string &	source1,
		const string &	source2,
		bool	use_log
	)

Computes differentiation power metric between two sources.

Calculates a standardized measure of how well each element differentiates between two sources. The metric quantifies separation in units of pooled standard deviations.

Formula: D = 2 × |μ₁ - μ₂| / (σ₁ + σ₂)

Interpretation:

D > 2: Strong differentiation (means differ by >1 pooled std dev)
D ≈ 1: Moderate differentiation
D < 0.5: Weak differentiation (substantial overlap)

Parameters

source1	Name of first source group
source2	Name of second source group
use_log	If true, compute using log-space statistics; if false, use linear-space statistics

Returns: Elemental_Profile containing differentiation power for each element

Note: Higher values indicate better discrimination ability; Complementary to p-values (considers effect size, not just significance)

See also: DifferentiationPower(bool, bool) for all pairwise comparisons

Definition at line 3256 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), ConcentrationSet::CalculateMean(), ConcentrationSet::CalculateMeanLog(), ConcentrationSet::CalculateStdDev(), ConcentrationSet::CalculateStdDevLog(), and GetElementNames().

◆ DifferentiationPower_P_value()

Elemental_Profile_Set SourceSinkData::DifferentiationPower_P_value ( bool include_target )

Computes t-test p-values for all source pairs.

Performs pairwise t-test analysis for all combinations of source groups, generating a comprehensive matrix of statistical significance values.

Parameters

include_target If true, include target group in comparisons

Returns: Elemental_Profile_Set with p-value profiles for each source pair

Note: Lower p-values indicate statistically significant differences

Definition at line 3379 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::AppendProfile(), t_TestPValue(), and target_group_.

Referenced by Conductor::ExecuteAutoSelect(), and Conductor::ExecuteEDPM().

◆ DifferentiationPower_Percentage() [1/2]

Elemental_Profile_Set SourceSinkData::DifferentiationPower_Percentage ( bool include_target )

Computes rank-based differentiation percentage for all source pairs.

Performs pairwise rank-based differentiation analysis for all combinations of source groups. Provides a non-parametric measure of source separability.

Parameters

include_target If true, include target group in comparisons

Returns: Elemental_Profile_Set with differentiation percentages for each source pair

Note: Values near 1.0 indicate excellent separation; Values near 0.5 indicate no separation

Definition at line 3356 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::AppendProfile(), DifferentiationPower_Percentage(), and target_group_.

Referenced by DifferentiationPower_Percentage(), Conductor::ExecuteEDP(), and Conductor::ExecuteEDPM().

◆ DifferentiationPower_Percentage() [2/2]

Elemental_Profile SourceSinkData::DifferentiationPower_Percentage	(	const string &	source1,
		const string &	source2
	)

Computes rank-based differentiation percentage between two sources.

Calculates the percentage of samples that can be correctly classified based on element concentrations using a rank-based approach. Provides a distribution-free measure of source separability.

Method:

Pool samples from both sources
Rank all concentrations for each element
Count correct classifications (Source1 samples with low ranks + Source2 samples with high ranks, or vice versa)
Return classification success percentage

Parameters

source1	Name of first source group
source2	Name of second source group

Returns: Elemental_Profile containing classification percentages (0-1) for each element

Note: Values near 1.0 indicate excellent separation; Values near 0.5 indicate no separation (random classification); Non-parametric (no distributional assumptions)

See also: DifferentiationPower_Percentage(bool) for all pairwise comparisons

Definition at line 3319 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), ConcentrationSet::AppendSet(), ConcentrationSet::CalculateRanks(), and GetElementNames().

◆ DiscriminantFunctionAnalysis() [1/3]

DFA_result SourceSinkData::DiscriminantFunctionAnalysis ( )

Performs discriminant function analysis for all source groups.

Conducts pairwise discriminant function analysis (DFA) comparing each source group against all others combined. DFA identifies the linear combination of elemental concentrations that best separates each source from the rest, which helps assess source uniqueness and potential confounding.

For each source, computes:

Wilks' Lambda (measure of group separation)
F-test p-value (significance of separation)
Discriminant p-value (multivariate normality test)
Eigenvector (discriminant function coefficients)
Projected samples (scores along discriminant axis)

Returns: DFA_result containing statistics and projections for all sources

Note: Target group is excluded from analysis; Each source is compared against all others combined ("one vs rest"); Lower Wilks' Lambda indicates better separation

Definition at line 2814 of file sourcesinkdata.cpp.

References CMBVectorSet::Append(), CMBVectorSetSet::Append(), counter, DiscriminantFunctionAnalysis(), DFA_result::eigen_vectors, DFA_result::F_test_P_value, DFA_result::multi_projected, DFA_result::p_values, DFA_result::projected, CMBVector::SetLabel(), target_group_, and DFA_result::wilkslambda.

Referenced by DiscriminantFunctionAnalysis(), Conductor::ExecuteDFA(), Conductor::ExecuteDFAM(), Conductor::ExecuteDFAOnevsRest(), StepwiseDiscriminantFunctionAnalysis(), and StepwiseDiscriminantFunctionAnalysis().

◆ DiscriminantFunctionAnalysis() [2/3]

DFA_result SourceSinkData::DiscriminantFunctionAnalysis ( const string & source1 )

Performs discriminant function analysis for one source vs all others.

Conducts DFA comparing a single source group against all other sources combined ("one vs rest" comparison).

Parameters

source1 Name of source group to test

Returns: DFA_result containing separation statistics and discriminant function

Note: All other source groups are pooled into an "Others" category

Definition at line 2894 of file sourcesinkdata.cpp.

References CMBVectorSet::Append(), AppendSampleSet(), DFA_eigvector(), DFA_P_Value(), DFA_Projected(), DFA_result::eigen_vectors, element_information_, DFA_result::F_test_P_value, CMBVectorSet::FTest_p_value(), DFA_result::p_values, PopulateElementInformation(), DFA_result::projected, TheRest(), DFA_result::wilkslambda, and WilksLambda().

◆ DiscriminantFunctionAnalysis() [3/3]

DFA_result SourceSinkData::DiscriminantFunctionAnalysis	(	const string &	source1,
		const string &	source2
	)

Performs discriminant function analysis between two specific sources.

Conducts pairwise DFA to assess how well two source groups can be distinguished based on their elemental compositions.

Parameters

source1	Name of first source group
source2	Name of second source group

Returns: DFA_result containing separation statistics and discriminant function

Note: Creates temporary dataset with only the two specified sources; Returns empty result if eigenvector computation fails

Definition at line 2854 of file sourcesinkdata.cpp.

References CMBVectorSet::Append(), AppendSampleSet(), DFA_eigvector(), DFA_P_Value(), DFA_Projected(), DFA_result::eigen_vectors, element_information_, DFA_result::F_test_P_value, CMBVectorSet::FTest_p_value(), DFA_result::p_values, PopulateElementInformation(), DFA_result::projected, CMBVector::size(), DFA_result::wilkslambda, and WilksLambda().

◆ ElementDataToJsonObject()

QJsonObject SourceSinkData::ElementDataToJsonObject ( ) const

Exports all elemental profile data to a JSON object.

Serializes all sample groups and their elemental profiles to JSON format.

Returns: QJsonObject containing all elemental profile sets

Definition at line 2673 of file sourcesinkdata.cpp.

Referenced by MainWindow::buildProjectJson().

◆ ElementInformationToJsonObject()

QJsonObject SourceSinkData::ElementInformationToJsonObject ( ) const

Exports element information metadata to a JSON object.

Serializes all element metadata (role, standard ratio, base element, inclusion flag) to a JSON object for saving or transmission.

Returns: QJsonObject containing all element information

Definition at line 2726 of file sourcesinkdata.cpp.

References element_information_, and Role().

Referenced by MainWindow::buildProjectJson().

◆ ElementOrder()

vector< string > SourceSinkData::ElementOrder ( )

private

Retrieves the ordering of chemical elements.

Returns: Vector of element names in analysis order

Definition at line 5095 of file sourcesinkdata.cpp.

References element_order_.

Referenced by GetObservedElementalProfile(), GetPredictedElementalProfile(), and MCMC().

◆ ElementsToBeUsedInCMB()

vector< string > SourceSinkData::ElementsToBeUsedInCMB ( )

Identifies chemical elements to be used in CMB analysis.

Scans the element information map and collects all elements marked with the 'element' role. Updates constituent count and ordering vectors.

Returns: Vector of element names to be included in CMB analysis

Note: Updates constituent_order_ and numberofconstituents_

Definition at line 2070 of file sourcesinkdata.cpp.

References constituent_order_, element_information::element, element_information_, and numberofconstituents_.

◆ ExtractChemicalElements()

SourceSinkData SourceSinkData::ExtractChemicalElements ( bool isotopes ) const

Extract only chemical elements (and optionally isotopes)

Creates a new SourceSinkData containing only elements with role 'element' (and optionally 'isotope'), excluding particle size, organic matter, and other non-element constituents.

Parameters

isotopes If true, also include isotope ratios

Returns: New SourceSinkData with only chemical elements

Definition at line 287 of file sourcesinkdata.cpp.

References AssignAllDistributions(), element_information_, ExtractChemicalElements(), omconstituent_, PopulateElementDistributions(), PopulateElementInformation(), regression_p_value_threshold_, sizeconsituent_, and target_group_.

Referenced by Conductor::ExecuteAutoSelect(), and ExtractChemicalElements().

◆ ExtractConcentrationData()

profiles_data SourceSinkData::ExtractConcentrationData ( const vector< vector< string > > & indicators ) const

Extract concentration data for specified samples.

Extracts elemental concentration data for a list of samples identified by [group_name, sample_name] pairs. Returns data in a structured format suitable for analysis.

Parameters

indicators Vector of [group_name, sample_name] pairs identifying samples

Returns: profiles_data structure containing element names, sample names, and concentration values

Definition at line 445 of file sourcesinkdata.cpp.

References profiles_data::element_names, Elemental_Profile_Set::GetConcentrationsForSample(), GetElementNames(), profiles_data::sample_names, and profiles_data::values.

Referenced by MainWindow::on_tree_selectionChanged().

◆ ExtractConcentrationSet()

map< string, ConcentrationSet > SourceSinkData::ExtractConcentrationSet ( )

Extracts concentration distributions for all elements across sources.

Collects all concentration values for each element across all source samples (excluding target) into ConcentrationSet objects. This aggregates data for statistical analysis of elemental distributions across the entire source dataset.

Returns: Map of element names to ConcentrationSet objects containing all source sample concentrations for that element

Note: Target group samples are excluded; Each ConcentrationSet contains values from all sources combined

Definition at line 3151 of file sourcesinkdata.cpp.

References ConcentrationSet::AppendValue(), GetElementNames(), and target_group_.

Referenced by OptimalBoxCoxParameters().

◆ ExtractElementConcentrations()

element_data SourceSinkData::ExtractElementConcentrations	(	const string &	element,
		const string &	group
	)		const

Extract concentration data for a specific element from a group.

Retrieves all concentration values for a single element across all samples within a specified group. Useful for univariate analysis of a single element's distribution within a source or target group.

Parameters

element	Element name to extract
group	Group name to extract from

Returns: element_data structure containing group name, sample names, and concentration values

Definition at line 495 of file sourcesinkdata.cpp.

References element_data::group_name, element_data::sample_names, and element_data::values.

Referenced by ExtractElementDataByGroup().

◆ ExtractElementDataByGroup()

map< string, vector< double > > SourceSinkData::ExtractElementDataByGroup ( const string & element ) const

Extract concentration data for a specific element from all groups.

Retrieves all concentration values for a single element organized by group. Returns a map where keys are group names and values are vectors of concentrations for that element within each group.

Parameters

element Element name to extract

Returns: Map of group_name -> vector of concentration values

Definition at line 542 of file sourcesinkdata.cpp.

References ExtractElementConcentrations(), GetGroupNames(), and element_data::values.

Referenced by MainWindow::on_tree_selectionChanged().

◆ ExtractSamplesAsProfileSet()

Elemental_Profile_Set SourceSinkData::ExtractSamplesAsProfileSet ( const vector< vector< string > > & indicators ) const

Extract samples as an Elemental_Profile_Set.

Extracts specified samples and combines them into a single profile set. Sample names are prefixed with their group name (e.g., "Agricultural-Sample1") to ensure uniqueness when combining samples from different groups.

Parameters

indicators Vector of [group_name, sample_name] pairs identifying samples

Returns: Elemental_Profile_Set containing the extracted samples

Definition at line 472 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::AppendProfile().

Referenced by MainWindow::on_tree_selectionChanged().

◆ ExtractSpecificElements()

SourceSinkData SourceSinkData::ExtractSpecificElements ( const vector< string > & element_list ) const

Extract specific elements from source groups.

Creates a new SourceSinkData containing only the specified elements from source groups. Target group is excluded. Useful for subset analysis or element selection optimization.

Parameters

element_list List of element names to extract

Returns: New SourceSinkData with only specified elements from sources

Definition at line 311 of file sourcesinkdata.cpp.

References AssignAllDistributions(), element_information_, PopulateElementDistributions(), PopulateElementInformation(), and target_group_.

Referenced by StepwiseDiscriminantFunctionAnalysis(), StepwiseDiscriminantFunctionAnalysis(), and StepwiseDiscriminantFunctionAnalysis().

◆ FirstOMConstituent()

string SourceSinkData::FirstOMConstituent ( )

Retrieves the name of the first organic matter constituent.

Searches element information for the first element designated as organic carbon/matter.

Returns: Name of first OM constituent, or empty string if none found

Note: Returns first match only

Definition at line 4416 of file sourcesinkdata.cpp.

References element_information_, and element_information::organic_carbon.

◆ FirstSizeConstituent()

string SourceSinkData::FirstSizeConstituent ( )

Retrieves the name of the first particle size constituent.

Searches element information for the first element designated as particle size.

Returns: Name of first size constituent, or empty string if none found

Note: Returns first match only

Definition at line 4433 of file sourcesinkdata.cpp.

References element_information_, and element_information::particle_size.

◆ GetCalculatedElementMeans()

ResultItem SourceSinkData::GetCalculatedElementMeans ( )

Computes estimated mean concentrations for all source elements.

Calculates the mean elemental concentrations for each source group using the current estimated distributions.

Returns: ResultItem containing Elemental_Profile_Set with mean profiles for each source

Note: Only includes source groups (target excluded); Uses estimated distributions (updated during optimization/MCMC)

Definition at line 2396 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), Elemental_Profile_Set::AppendProfile(), element_order_, elemental_profile_set, ResultItem::SetName(), ResultItem::SetResult(), ResultItem::SetType(), and target_group_.

◆ GetCalculatedElementMu()

ResultItem SourceSinkData::GetCalculatedElementMu ( )

Computes μ parameters from fitted log-normal distributions for source elements.

Calculates the μ (log-space mean) parameter for each element in each source group based on the fitted log-normal distributions from source sample data.

