Represents a multidimensional optimizer that uses a quasiNewton algorithm (DFP or BFGS).
SystemObject Extreme.Mathematics.AlgorithmsManagedIterativeAlgorithmVectorDouble,
Double,
OptimizationSolutionReport Extreme.Mathematics.OptimizationMultidimensionalOptimizer Extreme.Mathematics.OptimizationDirectionalOptimizer Extreme.Mathematics.OptimizationQuasiNewtonOptimizer
Namespace:
Extreme.Mathematics.Optimization
Assembly:
Extreme.Numerics (in Extreme.Numerics.dll) Version: 8.1.1
public sealed class QuasiNewtonOptimizer : DirectionalOptimizer
Public NotInheritable Class QuasiNewtonOptimizer
Inherits DirectionalOptimizer
public ref class QuasiNewtonOptimizer sealed : public DirectionalOptimizer
[<SealedAttribute>]
type QuasiNewtonOptimizer =
class
inherit DirectionalOptimizer
end
The QuasiNewtonOptimizer type exposes the following members.
Top
 Name  Description 

 ConvergenceTests 
Gets the collection of convergence tests for the algorithm.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 CurrentDirection 
Gets or sets the current search direction.
(Inherited from DirectionalOptimizer.) 
 Dimensions 
Gets or sets the number of dimensions of the optimization problem.
(Inherited from MultidimensionalOptimizer.) 
 EstimatedError 
Gets a value indicating the size of the absolute
error of the result.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 EvaluationsNeeded 
Gets the number of evaluations needed to execute the algorithm.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 Extremum 
Gets or sets the current best approximation to the extremum.
(Inherited from MultidimensionalOptimizer.) 
 ExtremumType 
Gets or sets the type of extremum.
(Inherited from MultidimensionalOptimizer.) 
 FastGradientFunction 
Gets or sets the function that computes the gradient of the objective funciton.
(Inherited from MultidimensionalOptimizer.) 
 GradientEvaluationsNeeded 
Gets the number of evaluations of the gradient of the objective function.
(Inherited from MultidimensionalOptimizer.) 
 GradientFunction 
Gets or sets the function that computes the gradient of the objective funciton.
(Inherited from MultidimensionalOptimizer.) 
 GradientTest  (Inherited from MultidimensionalOptimizer.) 
 GradientVector 
Gets or sets the current value of the gradient.
(Inherited from MultidimensionalOptimizer.) 
 HasSharedDegreeOfParallelism 
Indicates whether the degree of parallelism is a property that is shared
across instances.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 InitialGuess 
Gets or sets the initial value for the iteration.
(Inherited from MultidimensionalOptimizer.) 
 IterationsNeeded 
Gets the number of iterations needed by the
algorithm to reach the desired accuracy.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 LineSearch 
Gets or sets the algorithm used to perform a line search.
(Inherited from DirectionalOptimizer.) 
 MaxDegreeOfParallelism 
Gets or sets the maximum degree of parallelism enabled by this instance.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 MaxEvaluations 
Gets or sets the maximum number of evaluations during the calculation.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 MaxIterations  Gets or sets the maximum number of iterations
to use when approximating the roots of the target
function.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 Method 
Gets or sets the method used to update the conjugate gradient direction.

 MinIterations 
Gets or sets the minimum iterations that have to be performed.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 ObjectiveFunction 
Gets or sets the objective function.
(Inherited from MultidimensionalOptimizer.) 
 ObjectiveFunctionWithGradient 
Gets or sets a function that evaluates the value and gradient
of the objective function.
(Inherited from MultidimensionalOptimizer.) 
 ParallelOptions 
Gets or sets the configuration for the parallel behavior of the algorithm.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 RestartIterations 
Gets or sets a value that indicates when the Hessian approximation should be restarted
from a diagonal matrix.

 Result 
Gets the result of an algorithm after it has executed.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 SolutionReport 
Gets the result of an algorithm after it has executed.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 SolutionTest  (Inherited from MultidimensionalOptimizer.) 
 Status  (Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 SymbolicObjectiveFunction 
Gets or sets the objective function.
(Inherited from MultidimensionalOptimizer.) 
 ThrowExceptionOnFailure 
Gets or sets a value indicating whether to throw an
exception when the algorithm fails to converge.
(Inherited from ManagedIterativeAlgorithmT, TError, TReport.) 
 ValueAtExtremum 
Gets or sets the current value of the objective function.
(Inherited from MultidimensionalOptimizer.) 
 ValueTest  (Inherited from MultidimensionalOptimizer.) 
Top
Top
Use the QuasiNewtonOptimizer class to find an extremum of a multivariate function
using a quasiNewton method. Two variations of this method are available: the BFGS method of
Broyden, Fletcher, Goldfarb and Shanno, and the DFP method of Davison, Fletcher and Powell.
The default is the BFGS method.
A quasiNewton method is the preferred method for smaller problems when the gradient
of the objective function is available. For large problems, the
Conjugate Gradient method is usually more efficient.
The objective function must be supplied as a multivariate function
delegate to the ObjectiveFunction property. The gradient of the objective function
can be supplied either as a multivariate function returning a vector delegate (by setting the
GradientFunction property), or a multivariate function returning a vector in its second argument
delegate (by setting the FastGradientFunction property). The latter has the advantage
that the same Vector instance is reused to hold the result.
Sometimes, the gradient function is not available, or is very expensive to calculate.
In such instances, a numerical approximation may work better.
Before the algorithm is run, you must set the InitialGuess property to
a vector that contains an initial estimate for the extremum. The ExtremumType
property specifies whether a minimum or a maximum of the objective function is desired.
The FindExtremum method performs the actual
search for an extremum, and returns a Vector containing the best approximation.
The Extremum property also returns the best
approximation to the extremum. The ValueAtExtremum property
returns the value of the objective function at the extremum.
The Status
property is a AlgorithmStatus value that indicates the outcome of the algorithm.
A value of Normal shows normal termination.
A value of Divergent usually indicates that the objective
function is not bounded.
The algorithm has three convergence tests. By default, the algorithm terminates
when either of these is satisfied. You can deactivate either test by setting its Enabled
property to . If both tests are deactivated, then the algorithm always terminates when
the maximum number of iterations or function evaluations is reached.
The first test is based on the uncertainty in the location
of the approximate extremum. The SolutionTest property returns a
VectorConvergenceTestT object that allows you to specify the desired
Tolerance and
specific ConvergenceCriterion.
See the VectorConvergenceTestT class for details on how to further customize
this test.
The second test is based on the change in value of the objective function at the approximate extremum.
The test is successful when the change of the value of the objective function is within the tolerance.
Care should be taken with this test. When the tolerance is too large, the algorithm will terminate prematurely.
The ValueTest property returns a SimpleConvergenceTestT object
that can be used to customize the test.
The third test is based on the value of the gradient at the approximate extremum.
The GradientTest property returns a VectorConvergenceTestT object
that can be used to customize the test. By default, the error is set to the component
with the largest absolute value.
Reference