NonlinearObjectiveFunctionNonAllocatingBase Class

Base implementation for a nonlinear objective model that supports allocation-free evaluation for Levenberg-Marquardt-style algorithms.

Inheritance Hierarchy

SystemObject
  Altaxo.Calc.Optimization.ObjectiveFunctionsNonlinearObjectiveFunctionNonAllocatingBase
    Altaxo.Calc.Optimization.ObjectiveFunctionsNonlinearObjectiveFunctionNonAllocating
    Altaxo.Calc.Regression.NonlinearNonlinearModelOfFitEnsemble

Namespace: Altaxo.Calc.Optimization.ObjectiveFunctions
Assembly: AltaxoCore (in AltaxoCore.dll) Version: 4.8.3572.0 (4.8.3572.0)

Syntax

Copy

public abstract class NonlinearObjectiveFunctionNonAllocatingBase : IObjectiveModel, 
	IObjectiveModelEvaluation, IObjectiveModelNonAllocating

The NonlinearObjectiveFunctionNonAllocatingBase type exposes the following members.

Constructors

	Name	Description
	NonlinearObjectiveFunctionNonAllocatingBase	Initializes a new instance of the NonlinearObjectiveFunctionNonAllocatingBase class.

Top

Properties

	Name	Description
	DegreeOfFreedom	Get the degree of freedom.
	FunctionEvaluations	Get the number of calls to function.
	Gradient	Get the Gradient vector. G = J'(y - f(x; p))
	Hessian	Get the approximated Hessian matrix. H = J'J
	IsFixedByUser	Gets whether parameters are fixed or free (by the user).
	IsFixedByUserOrBoundary	Gets or sets an array of the same length as the parameter array. If an element in this vector is , that parameter is either fixed by the user or fixed because the corresponding parameter has reached a boundary. This array is updated only at the end of the minimization process.
	IsGradientSupported	Gets a value indicating whether the gradient can be provided by the model.
	IsHessianSupported	Gets a value indicating whether the Hessian can be provided by the model.
	JacobianEvaluations	Get the number of calls to jacobian.
	ModelValues	Get the y-values of the fitted model that correspond to the independent values.
	NegativeGradient	Gets the negative gradient vector. -G = -J'(y - f(x; p)).
	NumberOfObservations	Gets the number of observations.
	NumberOfParameters	Gets the number of unknown parameters.
	ObservedY	Get the y-values of the observations.
	Point	Get the values of the parameters.
	SigmaSquare	Gets Chi²/(N-F+1).
	Value	Get the residual sum of squares.
	Weights	Gets or sets the values of the weights for the observations.

Top

Methods

	Name	Description
	CreateNew	Creates a new instance of the objective model with identical configuration but independent state.
	Equals	Determines whether the specified object is equal to the current object. (Inherited from Object)
	EvaluateAt(IReadOnlyListDouble)	Evaluates the model at the given parameter vector and invalidates cached dependent values.
	EvaluateAt(VectorDouble)	Evaluates the model at the given parameter vector, updating dependent values.
	EvaluateFunction	Evaluates the objective function value and updates cached values.
	EvaluateJacobian	Evaluates the Jacobian and updates cached Jacobian-derived values (gradient and Hessian).
	Finalize	Allows an object to try to free resources and perform other cleanup operations before it is reclaimed by garbage collection. (Inherited from Object)
	Fork	Creates a forked copy of the model with independent mutable state.
	GetHashCode	Serves as the default hash function. (Inherited from Object)
	GetType	Gets the Type of the current instance. (Inherited from Object)
	MemberwiseClone	Creates a shallow copy of the current Object. (Inherited from Object)
	NumericalJacobian	Numerically approximates the Jacobian at the specified parameter vector.
	SetParameters(IReadOnlyListDouble, IReadOnlyListBoolean)	Sets model parameters and optional fixed flags for individual parameters.
	SetParameters(VectorDouble, ListBoolean)	Sets the model parameters and optional fixed flags for individual parameters.
	ToObjectiveFunction	Converts this model to an objective function suitable for minimizers.
	ToString	Returns a string that represents the current object. (Inherited from Object)

Top

Fields

	Name	Description
	_accuracyOrder	The desired accuracy order to evaluate the jacobian by numerical approximaiton.
	_coefficients	Coefficients for the model.
	_f1	Temporary function-evaluation vectors used for numerical differentiation.
	_f2	Temporary function-evaluation vectors used for numerical differentiation.
	_f3	Temporary function-evaluation vectors used for numerical differentiation.
	_f4	Temporary function-evaluation vectors used for numerical differentiation.
	_f5	Temporary function-evaluation vectors used for numerical differentiation.
	_f6	Temporary function-evaluation vectors used for numerical differentiation.
	_functionValue	The residual sum of squares, residuals * residuals.
	_hasFunctionValue	Indicates if the function value has been computed.
	_hasJacobianValue	Indicates if the jacobian has been computed.
	_hessianValue	The Hessian matrix.
	_jacobianValue	The Jacobian matrix.
	_jacobianValueTransposed	The Jacobian matrix, transposed.
	_negativeGradientValue	The Gradient vector.
	_residuals	The weighted error values.
	L	Gets or sets the Cholesky factorization of the weights = LL'.

Top

Remarks

This class provides cached values for the function value, Jacobian-derived quantities, and auxiliary vectors/matrices to minimize allocations during repeated evaluations.

Reference

Altaxo.Calc.Optimization.ObjectiveFunctions Namespace