Returns: ResultItem containing Elemental_Profile_Set with μ values for each source

Note: Uses CalculateMeanLog() - computed from fitted distribution

Definition at line 2479 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), Elemental_Profile_Set::AppendProfile(), element_order_, elemental_profile_set, ResultItem::SetName(), ResultItem::SetResult(), ResultItem::SetType(), and target_group_.

◆ GetCalculatedElementSigma()

ResultItem SourceSinkData::GetCalculatedElementSigma ( )

Computes estimated standard deviations for all source elements.

Calculates the standard deviations of elemental concentrations for each source group using the current estimated distributions.

Returns: ResultItem containing Elemental_Profile_Set with std dev profiles for each source

Note: Different calculations for normal vs log-normal distributions

Definition at line 2420 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), Elemental_Profile_Set::AppendProfile(), element_order_, elemental_profile_set, lognormal, normal, ResultItem::SetName(), ResultItem::SetResult(), ResultItem::SetType(), and target_group_.

◆ GetContribution()

ResultItem SourceSinkData::GetContribution ( )

Packages source contributions into a ResultItem for output.

Creates a ResultItem containing the current source contribution estimates, formatted for export or display.

Returns: ResultItem of type 'contribution' containing source fractions

Note: Creates new Contribution object on heap (managed by ResultItem)

Definition at line 2198 of file sourcesinkdata.cpp.

References contribution, GetContributionVector(), GetSourceOrder(), ResultItem::SetName(), ResultItem::SetResult(), and ResultItem::SetType().

Referenced by Conductor::ExecuteLevenbergMarquardt().

◆ GetContributionVector()

CVector SourceSinkData::GetContributionVector ( bool include_all = true )

private

Retrieves the current source contribution fractions.

Returns a vector of contribution fractions from each source. Contributions represent the fraction of the target sample originating from each source, subject to the constraint: Σ f_i = 1.

Parameters

include_all If true, returns all n contributions including the constrained one. If false, returns only the n-1 independent contributions used in optimization.

Returns: Vector of contribution fractions, size n or n-1

Note: All contributions satisfy: 0 ≤ f_i ≤ 1 and Σ f_i = 1

Definition at line 1583 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::GetContribution(), GetSampleSet(), numberofsourcesamplesets_, and samplesetsorder_.

Referenced by BootStrap(), BootStrap(), GetContribution(), InitializeContributionsRandomly(), LM_Batch(), PredictTarget(), PredictTarget_Isotope(), PredictTarget_Isotope_delta(), ResidualJacobian(), ResidualJacobian_arma(), SetContribution(), SetContributionSoftmax(), SetParameterValue(), SolveLevenberg_Marquardt(), and VerifySource().

◆ GetContributionVectorSoftmax()

CVector SourceSinkData::GetContributionVectorSoftmax ( )

private

Retrieves the softmax parameters for source contributions.

Returns the unconstrained softmax parameters x_i that are transformed to contribution fractions via: f_i = exp(x_i) / Σ exp(x_j)

Returns: Vector of softmax parameters, size n (all sources)

Note: Softmax parameters are unbounded: x_i ∈ (-∞, +∞)

Definition at line 1603 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::GetContributionSoftmax(), GetSampleSet(), numberofsourcesamplesets_, and samplesetsorder_.

Referenced by ResidualJacobian_softmax(), and SolveLevenberg_Marquardt().

◆ GetElementalProfile()

Elemental_Profile * SourceSinkData::GetElementalProfile ( const string & sample_name )

private

Finds and retrieves an elemental profile by sample name.

Searches all groups (sources and target) for a sample with the given name and returns a pointer to its elemental profile. Useful when the group containing the sample is unknown.

Parameters

sample_name Name of the sample to find

Returns: Pointer to the elemental profile if found, nullptr otherwise

Note: Searches all groups sequentially until sample is found; Returns pointer to internal data - do not delete

Definition at line 2178 of file sourcesinkdata.cpp.

Referenced by InitializeParametersAndObservations().

◆ GetElementDistribution() [1/2]

ConcentrationSet * SourceSinkData::GetElementDistribution ( const string & element_name )

Retrieves pointer to element distribution at dataset level.

Parameters

element_name Name of the element

Returns: Pointer to ConcentrationSet, or nullptr if not found

Definition at line 5055 of file sourcesinkdata.cpp.

References element_distributions_.

Referenced by BracketTest(), Conductor::ExecuteDistributionFitting(), Conductor::ExecuteKolmogorovSmirnovIndividual(), GetElementDistribution(), and SetParameterValue().

◆ GetElementDistribution() [2/2]

ConcentrationSet * SourceSinkData::GetElementDistribution	(	const string &	element_name,
		const string &	sample_group
	)

Retrieves pointer to element distribution for a specific group.

Parameters

element_name	Name of the element
sample_group	Name of the source or target group

Returns: Pointer to ConcentrationSet, or nullptr if not found

Definition at line 5063 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::GetElementDistribution(), GetElementDistribution(), and GetSampleSet().

◆ GetElementDistributionMuParameter()

Parameter * SourceSinkData::GetElementDistributionMuParameter	(	size_t	element_index,
		size_t	source_index
	)

private

Retrieves pointer to the μ (mean) parameter for an element distribution.

Accesses the parameter object representing the mean (μ) of the log-normal distribution for a specific element in a specific source.

Parameters

element_index	Index of the element in element_order_ (0-based)
source_index	Index of the source in samplesetsorder_ (0-based)

Returns: Pointer to the Parameter object, or nullptr if indices are out of bounds

Note: For log-normal distributions: actual mean = exp(μ + σ²/2)

Definition at line 1690 of file sourcesinkdata.cpp.

References numberofsourcesamplesets_, and parameters_.

Referenced by GetElementDistributionMuValue().

◆ GetElementDistributionMuValue()

double SourceSinkData::GetElementDistributionMuValue	(	size_t	element_index,
		size_t	source_index
	)

private

Retrieves the current value of the μ parameter for an element distribution.

Parameters

element_index	Index of the element in element_order_ (0-based)
source_index	Index of the source in samplesetsorder_ (0-based)

Returns: Current μ value, or 0.0 if parameter cannot be retrieved

Definition at line 1743 of file sourcesinkdata.cpp.

References GetElementDistributionMuParameter(), and Parameter::Value().

◆ GetElementDistributionSigmaParameter()

Parameter * SourceSinkData::GetElementDistributionSigmaParameter	(	size_t	element_index,
		size_t	source_index
	)

private

Retrieves pointer to the σ (std dev) parameter for an element distribution.

Accesses the parameter object representing the standard deviation (σ) of the log-normal distribution for a specific element in a specific source.

Parameters

element_index	Index of the element in element_order_ (0-based)
source_index	Index of the source in samplesetsorder_ (0-based)

Returns: Pointer to the Parameter object, or nullptr if indices are out of bounds

Note: σ represents log-space standard deviation for log-normal distributions

Definition at line 1716 of file sourcesinkdata.cpp.

References numberofconstituents_, numberofsourcesamplesets_, and parameters_.

Referenced by GetElementDistributionSigmaValue().

◆ GetElementDistributionSigmaValue()

double SourceSinkData::GetElementDistributionSigmaValue	(	size_t	element_index,
		size_t	source_index
	)

private

Retrieves the current value of the σ parameter for an element distribution.

Parameters

element_index	Index of the element in element_order_ (0-based)
source_index	Index of the source in samplesetsorder_ (0-based)

Returns: Current σ value, or 0.0 if parameter cannot be retrieved

Definition at line 1758 of file sourcesinkdata.cpp.

References GetElementDistributionSigmaParameter(), and Parameter::Value().

◆ GetElementInformation() [1/2]

map< string, element_information > * SourceSinkData::GetElementInformation ( )

Retrieves pointer to the element information map.

Provides access to element metadata including roles, standard ratios, base elements, and inclusion flags for all elements in the dataset.

Returns: Pointer to element information map

Definition at line 5042 of file sourcesinkdata.cpp.

References element_information_.

Referenced by ElementTableDelegate::createEditor(), ElementTableModel::data(), LM_Batch(), MCMC(), ElementTableModel::setData(), ElementTableDelegate::setEditorData(), and VerifySource().

◆ GetElementInformation() [2/2]

element_information * SourceSinkData::GetElementInformation ( const string & element_name )

Retrieves pointer to element information metadata by name.

Parameters

element_name Name of the element

Returns: Pointer to element_information, or nullptr if not found

Definition at line 5047 of file sourcesinkdata.cpp.

References element_information_.

◆ GetElementNames()

vector< string > SourceSinkData::GetElementNames ( ) const

Get all element names in the dataset.

Returns element names by examining the first sample of the first group. Assumes all samples have the same elements (which should be true after proper data loading).

Returns: Vector of element names, or empty vector if dataset is empty

Definition at line 420 of file sourcesinkdata.cpp.

Referenced by GenericForm::GenericForm(), ANOVA(), AssignAllDistributions(), BetweenGroupCovarianceMatrix(), BracketTest(), BracketTest(), ElementTableDelegate::createEditor(), ElementTableModel::data(), DFA_eigvector(), DFA_P_Value(), DFA_weight_vector(), DifferentiationPower(), DifferentiationPower_Percentage(), MainWindow::ElementsToQStandardItem(), ExtractConcentrationData(), ExtractConcentrationSet(), MeanElementalContent(), MeanElementalContent(), MainWindow::on_tool_executed(), OptimalBoxCoxParameters(), PopulateElementDistributions(), PopulateElementInformation(), ElementTableModel::rowCount(), ElementTableModel::setData(), StepwiseDiscriminantFunctionAnalysis(), StepwiseDiscriminantFunctionAnalysis(), StepwiseDiscriminantFunctionAnalysis(), t_TestPValue(), TotalScatterMatrix(), and WithinGroupCovarianceMatrix().

◆ GetEstimatedElementMean()

ResultItem SourceSinkData::GetEstimatedElementMean ( )

Computes actual mean concentrations from estimated log-normal parameters.

Calculates the actual concentration means for each element in each source using the estimated μ and σ parameters from Bayesian inference.

Formula: Mean = exp(μ + σ²/2)

Returns: ResultItem containing Elemental_Profile_Set with mean concentrations

Definition at line 2527 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), Elemental_Profile_Set::AppendProfile(), element_order_, elemental_profile_set, ResultItem::SetName(), ResultItem::SetResult(), ResultItem::SetType(), and target_group_.

◆ GetEstimatedElementMu()

ResultItem SourceSinkData::GetEstimatedElementMu ( )

Retrieves estimated μ parameters from Bayesian inference for source elements.

Returns the inferred μ (log-space mean) parameters for each element in each source group as estimated during MCMC sampling or optimization.

Returns: ResultItem containing Elemental_Profile_Set with inferred μ values

Note: Uses GetEstimatedMu() - updated during MCMC/optimization

Definition at line 2503 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), Elemental_Profile_Set::AppendProfile(), element_order_, elemental_profile_set, ResultItem::SetName(), ResultItem::SetResult(), ResultItem::SetType(), and target_group_.

◆ GetEstimatedElementSigma()

ResultItem SourceSinkData::GetEstimatedElementSigma ( )

Retrieves estimated σ parameters from Bayesian inference for source elements.

Returns the inferred σ (log-space standard deviation) parameters for each element in each source group as estimated during MCMC sampling or optimization.

Returns: ResultItem containing Elemental_Profile_Set with inferred σ values

Note: Values are in log-space (not actual concentration std dev)

Definition at line 2553 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), Elemental_Profile_Set::AppendProfile(), element_order_, elemental_profile_set, ResultItem::SetName(), ResultItem::SetResult(), ResultItem::SetType(), and target_group_.

◆ GetFittedDistribution()

Distribution * SourceSinkData::GetFittedDistribution ( const string & element_name )

Get the fitted distribution for a specific element at dataset level.

Returns a pointer to the Distribution object fitted to all data for this element (across all groups). This is the distribution used for likelihood calculations in MCMC.

Parameters

element_name Name of the element

Returns: Pointer to fitted Distribution, or nullptr if element not found

Definition at line 596 of file sourcesinkdata.cpp.

References element_distributions_.

Referenced by MainWindow::showdistributionsforelements().

◆ GetGroupNames()

vector< string > SourceSinkData::GetGroupNames ( ) const

Get all group names in the dataset.

Returns the names of all sample sets (source and target groups). These are the top-level keys of the SourceSinkData map.

Returns: Vector of all group names

Definition at line 407 of file sourcesinkdata.cpp.

Referenced by MainWindow::ElementsToQStandardItem(), ExtractElementDataByGroup(), MainWindow::ToQStandardItem(), and MainWindow::ToQStandardItemModel().

◆ GetMLRResults()

vector< ResultItem > SourceSinkData::GetMLRResults ( )

Retrieves multiple linear regression results for all sample groups.

Collects the organic matter (OM) and particle size regression models for each source group. These regressions are used to correct elemental concentrations for variations in OM content and particle size distribution.

Returns: Vector of ResultItems, one per sample group, containing MLR models

Note: Each ResultItem is configured for table display; Includes both source groups and target group; Regressions must be computed first via PerformRegressionVsOMAndSize()

Definition at line 2305 of file sourcesinkdata.cpp.

References ResultItem::SetName(), and ResultItem::SetShowTable().

Referenced by Conductor::ExecuteMLR().

◆ GetObjectiveFunctionValue()

double SourceSinkData::GetObjectiveFunctionValue ( )

Returns the objective function value for optimization algorithms.

Computes the negative log-likelihood based on the current parameter values and estimation mode. Used by optimization algorithms where minimizing -log(L) is equivalent to maximizing likelihood L.

Returns: Negative log-likelihood value (lower values indicate better fit)

Definition at line 1451 of file sourcesinkdata.cpp.

References LogLikelihood(), and parameter_estimation_mode_.

◆ GetObservedElementalProfile()

ResultItem SourceSinkData::GetObservedElementalProfile ( )

Retrieves the observed elemental concentrations for the selected target sample.

Extracts the measured elemental composition of the currently selected target sample. These are the observed values that the model attempts to reproduce.

Returns: ResultItem containing the observed Elemental_Profile for chemical elements

Note: Only includes chemical elements, not isotopes

Definition at line 2350 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), ElementOrder(), ObservedDataforSelectedSample(), predicted_concentration, selected_target_sample_, ResultItem::SetName(), ResultItem::SetResult(), and ResultItem::SetType().

Referenced by GetObservedvsModeledElementalProfile().

◆ GetObservedElementalProfile_Isotope()

ResultItem SourceSinkData::GetObservedElementalProfile_Isotope ( )

Retrieves the observed isotope delta values for the selected target sample.

Extracts the measured isotopic composition (δ values in ‰) of the currently selected target sample.

Returns: ResultItem containing the observed Elemental_Profile with isotope δ values

Note: Only includes isotopes, not chemical elements; Values are in delta notation (‰)

Definition at line 2373 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), IsotopeOrder(), ObservedDataforSelectedSample_Isotope_delta(), predicted_concentration, selected_target_sample_, ResultItem::SetName(), ResultItem::SetResult(), and ResultItem::SetType().

Referenced by GetObservedvsModeledElementalProfile_Isotope().

◆ GetObservedvsModeledElementalProfile()

ResultItem SourceSinkData::GetObservedvsModeledElementalProfile ( parameter_mode param_mode = parameter_mode::based_on_fitted_distribution )

Creates a comparison of observed vs modeled elemental profiles.

Generates a profile set containing both the observed target sample composition and the model's predicted composition for side-by-side comparison.

Parameters

param_mode Controls whether to use parametric or empirical source means

Returns: ResultItem containing Elemental_Profile_Set with "Observed" and "Modeled" profiles

Note: Only includes chemical elements, not isotopes

Definition at line 2280 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::AppendProfile(), elemental_profile_set, GetObservedElementalProfile(), GetPredictedElementalProfile(), ResultItem::Result(), ResultItem::SetName(), ResultItem::SetResult(), and ResultItem::SetType().

Referenced by Conductor::ExecuteLevenbergMarquardt().

◆ GetObservedvsModeledElementalProfile_Isotope()

ResultItem SourceSinkData::GetObservedvsModeledElementalProfile_Isotope ( parameter_mode param_mode = parameter_mode::based_on_fitted_distribution )

Creates a comparison of observed vs modeled isotope delta values.

Generates a profile set containing both the observed target sample isotopic composition and the model's predicted isotope delta values.

Parameters

param_mode Controls whether to use parametric or empirical source means

Returns: ResultItem containing Elemental_Profile_Set with "Observed" and "Modeled" isotope profiles

Note: Only includes isotopes, not chemical elements; All values are in delta notation (‰)

Definition at line 2325 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::AppendProfile(), elemental_profile_set, GetObservedElementalProfile_Isotope(), GetPredictedElementalProfile_Isotope(), ResultItem::Result(), ResultItem::SetName(), ResultItem::SetResult(), and ResultItem::SetType().

Referenced by Conductor::ExecuteLevenbergMarquardt().

◆ GetOptions()

QMap< QString, double > * SourceSinkData::GetOptions ( )

Retrieves pointer to the options map.

Returns: Pointer to QMap containing configuration options

Definition at line 5141 of file sourcesinkdata.cpp.

References options_.

Referenced by MainWindow::onIncludeExcludeSample().

◆ GetOutputPath()

string SourceSinkData::GetOutputPath ( ) const

Get the output directory path.

Returns the path where analysis results and output files will be saved.

Returns: Current output path

Definition at line 5148 of file sourcesinkdata.cpp.

References outputpath_.

◆ GetParameterName()

string SourceSinkData::GetParameterName ( int index ) const

Get the name of a parameter by its index.

Retrieves the descriptive name of an optimization parameter (e.g., "Agricultural_contribution", "Urban_Al_mu") by its position in the parameters vector.

Parameters

index Parameter index (0-based)

Returns: Parameter name, or empty string if index invalid

Definition at line 635 of file sourcesinkdata.cpp.

References Parameter::Name(), and parameter().

◆ GetParameterValue() [1/2]

CVector SourceSinkData::GetParameterValue ( ) const

private

Retrieves all current parameter values as a vector.

Returns a vector containing all parameter values in the order they appear in the parameters_ vector. The parameter layout depends on estimation_mode_.

Returns: Vector of all parameter values

Definition at line 1950 of file sourcesinkdata.cpp.

References parameters_.

Referenced by GradientUpdate().

◆ GetParameterValue() [2/2]

double SourceSinkData::GetParameterValue ( size_t index ) const

private

Retrieves a single parameter value by index.

Parameters

index Parameter index in the parameters_ vector

Returns: Parameter value at the specified index

Note: No bounds checking - ensure index is valid

Definition at line 1929 of file sourcesinkdata.cpp.

References parameters_.

◆ GetPredictedElementalProfile()

ResultItem SourceSinkData::GetPredictedElementalProfile ( parameter_mode param_mode = parameter_mode::based_on_fitted_distribution )

Generates predicted elemental concentrations for the target sample.

Computes the model's predicted elemental composition using the current source contributions and profiles.

Parameters

param_mode Controls whether to use parametric or empirical source means

Returns: ResultItem containing the predicted Elemental_Profile

Note: Only includes chemical elements, not isotopes

Definition at line 2220 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), ElementOrder(), predicted_concentration, PredictTarget(), ResultItem::SetName(), ResultItem::SetResult(), and ResultItem::SetType().

Referenced by Conductor::ExecuteLevenbergMarquardt(), and GetObservedvsModeledElementalProfile().

◆ GetPredictedElementalProfile_Isotope()

ResultItem SourceSinkData::GetPredictedElementalProfile_Isotope ( parameter_mode param_mode = parameter_mode::based_on_fitted_distribution )

Generates predicted isotope delta values for the target sample.

Computes the model's predicted isotopic composition (δ values in ‰) using current source contributions and isotope profiles.

Parameters

param_mode Controls whether to use parametric or empirical source means

Returns: ResultItem containing the predicted Elemental_Profile with isotope δ values

Note: Only includes isotopes, not chemical elements; Values are in delta notation (‰)

Definition at line 2257 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), IsotopeOrder(), predicted_concentration, PredictTarget_Isotope_delta(), ResultItem::SetName(), ResultItem::SetResult(), and ResultItem::SetType().

Referenced by GetObservedvsModeledElementalProfile_Isotope().

◆ GetPredictedValues()

CVector SourceSinkData::GetPredictedValues ( )

Retrieves predicted values for all observations.

Collects the predicted values from all observation objects for comparison with actual observations.

Returns: Vector of predicted values matching the observations vector

Definition at line 2243 of file sourcesinkdata.cpp.

References observation(), ObservationsCount(), and Observation::PredictedValue().

◆ GetSampleNames()

vector< string > SourceSinkData::GetSampleNames ( const string & group_name ) const

Get all sample names within a specific group.

Returns the names of all samples (elemental profiles) within a specified source or target group.

Parameters

group_name Name of the group to query

Returns: Vector of sample names, or empty vector if group not found

Definition at line 395 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::GetSampleNames().

Referenced by GenericForm::GenericForm(), SelectSampleTableModel::data(), SelectSampleTableModel::GetProfileSet(), and SelectSampleTableModel::GetProfileSet().

◆ GetSampleSet()

Elemental_Profile_Set * SourceSinkData::GetSampleSet ( const string & name )

Get a sample set (source or target group) by name.

Retrieves a pointer to the Elemental_Profile_Set for a specific group. Returns nullptr if the group does not exist.

Parameters

name	Group name to retrieve

Returns: Pointer to the profile set, or nullptr if not found

Definition at line 386 of file sourcesinkdata.cpp.

References name.

Referenced by BuildSourceMeanMatrix(), BuildSourceMeanMatrix_Isotopes(), GetContributionVector(), GetContributionVectorSoftmax(), GetElementDistribution(), InitializeContributionsRandomly(), InitializeContributionsRandomlySoftmax(), LM_Batch(), LogLikelihoodSourceElementalDistributions(), ObservedDataforSelectedSample(), ObservedDataforSelectedSample_Isotope(), ObservedDataforSelectedSample_Isotope_delta(), MainWindow::on_tree_selectionChanged(), SetContribution(), SetContributionSoftmax(), SetParameterValue(), MainWindow::showdistributionsforelements(), MainWindow::ToQStandardItem(), MainWindow::ToQStandardItemModel(), and VerifySource().

◆ GetSourceContributions()

CVector SourceSinkData::GetSourceContributions ( )

private

Retrieves the contributions from all sources.

Calculates source contributions where the first (size-2) sources have their contributions stored as parameters, and the last source's contribution is calculated as the remainder to ensure all contributions sum to 1.

Returns: CVector of size (size()-1) containing contribution values for each source

Note: The last element is computed as (1 - sum of other contributions)

Definition at line 941 of file sourcesinkdata.cpp.

References parameter(), and Parameter::Value().

Referenced by LogPriorContributions().

◆ GetSourceOrder()

vector< string > SourceSinkData::GetSourceOrder ( ) const

private

Retrieves the ordering of source groups.

Returns: Vector of source group names in parameter order

Definition at line 5080 of file sourcesinkdata.cpp.

References samplesetsorder_.

Referenced by BootStrap(), GetContribution(), InitializeContributionsRandomly(), InitializeContributionsRandomlySoftmax(), MCMC(), ResidualJacobian(), ResidualJacobian_arma(), and ResidualJacobian_softmax().

◆ GetSourceProfiles()

vector< ResultItem > SourceSinkData::GetSourceProfiles ( )

Retrieves elemental profiles for all source groups.

Collects the complete elemental profile sets for each source group, packaged as ResultItems for display.

Returns: Vector of ResultItems, one per source group

Note: Target group is excluded

Definition at line 2448 of file sourcesinkdata.cpp.

References elemental_profile_set, ResultItem::SetName(), ResultItem::SetResult(), ResultItem::SetShowAsString(), ResultItem::SetShowGraph(), ResultItem::SetShowTable(), ResultItem::SetType(), and target_group_.

Referenced by Conductor::ExecuteOMSizeCorrect().

◆ GetTargetGroup()

string SourceSinkData::GetTargetGroup ( ) const

Retrieves the name of the target group.

Returns: Name of the target group, or empty string if not set

Definition at line 5022 of file sourcesinkdata.cpp.

References target_group_.

Referenced by Conductor::ExecuteMLR(), InitializeParametersAndObservations(), ObservedDataforSelectedSample(), ObservedDataforSelectedSample_Isotope(), ObservedDataforSelectedSample_Isotope_delta(), and SelectSamples::SetData().

◆ Gradient()

CVector SourceSinkData::Gradient	(	const CVector &	parameters,
		estimation_mode	est_mode
	)

private

Computes the normalized gradient of the log-likelihood function.

Calculates the gradient vector ∇log(L) using numerical differentiation (finite differences). The gradient points in the direction of steepest ascent for the log-likelihood.

Numerical Differentiation: ∂log(L)/∂θ_i ≈ [log(L(θ + ε·e_i)) - log(L(θ))] / ε

Parameters

parameters	Parameter vector at which to evaluate the gradient
est_mode	Estimation mode determining which likelihood components to include

Returns: Normalized gradient vector (unit length: ||∇log(L)|| = 1)

Note: The gradient is normalized by its L2 norm for numerical stability

Definition at line 1963 of file sourcesinkdata.cpp.

References epsilon_, LogLikelihood(), and SetParameterValue().

Referenced by GradientUpdate().

◆ GradientUpdate()

CVector SourceSinkData::GradientUpdate ( estimation_mode estmode = estimation_mode::elemental_profile_and_contribution )

Performs one gradient ascent step with adaptive step size.

Executes a single iteration of gradient ascent optimization on the log-likelihood function using an adaptive step size (distance_coeff_).

Algorithm:

Compute gradient direction at current parameters
Try step sizes: distance_coeff_ and 2×distance_coeff_
Accept step that improves likelihood
Adjust step size for next iteration

Parameters

estmode Estimation mode determining which likelihood components to include

Returns: Updated parameter vector after the gradient step

Note: Modifies distance_coeff_ member variable for next iteration

Definition at line 1993 of file sourcesinkdata.cpp.

References distance_coeff_, GetParameterValue(), Gradient(), LogLikelihood(), and SetParameterValue().

◆ GrandMean()

double SourceSinkData::GrandMean	(	const string &	element,
		bool	use_log
	)

private

Computes grand mean concentration for an element across all sources.

Calculates the overall mean concentration weighted by sample sizes across all source groups (excluding target).

Formula: μ_grand = Σ(n_i × μ_i) / Σ(n_i)

Parameters

element	Name of the element
use_log	If true, compute using log-space means

Returns: Weighted grand mean concentration

Note: Target group is excluded

Definition at line 3437 of file sourcesinkdata.cpp.

References ConcentrationSet::CalculateMean(), ConcentrationSet::CalculateMeanLog(), and target_group_.

◆ IncludeExcludeAllElements()

void SourceSinkData::IncludeExcludeAllElements ( bool include_in_analysis )

Sets inclusion flag for all elements.

Enables or disables all elements for analysis in batch.

Parameters

include_in_analysis If true, include all; if false, exclude all

Note: Affects all elements regardless of role

Definition at line 3426 of file sourcesinkdata.cpp.

References element_information_.

◆ IncludeExcludeElementsBasedOn()

void SourceSinkData::IncludeExcludeElementsBasedOn ( const vector< string > & elements )

Sets element inclusion based on a specified list.

First excludes all elements from analysis, then includes only those specified in the provided list. Useful for restricting analysis to a subset of elements (e.g., elements with sufficient data quality).

Parameters

elements Vector of element names to include in analysis

Note: All elements not in the list are excluded; Elements in the list that don't exist are silently ignored; Does not affect element roles, only the inclusion flag

Definition at line 3603 of file sourcesinkdata.cpp.

References element_information_.

Referenced by Conductor::ExecuteANOVA(), Conductor::ExecuteAutoSelect(), and Conductor::ExecuteSDFAM().

◆ InitializeContributionsRandomly()

bool SourceSinkData::InitializeContributionsRandomly ( )

private

Initialize source contributions randomly (linear constraint)

Randomly initializes contribution values for all sources such that they sum to 1.0 and all values are non-negative. Uses uniform random sampling with rejection until valid contributions are found.

Returns: Always returns true

Definition at line 644 of file sourcesinkdata.cpp.

References GetContributionVector(), GetSampleSet(), GetSourceOrder(), samplesetsorder_, Elemental_Profile_Set::SetContribution(), and SetContribution().

Referenced by SolveLevenberg_Marquardt().

◆ InitializeContributionsRandomlySoftmax()

bool SourceSinkData::InitializeContributionsRandomlySoftmax ( )

private

Initialize source contributions randomly (softmax transformation)

Randomly initializes contribution values using softmax transformation of normally distributed random variables. This ensures contributions automatically sum to 1.0 and are non-negative without rejection sampling.

Softmax transformation: contribution_i = exp(X_i) / sum(exp(X_j)) where X ~ N(0,1)

Returns: Always returns true

Definition at line 661 of file sourcesinkdata.cpp.

References GetSampleSet(), GetSourceOrder(), samplesetsorder_, Elemental_Profile_Set::SetContribution(), and SetContributionSoftmax().

Referenced by SolveLevenberg_Marquardt().

◆ InitializeParametersAndObservations()

bool SourceSinkData::InitializeParametersAndObservations	(	const string &	targetsamplename,
		estimation_mode	est_mode = `estimation_mode::elemental_profile_and_contribution`
	)

Initialize parameters and observations for MCMC optimization.

Sets up the complete parameter space and observation vector for Bayesian inference of source contributions and/or source profiles. The exact parameters created depend on the estimation mode:

only_contributions: Only source contribution parameters
elemental_profile_and_contribution: Contributions + mu/sigma for each source-element
source_elemental_profiles_based_on_source_data: Only source profile parameters

Parameters

targetsamplename	Name of the target sample to unmix
est_mode	Estimation mode controlling which parameters to optimize

Returns: true if successful, false if data not loaded

Note: Must be called before running MCMC or optimization

Definition at line 676 of file sourcesinkdata.cpp.

References Observation::AppendValues(), dirichlet, element_information::element, element_information_, GetElementalProfile(), GetTargetGroup(), Elemental_Profile::GetValue(), element_information::isotope, lognormal, normal, numberofconstituents_, numberofisotopes_, numberofsourcesamplesets_, observations_, only_contributions, Distribution::parameters, parameters_, PopulateConstituentOrders(), samplesetsorder_, selected_target_sample_, Observation::SetName(), Parameter::SetName(), Parameter::SetPriorDistribution(), Parameter::SetRange(), source_elemental_profiles_based_on_source_data, and target_group_.

Referenced by BootStrap(), BootStrap(), Conductor::ExecuteErrorAnalysis(), Conductor::ExecuteGA(), Conductor::ExecuteGA_FixedProfile(), Conductor::ExecuteGA_NoTargets(), Conductor::ExecuteLevenbergMarquardt(), LM_Batch(), MCMC(), MCMC_Batch(), and VerifySource().

◆ IsotopeOrder()

vector< string > SourceSinkData::IsotopeOrder ( )

private

Retrieves the ordering of isotopes.

Returns: Vector of isotope names in analysis order

Definition at line 5100 of file sourcesinkdata.cpp.

References isotope_order_.

Referenced by GetObservedElementalProfile_Isotope(), GetPredictedElementalProfile_Isotope(), and MCMC().

◆ IsotopesToBeUsedInCMB()

vector< string > SourceSinkData::IsotopesToBeUsedInCMB ( )

Identifies isotopes to be used in CMB analysis.

Scans the element information map and collects all isotopes marked with the 'isotope' role AND flagged for inclusion in analysis.

Returns: Vector of isotope names to be included in CMB analysis

Note: Updates isotope_order_ and numberofisotopes_; Only includes isotopes where include_in_analysis = true

Definition at line 2089 of file sourcesinkdata.cpp.

References element_information_, element_information::isotope, isotope_order_, and numberofisotopes_.

◆ LM_Batch()

CMBTimeSeriesSet SourceSinkData::LM_Batch	(	transformation	transform,
		bool	apply_om_size_correction,
		map< string, vector< string > > &	negative_elements
	)

Solves CMB model for all target samples using Levenberg-Marquardt.

Performs batch source apportionment on all samples in the target group, solving the CMB model via Levenberg-Marquardt optimization for each sample.

Parameters

transform	Transformation type (linear or softmax) for contributions
apply_om_size_correction	If true, apply OM and size corrections before solving
negative_elements	[out] Map of sample names to vectors of elements with negative values

Returns: CMBTimeSeriesSet containing contribution estimates for all valid samples

Note: Samples with negative values are skipped and recorded; Progress updates sent to rtw_ if available

Definition at line 3934 of file sourcesinkdata.cpp.

References CreateCorrectedDataset(), GetContributionVector(), GetElementInformation(), GetSampleSet(), InitializeParametersAndObservations(), NegativeValueCheck(), numberofsourcesamplesets_, rtw_, samplesetsorder_, ProgressWindow::SetLabel(), CMBTimeSeriesSet::SetLabel(), ProgressWindow::SetProgress(), SolveLevenberg_Marquardt(), and target_group_.

◆ LogLikelihood()

double SourceSinkData::LogLikelihood ( estimation_mode est_mode = estimation_mode::elemental_profile_and_contribution )

Calculates the total log-likelihood for Bayesian source apportionment.

Computes the posterior log-likelihood by summing multiple components based on the estimation mode. Components include source data likelihood, observation likelihood (elements and isotopes), and contribution priors.

Parameters

est_mode Estimation mode determining which likelihood terms to include

Returns: Total log-likelihood (higher values indicate better model fit)

Definition at line 1456 of file sourcesinkdata.cpp.

References LogLikelihoodModelvsMeasured(), LogLikelihoodModelvsMeasured_Isotope(), LogLikelihoodSourceElementalDistributions(), LogPriorContributions(), only_contributions, and source_elemental_profiles_based_on_source_data.

Referenced by GetObjectiveFunctionValue(), Gradient(), and GradientUpdate().

◆ LogLikelihoodModelvsMeasured()

double SourceSinkData::LogLikelihoodModelvsMeasured ( estimation_mode est_mode = estimation_mode::elemental_profile_and_contribution )

private

Calculates the log-likelihood of the model prediction versus measured data.

Compares the model's predicted elemental concentrations against the observed data for the selected target sample. Computed in log-space assuming log-normally distributed errors.

Formula: log(L) = -n*log(σ) - ||log(C_pred) - log(C_obs)||² / (2σ²)

Parameters

est_mode Estimation mode determining how predictions are made

Returns: Log-likelihood value (or -1e10 if any predicted concentration is non-positive)

Definition at line 1020 of file sourcesinkdata.cpp.

References based_on_fitted_distribution, direct, elemental_profile_and_contribution, error_stdev_, log, ObservedDataforSelectedSample(), PredictTarget(), and selected_target_sample_.

Referenced by LogLikelihood().

◆ LogLikelihoodModelvsMeasured_Isotope()

double SourceSinkData::LogLikelihoodModelvsMeasured_Isotope ( estimation_mode est_mode = estimation_mode::elemental_profile_and_contribution )

private

Calculates the log-likelihood of model versus measured isotopic data.

Compares the model's predicted isotopic delta values against the observed isotopic data for the selected target sample. Operates in linear space assuming normally distributed errors in delta values.

Formula: log(L) = -n*log(σ_iso) - ||δ_pred - δ_obs||² / (2σ_iso²)

Parameters

est_mode Estimation mode determining how predictions are made

Returns: Log-likelihood value

Definition at line 1051 of file sourcesinkdata.cpp.

References based_on_fitted_distribution, direct, elemental_profile_and_contribution, error_stdev_isotope_, log, ObservedDataforSelectedSample_Isotope_delta(), PredictTarget_Isotope_delta(), and selected_target_sample_.

Referenced by LogLikelihood().

◆ LogLikelihoodSourceElementalDistributions()

double SourceSinkData::LogLikelihoodSourceElementalDistributions ( )

private

Calculates the log-likelihood of source elemental distributions.

Computes the total log-likelihood by evaluating how well each source sample's observed elemental concentrations fit their respective estimated distributions.

Returns: Sum of log-probabilities across all elements, all source groups, and all samples

Definition at line 961 of file sourcesinkdata.cpp.

References element_order_, Distribution::EvalLog(), Elemental_Profile_Set::GetElementDistribution(), ConcentrationSet::GetEstimatedDistribution(), GetSampleSet(), numberofsourcesamplesets_, and samplesetsorder_.

Referenced by LogLikelihood().

◆ LogPriorContributions()

double SourceSinkData::LogPriorContributions ( )

private

Calculates the log prior probability for source contributions.

Evaluates whether the current source contributions are physically valid by checking if all contributions are non-negative. Acts as a constraint in Bayesian inference.

Returns: -1e10 if any contribution is negative, 0 if all valid

Definition at line 953 of file sourcesinkdata.cpp.

References GetSourceContributions().

Referenced by LogLikelihood().

◆ LumpAllProfileSets()

Elemental_Profile_Set SourceSinkData::LumpAllProfileSets ( )

private

Combines all source samples into a single profile set.

Pools samples from all source groups into one unified Elemental_Profile_Set.

Returns: Elemental_Profile_Set containing all source samples combined

Note: Target group is excluded

Definition at line 3470 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::AppendProfiles(), target_group_, and Elemental_Profile_Set::UpdateElementDistributions().

Referenced by ANOVA().

◆ MCMC()

Results SourceSinkData::MCMC	(	const string &	target_sample,
		map< string, string >	arguments,
		CMCMC< SourceSinkData > *	mcmc,
		ProgressWindow *	progress_window,
		const string &	working_folder
	)

Performs Markov Chain Monte Carlo analysis for Bayesian source apportionment.

Conducts full Bayesian inference using MCMC sampling to estimate posterior distributions of source contributions and model parameters. Generates comprehensive uncertainty quantification including credible intervals, posterior distributions, and predicted concentrations.

Parameters

target_sample	Name of target sample to apportion
arguments	Map of MCMC settings (number of samples, chains, burnin, etc.)
mcmc	Pointer to MCMC sampler object
progress_window	Pointer to progress window for updates
working_folder	Base directory for output files

Returns: Results object containing all MCMC outputs and credible intervals

Note: Burnin samples excluded from posterior statistics; Last contribution computed from sum constraint

Definition at line 4047 of file sourcesinkdata.cpp.

References Results::Append(), Results::AppendError(), CreateCorrectedDataset(), distribution, distribution_with_observed, element_information::element, ElementOrder(), GetElementInformation(), GetSourceOrder(), high, CMCMC< T >::initialize(), InitializeParametersAndObservations(), element_information::isotope, IsotopeOrder(), log, low, mcmc_samples, Range::mean, CMCMC< T >::Model, NegativeValueCheck(), normal, observation(), CMCMC< T >::predicted, rangeset, rangeset_with_observed, Range::Set(), Results::SetError(), Range::SetMean(), Range::SetMedian(), ResultItem::SetName(), Results::SetName(), CMBTimeSeriesSet::SetObservedValue(), ProgressWindow::SetProgress(), CMCMC< T >::SetProperty(), ResultItem::SetResult(), ResultItem::SetShowAsString(), ResultItem::SetShowTable(), ProgressWindow::SetTitle(), ResultItem::SetType(), Range::SetValue(), ResultItem::SetYAxisMode(), ProgressWindow::SetYAxisTitle(), ResultItem::SetYLimit(), SourceGroupNames(), CMCMC< T >::step(), and Observation::Value().

Referenced by Conductor::ExecuteCMBBayesian(), and MCMC_Batch().

◆ MCMC_Batch()

CMBMatrix SourceSinkData::MCMC_Batch	(	map< string, string >	arguments,
		CMCMC< SourceSinkData > *	mcmc,
		ProgressWindow *	progress_window,
		const string &	working_folder
	)

Performs batch MCMC analysis on all target samples.

Conducts full Bayesian MCMC analysis for every sample in the target group, generating comprehensive results for each sample and summarizing credible intervals in a matrix.

Output matrix structure:

Rows: Target samples
Columns (4 per source): [source]_low, [source]_high, [source]_median, [source]_mean

Parameters

arguments	Map of MCMC settings
mcmc	Pointer to MCMC sampler object
progress_window	Pointer to progress window for updates
working_folder	Base directory for output folders

Returns: CMBMatrix containing credible interval statistics for all samples

Note: Creates subdirectory for each target sample; All ResultItems saved as text files

Definition at line 4314 of file sourcesinkdata.cpp.

References ProgressWindow::ClearGraph(), Range::Get(), high, InitializeParametersAndObservations(), low, MCMC(), Range::Mean(), Range::Median(), numberofsourcesamplesets_, samplesetsorder_, CMBMatrix::SetColumnLabel(), ProgressWindow::SetLabel(), ProgressWindow::SetProgress2(), CMBMatrix::SetRowLabel(), and target_group_.

◆ MeanElementalContent() [1/2]

CMBVector SourceSinkData::MeanElementalContent ( )

private

Computes weighted mean elemental concentrations across all sources.

Calculates the overall mean concentration for each element, weighted by the number of samples in each source group.

Formula: μ_overall = Σ(n_i × μ_i) / Σ(n_i)

Returns: CMBVector of weighted mean concentrations, labeled with element names

Note: Target group excluded; Weighting ensures larger groups have proportional influence

Definition at line 4744 of file sourcesinkdata.cpp.

References element_order_, GetElementNames(), MeanElementalContent(), CMBVector::SetLabels(), and target_group_.

Referenced by BetweenGroupCovarianceMatrix(), MeanElementalContent(), and TotalScatterMatrix().

◆ MeanElementalContent() [2/2]

CMBVector SourceSinkData::MeanElementalContent ( const string & group_name )

private

Computes mean elemental concentrations for a specific group.

Calculates the mean concentration for each element within a specified source or target group.

Parameters

group_name Name of the group (source or target)

Returns: CMBVector of mean concentrations, labeled with element names

Note: Returns empty vector if group doesn't exist

Definition at line 4728 of file sourcesinkdata.cpp.

References GetElementNames(), and CMBVector::SetLabels().

◆ NegativeValueCheck()

vector< string > SourceSinkData::NegativeValueCheck ( )

Checks for zero or negative concentration values across all sources.

Scans all source groups for elements with zero or negative concentrations, which are problematic for log-normal distributions and certain statistical analyses. Returns detailed error messages identifying problematic elements and their source groups.

Returns: Vector of error messages describing zero/negative values found

Note: Returns empty vector if no issues found; Only checks source groups (target excluded)

Definition at line 3402 of file sourcesinkdata.cpp.

References element_order_, and PopulateConstituentOrders().

Referenced by Conductor::CheckNegativeElements(), LM_Batch(), MCMC(), and VerifySource().

◆ observation()

Observation * SourceSinkData::observation ( size_t i )

Retrieves pointer to an observation by index.

Parameters

i	Observation index (0-based)

Returns: Pointer to Observation, or nullptr if invalid index

Definition at line 5008 of file sourcesinkdata.cpp.

References observations_.

Referenced by GetPredictedValues(), MCMC(), PredictTarget(), and PredictTarget_Isotope_delta().

◆ ObservationsCount()

size_t SourceSinkData::ObservationsCount ( )

Returns the number of observations.

Returns: Number of observations (measured values)

Definition at line 4987 of file sourcesinkdata.cpp.

References observations_.

Referenced by GetPredictedValues().

◆ ObservedDataforSelectedSample()

CVector SourceSinkData::ObservedDataforSelectedSample ( const string & SelectedTargetSample = "" )

private

Retrieves the observed elemental data for a selected target sample.

Extracts the elemental concentration values for a specific target sample, returning them in the order specified by element_order_.

Parameters

SelectedTargetSample Name of the target sample (empty = use selected_target_sample_)

Returns: CVector containing elemental concentration values

Definition at line 981 of file sourcesinkdata.cpp.

References element_order_, Elemental_Profile_Set::GetProfile(), GetSampleSet(), GetTargetGroup(), Elemental_Profile::GetValue(), selected_target_sample_, and SelectedTargetSample().

Referenced by GetObservedElementalProfile(), LogLikelihoodModelvsMeasured(), ResidualVector(), and ResidualVector_arma().

◆ ObservedDataforSelectedSample_Isotope()

CVector SourceSinkData::ObservedDataforSelectedSample_Isotope ( const string & SelectedTargetSample = "" )

private

Retrieves the observed isotopic data for a selected target sample.

Extracts and converts isotopic ratio values for a specific target sample into absolute concentrations using the delta notation formula: concentration = (δ/1000 + 1) × standard_ratio × base_element_concentration

Parameters

SelectedTargetSample Name of the target sample (empty = use selected_target_sample_)

Returns: CVector containing converted isotopic concentration values

Definition at line 993 of file sourcesinkdata.cpp.

References element_information_, Elemental_Profile_Set::GetProfile(), GetSampleSet(), GetTargetGroup(), Elemental_Profile::GetValue(), isotope_order_, selected_target_sample_, and SelectedTargetSample().

◆ ObservedDataforSelectedSample_Isotope_delta()

CVector SourceSinkData::ObservedDataforSelectedSample_Isotope_delta ( const string & SelectedTargetSample = "" )

private

Retrieves the observed isotopic data in delta notation.

Extracts isotopic ratio values for a specific target sample in their original delta notation (‰) format, without conversion to absolute concentrations.

Parameters

SelectedTargetSample Name of the target sample (empty = use selected_target_sample_)

Returns: CVector containing isotopic delta values (‰)

Definition at line 1006 of file sourcesinkdata.cpp.

References element_information_, Elemental_Profile_Set::GetProfile(), GetSampleSet(), GetTargetGroup(), Elemental_Profile::GetValue(), isotope_order_, selected_target_sample_, and SelectedTargetSample().

Referenced by GetObservedElementalProfile_Isotope(), LogLikelihoodModelvsMeasured_Isotope(), ResidualVector(), and ResidualVector_arma().

◆ OMandSizeConstituents()

vector< string > SourceSinkData::OMandSizeConstituents ( )

Retrieves the names of OM and particle size constituents.

Returns a vector containing the names of the organic matter and particle size constituents set via PerformRegressionVsOMAndSize().

Returns: Vector with [0]=OM constituent name, [1]=size constituent name

Definition at line 5131 of file sourcesinkdata.cpp.

References omconstituent_, and sizeconsituent_.

Referenced by Conductor::ExecuteANOVA(), Conductor::ExecuteBracketingAnalysisBatch(), Conductor::ExecuteDistributionFitting(), Conductor::ExecuteEDP(), Conductor::ExecuteEDPM(), Conductor::ExecuteSDFA(), Conductor::ExecuteSDFAM(), Conductor::ExecuteSDFAOnevsRest(), and MainWindow::onOMSizeCorrection().

◆ OneStepLevenberg_Marquardt()

CVector SourceSinkData::OneStepLevenberg_Marquardt ( double lambda )

private

Performs one iteration of the Levenberg-Marquardt optimization algorithm.

Computes the parameter update step for source contributions using the Levenberg-Marquardt algorithm, which interpolates between Gauss-Newton and gradient descent methods.

The algorithm solves: (J^T J + λ diag(J^T J)) dx = -J^T r

Parameters

lambda Damping parameter controlling trade-off (typical: 0.001 to 1000)

Returns: CVector parameter update vector dx of size (n-1)

Note: Uses standard contribution parameterization (last source implicit)

Definition at line 1238 of file sourcesinkdata.cpp.

References ResidualJacobian(), and ResidualVector().

Referenced by SolveLevenberg_Marquardt().

◆ OneStepLevenberg_Marquardt_softmax()

CVector SourceSinkData::OneStepLevenberg_Marquardt_softmax ( double lambda )

private

Performs one iteration of Levenberg-Marquardt using softmax parameterization.

Computes the parameter update step for softmax-parameterized source contributions.

Parameters

lambda Damping parameter

Returns: CVector parameter update vector dx of size n

Note: Uses softmax parameterization where all n sources are independent

Definition at line 1268 of file sourcesinkdata.cpp.

References ResidualJacobian_softmax(), and ResidualVector().

Referenced by SolveLevenberg_Marquardt().

◆ operator=()

SourceSinkData & SourceSinkData::operator= ( const SourceSinkData & other )

Assignment operator.

Assigns all data from another SourceSinkData object to this one. Performs deep copy of all member variables.

Parameters

other The SourceSinkData object to assign from

Returns: Reference to this object

Definition at line 74 of file sourcesinkdata.cpp.

References constituent_order_, distance_coeff_, element_distributions_, element_information_, element_order_, isotope_order_, numberofconstituents_, numberofisotopes_, numberofsourcesamplesets_, observations_, omconstituent_, options_, outputpath_, parameter_estimation_mode_, parameters_, regression_p_value_threshold_, samplesetsorder_, selected_target_sample_, size_om_order_, sizeconsituent_, target_group_, and tools_used_.

◆ OptimalBoxCoxParameters()

CMBVector SourceSinkData::OptimalBoxCoxParameters ( )

private

Computes optimal Box-Cox transformation parameters for all elements.

Determines the optimal λ (lambda) parameter for each element's Box-Cox transformation by maximizing the log-likelihood of achieving normality.

Box-Cox transformation: y = (x^λ - 1) / λ (or ln(x) if λ = 0)

Special cases:

λ = 1: No transformation (linear)
λ = 0.5: Square root transformation
λ = 0: Log transformation
λ = -1: Reciprocal transformation

Returns: CMBVector of optimal λ parameters, labeled with element names

Note: Search range: λ ∈ [-5, 5]; Optimization iterations: 10

Definition at line 3183 of file sourcesinkdata.cpp.

References ExtractConcentrationSet(), GetElementNames(), and CMBVector::SetLabels().

Referenced by BoxCoxTransformed().

◆ OptionsToJsonObject()

QJsonObject SourceSinkData::OptionsToJsonObject ( ) const

Exports analysis options/settings to a JSON object.

Serializes all analysis configuration options (numerical parameters, thresholds, flags) as key-value pairs.

Returns: QJsonObject containing option names and values

Definition at line 2593 of file sourcesinkdata.cpp.

References options_.

Referenced by MainWindow::buildProjectJson().

◆ OutlierAnalysisForAll()

void SourceSinkData::OutlierAnalysisForAll	(	const double &	lower_threshold = `-3`,
		const double &	upper_threshold = `3`
	)

Performs outlier detection on all source groups.

Applies Box-Cox transformation and identifies outliers based on standardized residuals. Samples with standardized values outside the threshold range are flagged as outliers in their notes.

Detection method:

Box-Cox transform each element's distribution
Calculate z-scores: z = (x_transformed - μ) / σ
Flag outliers: z < lower_threshold OR z > upper_threshold

Parameters

lower_threshold	Lower bound for standardized values (e.g., -3.0)
upper_threshold	Upper bound for standardized values (e.g., +3.0)

Note: Only analyzes source groups (target group excluded); Outliers are flagged in sample notes, not removed; Uses Box-Cox transformation for normalization

Definition at line 2801 of file sourcesinkdata.cpp.

References target_group_.

Referenced by MainWindow::onIncludeExcludeSample().

◆ parameter() [1/2]

Parameter * SourceSinkData::parameter ( size_t i )

Retrieves pointer to a parameter by index.

Parameters

i	Parameter index (0-based)

Returns: Pointer to Parameter, or nullptr if invalid index

Definition at line 4992 of file sourcesinkdata.cpp.

References parameters_.

Referenced by GetParameterName(), and GetSourceContributions().

◆ parameter() [2/2]

const Parameter * SourceSinkData::parameter ( size_t i ) const

Retrieves const pointer to a parameter by index.

Parameters

i	Parameter index (0-based)

Returns: Const pointer to Parameter, or nullptr if invalid index

Definition at line 5000 of file sourcesinkdata.cpp.

References parameters_.

◆ ParameterEstimationMode()

estimation_mode SourceSinkData::ParameterEstimationMode ( )

Retrieves the current estimation mode.

Returns: Current estimation mode setting

Definition at line 5032 of file sourcesinkdata.cpp.

References parameter_estimation_mode_.

◆ Parameters()

vector< Parameter > & SourceSinkData::Parameters ( )

Retrieves reference to the parameters vector.

Returns: Reference to parameters vector

Definition at line 4977 of file sourcesinkdata.cpp.

References parameters_.

◆ ParametersCount()

size_t SourceSinkData::ParametersCount ( )

Returns the number of parameters.

Returns: Number of parameters in the optimization

Definition at line 4982 of file sourcesinkdata.cpp.

References parameters_.

◆ PerformRegressionVsOMAndSize()

bool SourceSinkData::PerformRegressionVsOMAndSize	(	const string &	om,
		const string &	particle_size,
		regression_form	form,
		const double &	p_value_threshold = `0.05`
	)

Performs multiple linear regression of elements vs OM and particle size.

Computes regression models for all sample groups to quantify how elemental concentrations vary with organic matter (OM) content and particle size. These models are used to correct elemental profiles for OM/size variations.

Regression forms:

Linear: C_corrected = C + β₁(OM_ref - OM) + β₂(Size_ref - Size)
Multiplicative: C_corrected = C × (OM_ref/OM)^β₁ × (Size_ref/Size)^β₂

Parameters

om	Name of the organic matter constituent (e.g., "OC", "OM")
particle_size	Name of the particle size constituent (e.g., "D50", "PM2.5")
form	Regression form (linear or multiplicative)
p_value_threshold	Significance threshold for including predictors (default: 0.05)

Returns: true if regressions were computed successfully

Note: Stores OM and size constituent names for later correction; Regressions are computed for all groups (sources and target); Predictors with p > threshold are excluded from the model

See also: GetMLRResults() to retrieve regression results

Definition at line 2781 of file sourcesinkdata.cpp.

References omconstituent_, regression_p_value_threshold_, and sizeconsituent_.

Referenced by Conductor::ExecuteMLR().

◆ PopulateConstituentOrders()

void SourceSinkData::PopulateConstituentOrders ( )

Populates all element ordering vectors used throughout CMB analysis.

Scans the element information map and organizes elements into separate ordering vectors based on their roles. These vectors define the order in which elements appear in matrices, parameter vectors, and observations.

Populated Vectors:

constituent_order_: All species in the dataset
element_order_: Chemical elements flagged for analysis
isotope_order_: Isotopes flagged for analysis
size_om_order_: Particle size and organic carbon parameters

Note: Must be called before InitializeParametersAndObservations()

Definition at line 2108 of file sourcesinkdata.cpp.

References constituent_order_, element_information::element, element_information_, element_order_, element_information::isotope, isotope_order_, numberofconstituents_, element_information::organic_carbon, element_information::particle_size, and size_om_order_.

Referenced by InitializeParametersAndObservations(), and NegativeValueCheck().

◆ PopulateElementDistributions()

void SourceSinkData::PopulateElementDistributions ( )

Populate element distributions from all groups.

Builds concentration distributions for each element by collecting all concentration values across all groups (sources and target). First updates distributions within each group, then aggregates them into the overall element_distributions_ map.

Note: Should be called after loading data or modifying profiles

Definition at line 521 of file sourcesinkdata.cpp.

References element_distributions_, and GetElementNames().

Referenced by BoxCoxTransformed(), CreateCorrectedAndFilteredDataset(), CreateCorrectedDataset(), ExtractChemicalElements(), ExtractSpecificElements(), MainWindow::LoadModel(), RandomlyEliminateSourceSamples(), MainWindow::ReadExcel(), and ReplaceSourceAsTarget().

◆ PopulateElementInformation()

void SourceSinkData::PopulateElementInformation ( const map< string, element_information > * ElementInfo = nullptr )

Populate element information metadata.

Initializes or updates the element_information_ map with metadata for all elements in the dataset. If ElementInfo is provided, copies metadata from it; otherwise initializes with default values.

Also updates the count of source sample sets (total groups minus target group if present).

Parameters

ElementInfo Optional source of element metadata (nullptr for defaults)

Definition at line 605 of file sourcesinkdata.cpp.

References element_information_, GetElementNames(), numberofsourcesamplesets_, and target_group_.

Referenced by CreateCorrectedAndFilteredDataset(), DiscriminantFunctionAnalysis(), DiscriminantFunctionAnalysis(), ExtractChemicalElements(), ExtractSpecificElements(), RandomlyEliminateSourceSamples(), MainWindow::ReadExcel(), and ReplaceSourceAsTarget().

◆ PredictTarget()

CVector SourceSinkData::PredictTarget ( parameter_mode param_mode = parameter_mode::direct )

private

Predicts target sample elemental concentrations based on source contributions.

Calculates the predicted elemental composition of the target sample as a linear combination of source profiles weighted by their contributions: C_predicted = Σ (source_mean_i × contribution_i)

Parameters

param_mode Parameter mode for obtaining source mean values

Returns: CVector of predicted elemental concentrations

Definition at line 1403 of file sourcesinkdata.cpp.

References BuildSourceMeanMatrix(), element_order_, GetContributionVector(), observation(), and Observation::SetPredictedValue().

Referenced by GetPredictedElementalProfile(), LogLikelihoodModelvsMeasured(), ResidualVector(), and ResidualVector_arma().

◆ PredictTarget_Isotope()

CVector SourceSinkData::PredictTarget_Isotope ( parameter_mode param_mode = parameter_mode::direct )

private

Predicts target sample isotopic compositions based on source contributions.

Calculates the predicted isotopic composition (as absolute concentrations) of the target sample as a linear combination of source isotopic profiles.

Parameters

param_mode Parameter mode for obtaining source mean values

Returns: CVector of predicted isotopic concentrations (absolute values)

Definition at line 1419 of file sourcesinkdata.cpp.

References BuildSourceMeanMatrix_Isotopes(), and GetContributionVector().

◆ PredictTarget_Isotope_delta()

CVector SourceSinkData::PredictTarget_Isotope_delta ( parameter_mode param_mode = parameter_mode::based_on_fitted_distribution )

private

Predicts target sample isotopic compositions in delta notation.

Calculates the predicted isotopic composition in delta notation (‰) by first computing absolute concentrations, then converting back to delta values.

Parameters

param_mode Parameter mode for obtaining source mean values

Returns: CVector of predicted isotopic delta values (‰)

Definition at line 1429 of file sourcesinkdata.cpp.

References BuildSourceMeanMatrix(), BuildSourceMeanMatrix_Isotopes(), element_information_, element_order_, GetContributionVector(), isotope_order_, numberofisotopes_, observation(), and Observation::SetPredictedValue().

Referenced by GetPredictedElementalProfile_Isotope(), LogLikelihoodModelvsMeasured_Isotope(), ResidualVector(), and ResidualVector_arma().

◆ RandomlyEliminateSourceSamples()

SourceSinkData SourceSinkData::RandomlyEliminateSourceSamples ( const double & percentage )

private

Creates dataset with randomly excluded source samples for validation.

Generates a new dataset with a specified percentage of source samples randomly removed. Useful for bootstrap validation and uncertainty analysis.

Parameters

percentage Percentage of samples to exclude (0-100)

Returns: New SourceSinkData object with reduced source samples

Note: Target group samples are preserved; Element distributions are recalculated

Definition at line 3623 of file sourcesinkdata.cpp.

References AssignAllDistributions(), element_information_, omconstituent_, PopulateElementDistributions(), PopulateElementInformation(), RandomlypickSamples(), sizeconsituent_, and target_group_.

Referenced by BootStrap(), and BootStrap().

◆ RandomlypickSamples()

vector< string > SourceSinkData::RandomlypickSamples ( const double & percentage ) const

private

Randomly selects a subset of source samples.

Performs Bernoulli sampling on all source samples, where each sample is independently included with the specified probability.

Parameters

percentage Probability of selecting each sample (0.0 to 1.0)

Returns: Vector of randomly selected sample names

Note: Expected number of samples: N × percentage; Each sample is selected independently

Definition at line 4025 of file sourcesinkdata.cpp.

References AllSourceSampleNames(), and GADistribution::GetRndUniF().

Referenced by RandomlyEliminateSourceSamples().

◆ ReadElementDatafromJsonObject()

bool SourceSinkData::ReadElementDatafromJsonObject ( const QJsonObject & jsonobject )

Deserializes elemental profile data from a JSON object.

Loads all sample groups and their elemental profiles from JSON format. Clears existing data before loading.

Parameters

jsonobject JSON object containing elemental profile sets

Returns: true if successfully loaded

Note: Clears all existing data before loading

Definition at line 2646 of file sourcesinkdata.cpp.

References elemental_profile_set, and Elemental_Profile_Set::ReadFromJsonObject().

Referenced by ReadFromFile().

◆ ReadElementInformationfromJsonObject()

bool SourceSinkData::ReadElementInformationfromJsonObject ( const QJsonObject & jsonobject )

Deserializes element information metadata from a JSON object.

Loads element roles, standard ratios, base elements, and inclusion flags from JSON format. Clears existing element information before loading.

Parameters

jsonobject JSON object containing element metadata

Returns: true if successfully loaded

Note: Clears element_information_ before loading

Definition at line 2626 of file sourcesinkdata.cpp.

References element_information::base_element, element_information_, element_information::include_in_analysis, element_information::Role, Role(), and element_information::standard_ratio.

Referenced by ReadFromFile().

◆ ReadFromFile()

bool SourceSinkData::ReadFromFile ( QFile * fil )

Loads complete dataset from a JSON file.

Deserializes all dataset components from a JSON file: elemental profiles, element information, tools used, options, and target group designation. Clears existing data before loading.

Parameters

fil	Pointer to an open QFile for reading

Returns: true if successfully loaded

Note: Clears all existing data before loading

Definition at line 2708 of file sourcesinkdata.cpp.

References Clear(), ReadElementDatafromJsonObject(), ReadElementInformationfromJsonObject(), ReadOptionsfromJsonObject(), ReadToolsUsedFromJsonObject(), and target_group_.

Referenced by MainWindow::LoadModel().

◆ ReadOptionsfromJsonObject()

bool SourceSinkData::ReadOptionsfromJsonObject ( const QJsonObject & jsonobject )

Deserializes analysis options from a JSON object.

Loads configuration options and settings from JSON format.

Parameters

jsonobject JSON object containing option key-value pairs

Returns: true if successfully loaded

Definition at line 2662 of file sourcesinkdata.cpp.

References options_.

Referenced by ReadFromFile().

◆ ReadToolsUsedFromJsonObject()

bool SourceSinkData::ReadToolsUsedFromJsonObject ( const QJsonArray & jsonarray )

Deserializes the list of analysis tools from a JSON array.

Parameters

jsonarray JSON array containing tool names

Returns: true if successfully loaded

Definition at line 2615 of file sourcesinkdata.cpp.

References AddtoToolsUsed().

Referenced by ReadFromFile().

◆ ReplaceSourceAsTarget()

SourceSinkData SourceSinkData::ReplaceSourceAsTarget ( const string & source_sample_name ) const

private

Creates a new dataset with a source sample designated as the target.

Generates a modified dataset where a specified source sample becomes the new target sample. Useful for leave-one-out validation.

Parameters

source_sample_name Name of the source sample to use as new target

Returns: New SourceSinkData object with specified sample as target

Note: Source sample is copied to target group (not moved)

Definition at line 3673 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::AppendProfile(), AssignAllDistributions(), element_information_, omconstituent_, PopulateElementDistributions(), PopulateElementInformation(), Sample(), sizeconsituent_, and target_group_.

Referenced by VerifySource().

◆ ResidualJacobian()

CMatrix SourceSinkData::ResidualJacobian ( )

private

Calculates the Jacobian matrix of residuals with respect to contributions.

Computes the numerical derivative of the residual vector with respect to the first (n-1) source contributions using finite differences.

Returns: CMatrix Jacobian matrix

Note: Uses adaptive epsilon based on distance from contribution = 0.5

Definition at line 1172 of file sourcesinkdata.cpp.

References element_order_, GetContributionVector(), GetSourceOrder(), isotope_order_, ResidualVector(), and SetContribution().

Referenced by OneStepLevenberg_Marquardt().

◆ ResidualJacobian_arma()

CMatrix_arma SourceSinkData::ResidualJacobian_arma ( )

private

Calculates the Jacobian matrix of residuals with respect to contributions (Armadillo)

Computes the numerical derivative of the residual vector with respect to the first (n-1) source contributions using finite differences.

The Jacobian matrix J has dimensions [(n-1) sources × (elements + isotopes)], where J[i,j] = ∂residual_j / ∂contribution_i

Returns: CMatrix_arma Jacobian matrix in Armadillo format

Note: Uses adaptive epsilon based on distance from contribution = 0.5

Definition at line 1140 of file sourcesinkdata.cpp.

References element_order_, GetContributionVector(), GetSourceOrder(), isotope_order_, ResidualVector_arma(), and SetContribution().

◆ ResidualJacobian_softmax()

CMatrix SourceSinkData::ResidualJacobian_softmax ( )

private

Calculates the Jacobian using softmax parameterization of contributions.

Computes the numerical derivative of residuals with respect to unconstrained softmax parameters. Includes all n sources since softmax transformation ensures contributions sum to 1 automatically.

Returns: CMatrix Jacobian matrix (n sources × m observations)

Note: Uses sign-dependent epsilon for better numerical behavior

Definition at line 1204 of file sourcesinkdata.cpp.

References element_order_, GetContributionVectorSoftmax(), GetSourceOrder(), isotope_order_, ResidualVector(), and SetContributionSoftmax().

Referenced by OneStepLevenberg_Marquardt_softmax().

◆ ResidualVector()

CVector SourceSinkData::ResidualVector ( )

private

Calculates the combined residual vector for elemental and isotopic predictions.

Computes residuals between predicted and observed values for both elemental concentrations (in log-space) and isotopic delta values (in linear space).

Combined residual vector structure: [log(C_pred/C_obs) for each element, δ_pred - δ_obs for each isotope]

Returns: CVector of combined residuals

Note: Returns vector with non-finite values if predictions are invalid

Definition at line 1077 of file sourcesinkdata.cpp.

References direct, ObservedDataforSelectedSample(), ObservedDataforSelectedSample_Isotope_delta(), PredictTarget(), PredictTarget_Isotope_delta(), and selected_target_sample_.

Referenced by OneStepLevenberg_Marquardt(), OneStepLevenberg_Marquardt_softmax(), ResidualJacobian(), ResidualJacobian_softmax(), and SolveLevenberg_Marquardt().

◆ ResidualVector_arma()

CVector_arma SourceSinkData::ResidualVector_arma ( )

private

Calculates the combined residual vector using Armadillo vector format.

Armadillo-based implementation of residual calculation for compatibility with linear algebra operations and optimization routines that use Armadillo types.

Returns: CVector_arma of combined residuals in Armadillo format

Definition at line 1115 of file sourcesinkdata.cpp.

References ObservedDataforSelectedSample(), ObservedDataforSelectedSample_Isotope_delta(), PredictTarget(), PredictTarget_Isotope_delta(), and selected_target_sample_.

Referenced by ResidualJacobian_arma().

◆ Role() [1/2]

QString SourceSinkData::Role ( const element_information::role & role ) const

private

Converts element role enum to string representation.

Converts the element_information::role enumeration to a human-readable string for serialization, display, or file output.

Mappings:

do_not_include → "DoNotInclude"
element → "Element"
isotope → "Isotope"
particle_size → "ParticleSize"
organic_carbon → "OM"

Parameters

role	Element role enumeration value

Returns: QString representation of the role

Note: Returns "DoNotInclude" for unrecognized values

Definition at line 2745 of file sourcesinkdata.cpp.

References element_information::do_not_include, element_information::element, element_information::isotope, element_information::organic_carbon, and element_information::particle_size.

Referenced by AssignAllDistributions(), ElementInformationToJsonObject(), and ReadElementInformationfromJsonObject().

◆ Role() [2/2]

element_information::role SourceSinkData::Role ( const QString & role_string ) const

private

Converts string representation to element role enum.

Parses a string and returns the corresponding element_information::role enumeration for deserialization or configuration loading.

Mappings:

"DoNotInclude" → do_not_include
"Element" → element
"Isotope" → isotope
"ParticleSize" → particle_size
"OM" → organic_carbon

Parameters

role_string QString representation of the role

Returns: element_information::role enumeration value

Note: Returns do_not_include for unrecognized strings

Definition at line 2763 of file sourcesinkdata.cpp.

References element_information::do_not_include, element_information::element, element_information::isotope, element_information::organic_carbon, and element_information::particle_size.

◆ Sample()

Elemental_Profile SourceSinkData::Sample ( const string & sample_name ) const

Retrieves an elemental profile by sample name.

Searches all groups (sources and target) for a sample with the specified name and returns its elemental profile. Returns empty profile if not found.

Parameters

sample_name Name of the sample to retrieve

Returns: Elemental_Profile for the sample, or empty profile if not found

Note: Searches all groups sequentially; Returns copy of profile, not reference; If multiple samples have the same name, returns first match

Definition at line 3659 of file sourcesinkdata.cpp.

Referenced by ReplaceSourceAsTarget().

◆ SamplesetsOrder()

vector< string > SourceSinkData::SamplesetsOrder ( )

private

Retrieves the ordering of sample sets.

Returns: Vector of sample set names

Definition at line 5085 of file sourcesinkdata.cpp.

References samplesetsorder_.

◆ SelectedTargetSample()

string SourceSinkData::SelectedTargetSample ( ) const

Retrieves the name of the currently selected target sample.

Returns: Name of the target sample being analyzed

Definition at line 5174 of file sourcesinkdata.cpp.

References selected_target_sample_.

Referenced by ObservedDataforSelectedSample(), ObservedDataforSelectedSample_Isotope(), and ObservedDataforSelectedSample_Isotope_delta().

◆ SetContribution() [1/2]

void SourceSinkData::SetContribution ( const CVector & contributions )

private

Sets all source contributions from a vector.

Updates contributions for multiple sources. If the vector has n-1 elements, the last contribution is computed automatically.

Parameters

contributions Vector of contribution values

Note: Validates that all contributions are non-negative

Definition at line 1646 of file sourcesinkdata.cpp.

References SetContribution().

◆ SetContribution() [2/2]

void SourceSinkData::SetContribution	(	size_t	source_index,
		double	contribution_value
	)

private

Sets a single source contribution value.

Updates the contribution fraction for the specified source and automatically recalculates the last source's contribution to maintain the sum constraint.

Parameters

source_index	Index of the source to update (0-based)
contribution_value	New contribution fraction (must satisfy 0 ≤ f ≤ 1)

Note: Last contribution is automatically updated: f_n = 1 - Σ(f_1...f_{n-1})

Definition at line 1622 of file sourcesinkdata.cpp.

References GetContributionVector(), GetSampleSet(), samplesetsorder_, and Elemental_Profile_Set::SetContribution().

Referenced by InitializeContributionsRandomly(), ResidualJacobian(), ResidualJacobian_arma(), SetContribution(), SetContributionSoftmax(), and SolveLevenberg_Marquardt().

◆ SetContributionSoftmax() [1/2]

void SourceSinkData::SetContributionSoftmax ( const CVector & softmax_params )

private

Sets all source contributions using softmax transformation.

Applies the softmax transformation to convert unconstrained parameters to valid contribution fractions: f_i = exp(x_i) / Σ exp(x_j)

Parameters

softmax_params Vector of unconstrained softmax parameters

Note: This is the correct way to update contributions in softmax mode

Definition at line 1661 of file sourcesinkdata.cpp.

References contribution, GetContributionVector(), SetContribution(), and SetContributionSoftmax().

◆ SetContributionSoftmax() [2/2]

void SourceSinkData::SetContributionSoftmax	(	size_t	source_index,
		double	softmax_value
	)

private

Sets a single softmax parameter value.

Updates the unconstrained softmax parameter for the specified source. Does NOT automatically update the actual contribution fractions.

Parameters

source_index	Index of the source to update (0-based)
softmax_value	New softmax parameter (unbounded: x ∈ ℝ)

Note: This only updates the softmax parameter, not the contribution

Definition at line 1639 of file sourcesinkdata.cpp.

References GetSampleSet(), samplesetsorder_, and Elemental_Profile_Set::SetContributionSoftmax().

Referenced by InitializeContributionsRandomlySoftmax(), ResidualJacobian_softmax(), SetContributionSoftmax(), and SolveLevenberg_Marquardt().

◆ SetOMandSizeConstituents() [1/2]

void SourceSinkData::SetOMandSizeConstituents	(	const string &	_omconstituent,
		const string &	_sizeconsituent
	)

Sets the names of OM and particle size constituents.

Parameters

_omconstituent	Name of organic matter constituent
_sizeconsituent	Name of particle size constituent

Definition at line 5112 of file sourcesinkdata.cpp.

References omconstituent_, and sizeconsituent_.

Referenced by Conductor::ExecuteMLR(), and MainWindow::LoadModel().

◆ SetOMandSizeConstituents() [2/2]

void SourceSinkData::SetOMandSizeConstituents ( const vector< string > & _omsizeconstituents )

Sets the names of OM and particle size constituents from a vector.

Parameters

_omsizeconstituents Vector with [0]=OM name, [1]=size name

Definition at line 5118 of file sourcesinkdata.cpp.

References omconstituent_, and sizeconsituent_.

◆ SetOutputPath()

bool SourceSinkData::SetOutputPath ( const string & output_path )

Set the output directory path.

Specifies the directory where analysis results and output files should be saved.

Parameters

output_path Path to output directory

Returns: Always returns true

Definition at line 5153 of file sourcesinkdata.cpp.

References outputpath_.

◆ SetParameterEstimationMode()

void SourceSinkData::SetParameterEstimationMode ( estimation_mode est_mode )

Sets the estimation mode for parameter optimization.

Specifies which parameters will be optimized during inference:

only_contributions: Estimate only source contributions (fixed source profiles)
elemental_profile_and_contribution: Estimate both contributions and source profiles
source_elemental_profiles_based_on_source_data: Estimate only source profiles

Parameters

est_mode Estimation mode to use

Definition at line 5027 of file sourcesinkdata.cpp.

References parameter_estimation_mode_.

Referenced by Conductor::ExecuteGA_FixedProfile(), and Conductor::ExecuteGA_NoTargets().

◆ SetParameterValue() [1/2]

bool SourceSinkData::SetParameterValue ( const CVector & values )

private

Sets multiple parameter values from a vector.

Updates all parameters using values from the provided vector. Each parameter update triggers synchronization with the corresponding model component.

Parameters

values Vector containing new parameter values

Returns: true if all parameters were successfully updated, false if any update failed

Note: Updates are applied sequentially; partial updates possible if one fails

Definition at line 1934 of file sourcesinkdata.cpp.

References SetParameterValue().

◆ SetParameterValue() [2/2]

bool SourceSinkData::SetParameterValue	(	size_t	index,
		double	value
	)

Sets a parameter value and updates corresponding model components.

Updates a parameter in the parameter vector and synchronizes the change with the appropriate model component (contributions, distribution parameters, or error terms). The parameter vector layout depends on the estimation mode.

Parameter Vector Layout (full mode: elemental_profile_and_contribution): [0 to n-2]: Contribution parameters (n-1 sources) [n-1 to end of element μ]: Element μ parameters [element μ to end of isotope μ]: Isotope μ parameters [isotope μ to end of element σ]: Element σ parameters [element σ to end of isotope σ]: Isotope σ parameters [end-1]: Error std dev for elements [end]: Error std dev for isotopes

Parameters

index	Parameter index in the parameters_ vector
value	New parameter value

Returns: true if parameter was successfully updated, false if index is invalid

Note: Automatically updates last contribution to maintain sum constraint; Validates that standard deviations are non-negative

Definition at line 1773 of file sourcesinkdata.cpp.

References element_order_, elemental_profile_and_contribution, error_stdev_, error_stdev_isotope_, GetContributionVector(), GetElementDistribution(), GetSampleSet(), isotope_order_, numberofconstituents_, numberofisotopes_, numberofsourcesamplesets_, only_contributions, parameter_estimation_mode_, parameters_, samplesetsorder_, Elemental_Profile_Set::SetContribution(), ConcentrationSet::SetEstimatedMu(), ConcentrationSet::SetEstimatedSigma(), and source_elemental_profiles_based_on_source_data.

Referenced by Gradient(), GradientUpdate(), and SetParameterValue().

◆ SetProgressWindow()

void SourceSinkData::SetProgressWindow ( ProgressWindow * _rtw )

Sets the progress window for displaying optimization progress.

Assigns a progress window that will receive updates during long-running operations like MCMC sampling, bootstrap analysis, or batch processing.

Parameters

_rtw	Pointer to ProgressWindow object (nullptr to disable progress display)

Definition at line 5037 of file sourcesinkdata.cpp.

References rtw_.

Referenced by Conductor::ExecuteDFA(), Conductor::ExecuteDFAM(), Conductor::ExecuteDFAOnevsRest(), Conductor::ExecuteErrorAnalysis(), Conductor::ExecuteLevenbergMarquardt(), Conductor::ExecuteLevenbergMarquardtBatch(), Conductor::ExecuteSDFA(), Conductor::ExecuteSDFAM(), Conductor::ExecuteSDFAOnevsRest(), Conductor::ExecuteSourceVerify(), and VerifySource().

◆ SetSelectedTargetSample()

bool SourceSinkData::SetSelectedTargetSample ( const string & sample_name )

Sets the currently selected target sample for analysis.

Specifies which sample from the target group will be used as the receptor in CMB calculations. Validates that the sample exists before setting.

Parameters

sample_name Name of the target sample to analyze

Returns: true if sample exists and was selected, false if sample not found

Note: Must be called before InitializeParametersAndObservations()

Definition at line 5161 of file sourcesinkdata.cpp.

References selected_target_sample_, and target_group_.

Referenced by Conductor::ExecuteOMSizeCorrect().

◆ SetTargetGroup()

bool SourceSinkData::SetTargetGroup ( const string & targroup )

Sets the target/sink group designation.

Specifies which group in the dataset represents the target samples (receptor sites to be unmixed via source apportionment).

Parameters

targroup Name of the target group

Returns: Always returns true

Definition at line 5016 of file sourcesinkdata.cpp.

References target_group_.

◆ SizeOMOrder()

vector< string > SourceSinkData::SizeOMOrder ( )

private

Retrieves the ordering of size and OM constituents.

Returns: Vector of size/OM constituent names

Definition at line 5105 of file sourcesinkdata.cpp.

References size_om_order_.

◆ SolveLevenberg_Marquardt()

bool SourceSinkData::SolveLevenberg_Marquardt ( transformation trans = transformation::linear )

Solves for optimal source contributions using the Levenberg-Marquardt algorithm.

Iteratively optimizes source contributions to minimize the residual between predicted and observed elemental/isotopic compositions. The algorithm adaptively adjusts the damping parameter based on convergence behavior.

Convergence criteria (any of):

Residual norm < tolerance (1e-10)
Parameter change norm < tolerance (1e-10)
Maximum iterations reached (1000)

Parameters

trans

Parameterization method:

transformation::linear: Direct contribution values with constraint Σc_i = 1
transformation::softmax: Unconstrained parameters transformed via softmax

Returns: bool Currently always returns false (legacy)

Note: Updates internal state with optimized contributions; Progress displayed in rtw_ if set

Definition at line 1298 of file sourcesinkdata.cpp.

References ProgressWindow::AppendPoint(), GetContributionVector(), GetContributionVectorSoftmax(), InitializeContributionsRandomly(), InitializeContributionsRandomlySoftmax(), linear, OneStepLevenberg_Marquardt(), OneStepLevenberg_Marquardt_softmax(), ResidualVector(), rtw_, SetContribution(), SetContributionSoftmax(), ProgressWindow::SetProgress(), ProgressWindow::SetXRange(), and softmax.

Referenced by BootStrap(), BootStrap(), Conductor::ExecuteLevenbergMarquardt(), LM_Batch(), and VerifySource().

◆ SourceGroupNames()

vector< string > SourceSinkData::SourceGroupNames ( ) const

Retrieves the names of all source groups (excluding target)

Returns a list of source group names, which are the pollution/emission sources being apportioned in the CMB analysis.

Returns: Vector of source group names

Note: Target group is filtered out

Definition at line 2162 of file sourcesinkdata.cpp.

References target_group_.

Referenced by GenericForm::GenericForm(), and MCMC().

◆ StepwiseDiscriminantFunctionAnalysis() [1/3]

vector< CMBVector > SourceSinkData::StepwiseDiscriminantFunctionAnalysis ( )

Performs stepwise discriminant analysis across all source groups.

Iteratively selects elements that best discriminate among all source groups using forward selection.

Returns: Vector of 3 CMBVectors containing selection order with: [0] = p-values at each step [1] = Wilks' Lambda at each step [2] = F-test p-values at each step

Note: Uses multi-group DFA statistics

Definition at line 4843 of file sourcesinkdata.cpp.

References DFA_P_Value(), DFA_Projected(), ExtractSpecificElements(), CMBVectorSet::FTest_p_value(), GetElementNames(), rtw_, ProgressWindow::SetProgress(), and WilksLambda().

◆ StepwiseDiscriminantFunctionAnalysis() [2/3]

vector< CMBVector > SourceSinkData::StepwiseDiscriminantFunctionAnalysis ( const string & source1 )

Performs stepwise discriminant analysis for one source vs all others.

Iteratively selects elements that best discriminate one source from all other sources combined using forward selection.

Parameters

source1 Name of source group to test against others

Returns: Vector of 3 CMBVectors containing selection order

Note: Other sources pooled into "Others" group

Definition at line 4908 of file sourcesinkdata.cpp.

References CMBVector::append(), DiscriminantFunctionAnalysis(), ExtractSpecificElements(), DFA_result::F_test_P_value, GetElementNames(), DFA_result::p_values, rtw_, ProgressWindow::SetProgress(), and DFA_result::wilkslambda.

◆ StepwiseDiscriminantFunctionAnalysis() [3/3]

vector< CMBVector > SourceSinkData::StepwiseDiscriminantFunctionAnalysis	(	const string &	source1,
		const string &	source2
	)

Performs stepwise discriminant analysis between two specific sources.

Iteratively selects elements that best discriminate between two source groups using a forward selection procedure. At each step, adds the element that most improves group separation (lowest p-value).

Parameters

source1	Name of first source group
source2	Name of second source group

Returns: Vector of 3 CMBVectors containing selection order with: [0] = p-values at each step [1] = Wilks' Lambda at each step [2] = F-test p-values at each step

Note: Elements ordered by discriminating power (best first); Progress updates sent to rtw_ if available

Definition at line 4773 of file sourcesinkdata.cpp.

References CMBVector::append(), DiscriminantFunctionAnalysis(), DFA_result::eigen_vectors, ExtractSpecificElements(), DFA_result::F_test_P_value, GetElementNames(), DFA_result::p_values, rtw_, ProgressWindow::SetProgress(), and DFA_result::wilkslambda.

Referenced by Conductor::ExecuteSDFA(), Conductor::ExecuteSDFAM(), and Conductor::ExecuteSDFAOnevsRest().

◆ t_TestPValue()

Elemental_Profile SourceSinkData::t_TestPValue	(	const string &	source1,
		const string &	source2,
		bool	use_log
	)

Computes t-test p-values for element-wise differences between two sources.

Performs independent two-sample t-tests for each element to assess whether concentrations differ significantly between two source groups. Lower p-values indicate stronger evidence that the sources differ for that element.

T-statistic: t = (μ₁ - μ₂) / √(σ₁²/n₁ + σ₂²/n₂)

Parameters

source1	Name of first source group
source2	Name of second source group
use_log	If true, perform test on log-transformed concentrations; if false, use linear-space concentrations

Returns: Elemental_Profile containing two-tailed p-values for each element

Note: P-values are two-tailed (tests for any difference, not directional); Lower p-values indicate better source differentiation; Log-space testing appropriate for log-normally distributed data

See also: DifferentiationPower_P_value() for all pairwise comparisons

Definition at line 3204 of file sourcesinkdata.cpp.

References Elemental_Profile::AppendElement(), ConcentrationSet::CalculateMean(), ConcentrationSet::CalculateMeanLog(), ConcentrationSet::CalculateStdDev(), ConcentrationSet::CalculateStdDevLog(), and GetElementNames().

Referenced by DifferentiationPower_P_value(), and Conductor::ExecuteEDP().

◆ TheRest()

Elemental_Profile_Set SourceSinkData::TheRest ( const string & excluded_source )

private

Collects all source samples except those from a specified source group.

Creates a combined profile set containing samples from all source groups except the specified one and the target group. Useful for "one vs rest" comparisons.

Parameters

excluded_source Name of the source group to exclude

Returns: Elemental_Profile_Set containing all samples from other sources

Note: Target group is always excluded

Definition at line 2968 of file sourcesinkdata.cpp.

References Elemental_Profile_Set::AppendProfile(), and target_group_.

Referenced by DiscriminantFunctionAnalysis().

◆ ToolsUsed()

bool SourceSinkData::ToolsUsed ( const string & tool_name )

Checks if a specific analysis tool has been used.

Queries the tools_used list to determine if a particular analysis method or tool has been applied to this dataset.

Parameters

tool_name Name of the tool to check (e.g., "MCMC", "Bootstrap")

Returns: true if tool has been used, false otherwise

Definition at line 4407 of file sourcesinkdata.cpp.

References tools_used_.

Referenced by AddtoToolsUsed(), and ToolBoxItem::data().

◆ ToolsUsedToJsonObject()

QJsonArray SourceSinkData::ToolsUsedToJsonObject ( ) const

Exports the list of analysis tools used to a JSON array.

Serializes the names of all statistical/analytical tools that have been applied to this dataset (e.g., "MCMC", "Levenberg-Marquardt", "BoxCox").

Returns: QJsonArray containing tool names as strings

Definition at line 2580 of file sourcesinkdata.cpp.

References tools_used_.

Referenced by MainWindow::buildProjectJson().

◆ TotalNumberofSourceSamples()

int SourceSinkData::TotalNumberofSourceSamples ( ) const

private

Counts the total number of source samples across all source groups.

Sums the number of samples in all source groups, excluding the target group.

Returns: Total number of source samples

Note: Target group samples are excluded

Definition at line 2930 of file sourcesinkdata.cpp.

References target_group_.

Referenced by DFA_P_Value().

◆ TotalScatterMatrix()

CMatrix SourceSinkData::TotalScatterMatrix ( )

private

Computes total scatter matrix.

Calculates the overall covariance matrix treating all source samples as a single population. Represents total variability.

Formula: S_T = Σ Σ[(x_ij - μ_overall)(x_ij - μ_overall)ᵀ] / N

Relationship: S_T = S_W + S_B

Returns: Total scatter matrix (num_elements × num_elements)

Note: Target group excluded

Definition at line 4512 of file sourcesinkdata.cpp.

References GetElementNames(), MeanElementalContent(), CMBVector::size(), and target_group_.

◆ VerifySource()

CMBTimeSeriesSet SourceSinkData::VerifySource	(	const string &	source_group,
		bool	use_softmax,
		bool	apply_om_size_correction
	)

Performs leave-one-out validation on a source group.

Tests source apportionment accuracy by iteratively treating each sample from a source group as an unknown target, solving the CMB model, and comparing estimated contributions against the known true source.

Ideal result: Sample shows ~100% contribution from its true source

Parameters

source_group	Name of source group to validate
use_softmax	If true, use softmax transformation for contributions
apply_om_size_correction	If true, apply OM and size corrections

Returns: CMBTimeSeriesSet containing contribution estimates for each sample

Note: Samples with negative values after correction are skipped; Progress updates sent to rtw_ if available

Definition at line 3861 of file sourcesinkdata.cpp.

References CreateCorrectedDataset(), GetContributionVector(), GetElementInformation(), GetSampleSet(), InitializeParametersAndObservations(), linear, NegativeValueCheck(), numberofsourcesamplesets_, ReplaceSourceAsTarget(), rtw_, samplesetsorder_, CMBTimeSeriesSet::SetLabel(), ProgressWindow::SetProgress(), SetProgressWindow(), softmax, SolveLevenberg_Marquardt(), and target_group_.

Referenced by Conductor::ExecuteSourceVerify().

◆ WilksLambda()

double SourceSinkData::WilksLambda ( )

private

Computes Wilks' Lambda statistic for multivariate group separation.

Calculates Wilks' Lambda, a multivariate test statistic measuring the ratio of within-group to total variance.

Formula: Λ = |S_W| / |S_T| = |S_W| / |S_W + S_B|

Interpretation:

Λ = 1: No group separation
Λ → 0: Perfect separation

Returns: Wilks' Lambda statistic (0 to 1)

Note: Uses absolute values of determinants for numerical stability

Definition at line 4553 of file sourcesinkdata.cpp.

References BetweenGroupCovarianceMatrix(), and WithinGroupCovarianceMatrix().

Referenced by DFA_P_Value(), DiscriminantFunctionAnalysis(), DiscriminantFunctionAnalysis(), and StepwiseDiscriminantFunctionAnalysis().

◆ WithinGroupCovarianceMatrix()

CMatrix SourceSinkData::WithinGroupCovarianceMatrix ( )

private

Computes pooled within-group covariance matrix.

Calculates the weighted average of covariance matrices within each source group. Represents the "error" or within-group scatter in discriminant analysis.

Formula: S_W = Σ[(n_i - 1) × Σ_i] / Σ(n_i - 1)

Returns: Within-group covariance matrix (num_elements × num_elements)

Note: Target group excluded; Weighted by degrees of freedom (n_i - 1)

Definition at line 4450 of file sourcesinkdata.cpp.

References GetElementNames(), and target_group_.

Referenced by DFA_eigvector(), and WilksLambda().

◆ WriteDataToFile()

bool SourceSinkData::WriteDataToFile ( QFile * file )

Writes elemental profile data to a text file.

Outputs all sample groups and their elemental profiles in tab-delimited text format for documentation or debugging purposes.

Parameters

file	Pointer to an open QFile for writing

Returns: true if write operation succeeded

Note: File must be opened in write mode before calling

Definition at line 2686 of file sourcesinkdata.cpp.

Referenced by WriteToFile().

◆ WriteToFile()

bool SourceSinkData::WriteToFile ( QFile * file )

Writes dataset to a text file.

Public interface for writing elemental profile data to a file.

Parameters

file	Pointer to an open QFile for writing

Returns: true if write operation succeeded

Definition at line 2703 of file sourcesinkdata.cpp.

References WriteDataToFile().

Member Data Documentation

◆ constituent_order_

vector<string> SourceSinkData::constituent_order_

private

Definition at line 1798 of file sourcesinkdata.h.

Referenced by Clear(), ConstituentOrder(), CreateCorrectedDataset(), ElementsToBeUsedInCMB(), operator=(), and PopulateConstituentOrders().

◆ distance_coeff_

double SourceSinkData::distance_coeff_

private

Definition at line 1808 of file sourcesinkdata.h.

Referenced by GradientUpdate(), and operator=().

◆ element_distributions_

map<string, ConcentrationSet> SourceSinkData::element_distributions_

private

Definition at line 1778 of file sourcesinkdata.h.

Referenced by AssignAllDistributions(), Clear(), CreateCorrectedDataset(), GetElementDistribution(), GetFittedDistribution(), operator=(), and PopulateElementDistributions().

◆ element_information_

map<string, element_information> SourceSinkData::element_information_

private

Definition at line 1777 of file sourcesinkdata.h.

Referenced by AssignAllDistributions(), BuildSourceMeanMatrix_Isotopes(), Clear(), CountElements(), CreateCorrectedAndFilteredDataset(), CreateCorrectedDataset(), DiscriminantFunctionAnalysis(), DiscriminantFunctionAnalysis(), ElementInformationToJsonObject(), ElementsToBeUsedInCMB(), ExtractChemicalElements(), ExtractSpecificElements(), FirstOMConstituent(), FirstSizeConstituent(), GetElementInformation(), GetElementInformation(), IncludeExcludeAllElements(), IncludeExcludeElementsBasedOn(), InitializeParametersAndObservations(), IsotopesToBeUsedInCMB(), ObservedDataforSelectedSample_Isotope(), ObservedDataforSelectedSample_Isotope_delta(), operator=(), PopulateConstituentOrders(), PopulateElementInformation(), PredictTarget_Isotope_delta(), RandomlyEliminateSourceSamples(), ReadElementInformationfromJsonObject(), and ReplaceSourceAsTarget().

◆ element_order_

vector<string> SourceSinkData::element_order_

private

◆ epsilon_

double SourceSinkData::epsilon_

private

Definition at line 1811 of file sourcesinkdata.h.

Referenced by Gradient().

◆ error_stdev_

double SourceSinkData::error_stdev_

private

Definition at line 1809 of file sourcesinkdata.h.

Referenced by LogLikelihoodModelvsMeasured(), and SetParameterValue().

◆ error_stdev_isotope_

double SourceSinkData::error_stdev_isotope_

private

Definition at line 1810 of file sourcesinkdata.h.

Referenced by LogLikelihoodModelvsMeasured_Isotope(), and SetParameterValue().

◆ isotope_order_

vector<string> SourceSinkData::isotope_order_

private

Definition at line 1800 of file sourcesinkdata.h.

Referenced by BuildSourceMeanMatrix_Isotopes(), Clear(), CreateCorrectedDataset(), IsotopeOrder(), IsotopesToBeUsedInCMB(), ObservedDataforSelectedSample_Isotope(), ObservedDataforSelectedSample_Isotope_delta(), operator=(), PopulateConstituentOrders(), PredictTarget_Isotope_delta(), ResidualJacobian(), ResidualJacobian_arma(), ResidualJacobian_softmax(), and SetParameterValue().

◆ numberofconstituents_

int SourceSinkData::numberofconstituents_

private

Definition at line 1781 of file sourcesinkdata.h.

Referenced by Clear(), CreateCorrectedDataset(), ElementsToBeUsedInCMB(), GetElementDistributionSigmaParameter(), InitializeParametersAndObservations(), operator=(), PopulateConstituentOrders(), and SetParameterValue().

◆ numberofisotopes_

int SourceSinkData::numberofisotopes_

private

Definition at line 1782 of file sourcesinkdata.h.

Referenced by Clear(), CreateCorrectedDataset(), InitializeParametersAndObservations(), IsotopesToBeUsedInCMB(), operator=(), PredictTarget_Isotope_delta(), and SetParameterValue().

◆ numberofsourcesamplesets_

int SourceSinkData::numberofsourcesamplesets_

private

◆ observations_

vector<Observation> SourceSinkData::observations_

private

Definition at line 1786 of file sourcesinkdata.h.

Referenced by Clear(), CreateCorrectedDataset(), InitializeParametersAndObservations(), observation(), ObservationsCount(), and operator=().

◆ omconstituent_

string SourceSinkData::omconstituent_

private

Definition at line 1805 of file sourcesinkdata.h.

Referenced by CreateCorrectedAndFilteredDataset(), CreateCorrectedDataset(), ExtractChemicalElements(), OMandSizeConstituents(), operator=(), PerformRegressionVsOMAndSize(), RandomlyEliminateSourceSamples(), ReplaceSourceAsTarget(), SetOMandSizeConstituents(), and SetOMandSizeConstituents().

◆ options_

QMap<QString, double> SourceSinkData::options_

private

Definition at line 1816 of file sourcesinkdata.h.

Referenced by SourceSinkData(), CreateCorrectedAndFilteredDataset(), CreateCorrectedDataset(), GetOptions(), operator=(), OptionsToJsonObject(), and ReadOptionsfromJsonObject().

◆ outputpath_

string SourceSinkData::outputpath_

private

Definition at line 1790 of file sourcesinkdata.h.

Referenced by Clear(), CreateCorrectedDataset(), GetOutputPath(), operator=(), and SetOutputPath().

◆ parameter_estimation_mode_

estimation_mode SourceSinkData::parameter_estimation_mode_

private

Definition at line 1804 of file sourcesinkdata.h.

Referenced by CreateCorrectedDataset(), GetObjectiveFunctionValue(), operator=(), ParameterEstimationMode(), SetParameterEstimationMode(), and SetParameterValue().

◆ parameters_

vector<Parameter> SourceSinkData::parameters_

private

Definition at line 1787 of file sourcesinkdata.h.

Referenced by Clear(), GetElementDistributionMuParameter(), GetElementDistributionSigmaParameter(), GetParameterValue(), GetParameterValue(), InitializeParametersAndObservations(), operator=(), parameter(), parameter(), Parameters(), ParametersCount(), and SetParameterValue().

◆ regression_p_value_threshold_

double SourceSinkData::regression_p_value_threshold_

private

Definition at line 1807 of file sourcesinkdata.h.

Referenced by CreateCorrectedAndFilteredDataset(), CreateCorrectedDataset(), ExtractChemicalElements(), operator=(), and PerformRegressionVsOMAndSize().

◆ rtw_

ProgressWindow* SourceSinkData::rtw_ = nullptr

private

Definition at line 1814 of file sourcesinkdata.h.

Referenced by SourceSinkData(), SourceSinkData(), BootStrap(), BootStrap(), LM_Batch(), SetProgressWindow(), SolveLevenberg_Marquardt(), StepwiseDiscriminantFunctionAnalysis(), StepwiseDiscriminantFunctionAnalysis(), StepwiseDiscriminantFunctionAnalysis(), and VerifySource().

◆ samplesetsorder_

vector<string> SourceSinkData::samplesetsorder_

private

◆ selected_target_sample_

string SourceSinkData::selected_target_sample_

private

Definition at line 1794 of file sourcesinkdata.h.

Referenced by Clear(), CreateCorrectedAndFilteredDataset(), CreateCorrectedDataset(), GetObservedElementalProfile(), GetObservedElementalProfile_Isotope(), InitializeParametersAndObservations(), LogLikelihoodModelvsMeasured(), LogLikelihoodModelvsMeasured_Isotope(), ObservedDataforSelectedSample(), ObservedDataforSelectedSample_Isotope(), ObservedDataforSelectedSample_Isotope_delta(), operator=(), ResidualVector(), ResidualVector_arma(), SelectedTargetSample(), and SetSelectedTargetSample().

◆ size_om_order_

vector<string> SourceSinkData::size_om_order_

private

Definition at line 1801 of file sourcesinkdata.h.

Referenced by Clear(), CreateCorrectedDataset(), operator=(), PopulateConstituentOrders(), and SizeOMOrder().

◆ sizeconsituent_

string SourceSinkData::sizeconsituent_

private

Definition at line 1806 of file sourcesinkdata.h.

Referenced by CreateCorrectedAndFilteredDataset(), CreateCorrectedDataset(), ExtractChemicalElements(), OMandSizeConstituents(), operator=(), PerformRegressionVsOMAndSize(), RandomlyEliminateSourceSamples(), ReplaceSourceAsTarget(), SetOMandSizeConstituents(), and SetOMandSizeConstituents().

◆ target_group_

string SourceSinkData::target_group_

private

Definition at line 1793 of file sourcesinkdata.h.

Referenced by AllSourceSampleNames(), ANOVA(), BetweenGroupCovarianceMatrix(), BoxCoxTransformed(), BracketTest(), BracketTest(), Clear(), CreateCorrectedAndFilteredDataset(), CreateCorrectedDataset(), DFA_P_Value(), DFA_Projected(), DifferentiationPower(), DifferentiationPower_P_value(), DifferentiationPower_Percentage(), DiscriminantFunctionAnalysis(), ExtractChemicalElements(), ExtractConcentrationSet(), ExtractSpecificElements(), GetCalculatedElementMeans(), GetCalculatedElementMu(), GetCalculatedElementSigma(), GetEstimatedElementMean(), GetEstimatedElementMu(), GetEstimatedElementSigma(), GetSourceProfiles(), GetTargetGroup(), GrandMean(), InitializeParametersAndObservations(), LM_Batch(), LumpAllProfileSets(), MCMC_Batch(), MeanElementalContent(), operator=(), OutlierAnalysisForAll(), PopulateElementInformation(), RandomlyEliminateSourceSamples(), ReadFromFile(), ReplaceSourceAsTarget(), SetSelectedTargetSample(), SetTargetGroup(), SourceGroupNames(), TheRest(), TotalNumberofSourceSamples(), TotalScatterMatrix(), VerifySource(), and WithinGroupCovarianceMatrix().

◆ tools_used_

list<string> SourceSinkData::tools_used_

private

Definition at line 1815 of file sourcesinkdata.h.

Referenced by AddtoToolsUsed(), Clear(), operator=(), ToolsUsed(), and ToolsUsedToJsonObject().

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

Public Member Functions

Private Member Functions

Private Attributes

Detailed Description

Constructor & Destructor Documentation

◆ SourceSinkData() [1/2]

◆ SourceSinkData() [2/2]

Member Function Documentation

◆ AddtoToolsUsed()

◆ AllSourceSampleNames()

◆ ANOVA() [1/2]

◆ ANOVA() [2/2]

◆ AppendSampleSet()

◆ AssignAllDistributions()

◆ BetweenGroupCovarianceMatrix()

◆ BootStrap() [1/2]

◆ BootStrap() [2/2]

◆ BoxCoxTransformed()

◆ BracketTest() [1/2]

◆ BracketTest() [2/2]

◆ BuildSourceMeanMatrix()

◆ BuildSourceMeanMatrix_Isotopes()

◆ Clear()

◆ ConstituentOrder()

◆ CountElements()

◆ CreateCorrectedAndFilteredDataset()

◆ CreateCorrectedDataset()

◆ DFA_eigvector()

◆ DFA_P_Value()

◆ DFA_Projected() [1/3]

◆ DFA_Projected() [2/3]

◆ DFA_Projected() [3/3]

◆ DFA_weight_vector()

◆ DFATransformed()

◆ DifferentiationPower() [1/2]

◆ DifferentiationPower() [2/2]

◆ DifferentiationPower_P_value()

◆ DifferentiationPower_Percentage() [1/2]

◆ DifferentiationPower_Percentage() [2/2]

◆ DiscriminantFunctionAnalysis() [1/3]

◆ DiscriminantFunctionAnalysis() [2/3]

◆ DiscriminantFunctionAnalysis() [3/3]

◆ ElementDataToJsonObject()

◆ ElementInformationToJsonObject()

◆ ElementOrder()

◆ ElementsToBeUsedInCMB()

◆ ExtractChemicalElements()

◆ ExtractConcentrationData()

◆ ExtractConcentrationSet()

◆ ExtractElementConcentrations()

◆ ExtractElementDataByGroup()

◆ ExtractSamplesAsProfileSet()

◆ ExtractSpecificElements()

◆ FirstOMConstituent()

◆ FirstSizeConstituent()

◆ GetCalculatedElementMeans()

◆ GetCalculatedElementMu()

◆ GetCalculatedElementSigma()

◆ GetContribution()

◆ GetContributionVector()

◆ GetContributionVectorSoftmax()

◆ GetElementalProfile()

◆ GetElementDistribution() [1/2]

◆ GetElementDistribution() [2/2]

◆ GetElementDistributionMuParameter()

◆ GetElementDistributionMuValue()

◆ GetElementDistributionSigmaParameter()

◆ GetElementDistributionSigmaValue()

◆ GetElementInformation() [1/2]

◆ GetElementInformation() [2/2]

◆ GetElementNames()

◆ GetEstimatedElementMean()

◆ GetEstimatedElementMu()

◆ GetEstimatedElementSigma()

◆ GetFittedDistribution()

◆ GetGroupNames()

◆ GetMLRResults()

◆ GetObjectiveFunctionValue()

◆ GetObservedElementalProfile()

◆ GetObservedElementalProfile_Isotope()