Optimization.Optimizer

⊟ Remarks

This class forms the basis of all optimizers. In itself it does not provide an optimization, but it provides all the infrastructure to make it easy to implement different optimization algorithms on top of this class.

There are a few key concepts to understand when implementing an optimizer based on this class. The optimizer consists of the following objects:
A list of Optimization.Solution which make up the population
A Optimization.Optimizer.Settings object providing optimizer settings (such as maximum number of iterations and population size)
A Optimization.Storage.Storage object providing a way to store the optimization results
Templates for the parameters and associated boundaries used in the optimization process

When implementing a new optimizer, you might want to use a subclassed settings class (Optimization.Optimizer.Settings) (if you want to add some settings specific to your optimizer and/or a custom subclass of a Optimization.Solution. The optimizer class implements a automatic discovery mechanism for finding subclasses of these two classes and will use them if it finds them. Therefore, you only need to subclass from for example Optimization.Solution and this class will be used to create solutions. This behavior can be overridden by overriding the Optimizer.CreateSettings and Optimizer.CreateSolution methods.

The usual optimization process usually consists of the following phases:
Create a new optimizer object
If needed, create a new storage object and assign it
Call Optimizer.Initialize to initialize the initial population and the storage
Evaluate all the solutions in the population (this should update the fitness of each solution)
Call Optimizer.Next to generate a new population
Repeat running solutions in the population as long as Optimizer.Next returns true. There is a helper class which can drive this process for you, including communicating with the master and dispatching the solutions (see Optimization.Application).

Subclasses of an optimizer usually only need to override certain methods. It's useful to know what exactly happens when calling Optimizer.Next so that you can override the relevant methods.
Update the current best solution (Optimizer.UpdateBest)
Save the current iteration in the storage (Optimization.Storage.Storage.SaveIteration)
Increment the current iteration (Optimizer.IncreaseIteration)
Check if the optimization is finished (Optimizer.Finished)
Create a new population (Optimizer.Update)

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Members

See Also: Inherited members from object.

⊟ Public Constructors

Optimizer ()
Default constructor.

⊟ Public Properties

[read-only]
Best Solution . Get the best best solution so far.

[read-only]
Boundaries List<Boundary> . Get list of boundaries.

[read-only]
Configuration Optimizer.Settings . Get optimizer configuration.

CurrentIteration uint . The current iteration.

[read-only]
Description string . Optimizer description.

[read-only]
Extensions List<Extension> . Extensions.

[read-only]
Fitness Fitness . The fitness object template.

[read-only]
Name string . The optimizer name.

[read-only]
Parameters List<Parameter> . Get parameters.

[read-only]
Population List<Solution> . Current population of solutions.

[read-only]
State State . Optimizer state.

Storage Optimization.Storage.Storage . Storage object.

⊟ Public Methods

Add (Solution)
Add a new solution to the optimizer.

AddExtension (Extension)
Add extension.

Boundary (string) : Boundary
Retrieve boundary by name.

CreateSolution (uint) : Solution
Create a new solution.

FromStorage (Optimization.Storage.Storage, Optimization.Storage.Records.Optimizer)
Restore optimizer from storage.

FromStorage (Optimization.Storage.Storage, Optimization.Storage.Records.Optimizer, Optimization.Storage.Records.Solution, Solution)
Restore solution from storage.

FromXml (System.Xml.XmlNode)
Load optimizer from XML.

static
GetDescription (Type) : string
Get optimizer description.

GetEnumerator () : IEnumerator<Solution>
Get solution enumerator.

static
GetName (Type) : string
Get the optimizer name.

HasBoundary (string) : bool
Check boundary existence.

HasParameter (string) : bool
Check parameter existence.

Initialize ()
Initialize optimizer.

InitializePopulation ()
Initialize solution population.

Log (string, string)
Generic message logging.

Log (string, string, params object[])
Generic message logging.

Next () : bool
Create new optimization iteration.

Parameter (string) : Parameter
Retrieve parameter by name.

Remove (Solution)
Remove solution from the population.

Update ()
Update the current population.

Update (Solution)
Update the solution.

⊟ Protected Methods

AddBoundary (Boundary)
Add boundary.

AddParameter (Parameter)
Add parameter.

CreateSettings () : Optimizer.Settings
Create settings object.

CreateState () : State
Create state.

Finished () : bool
Check whether the optimization is finished.

IncrementIteration ()
Increment the current iteration.

SaveIteration ()
Save iteration to storage.

Setup ()
Setup the optimizer.

UpdateBest ()
Update the best solution so far.

⊟ Explicitly Implemented Interface Members

IEnumerable.GetEnumerator Get solution enumerator.

⊟ Member Details

⊟ Optimizer Constructor

Default constructor.

public Optimizer ()

⊟ Remarks

Create new optimizer object.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Add Method

Add a new solution to the optimizer.

public virtual void Add (Solution solution)

⊟ Parameters

solution

The solution to add.

⊟ Remarks

Add a new solution to the population of solutions in the optimizer. This can be useful if you want to generate an initial population externally, or just want to add some solutions at some point from your application.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ AddBoundary Method

Add boundary.

protected void AddBoundary (Boundary boundary)

⊟ Parameters

boundary

Boundary.

⊟ Remarks

Add a boundary to the optimizer.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ AddExtension Method

Add extension.

public virtual void AddExtension (Extension ext)

⊟ Parameters

ext

Extension.

⊟ Remarks

Add an extension to the optimizer.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ AddParameter Method

Add parameter.

protected void AddParameter (Parameter parameter)

⊟ Parameters

parameter

A Parameter.

⊟ Remarks

Add parameter to the optimizer.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Best Property

Get the best best solution so far.

public Solution Best { get; }

⊟ Value

The best solution so far.

⊟ Remarks

The optimizer keeps track of the best solution so far and it can be accessed by this property.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Boundaries Property

Get list of boundaries.

public List<Boundary> Boundaries { get; }

⊟ Value

The list of boundaries.

⊟ Remarks

Get the list of boundaries registered on the optimizer.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Boundary Method

Retrieve boundary by name.

public Boundary Boundary (string name)

⊟ Parameters

name

Boundary name.

⊟ Returns

The boundary.

⊟ Remarks

Get a boundary by name.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Configuration Property

Get optimizer configuration.

public Optimizer.Settings Configuration { get; }

⊟ Value

Optimizer configuration.

⊟ Remarks

The optimizer configuration contains settings which can easily be set by a job. Default settings for optimizer include the maximum number of iterations and the population size.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ CreateSettings Method

Create settings object.

protected virtual Optimizer.Settings CreateSettings ()

⊟ Returns

A settings object.

⊟ Remarks

If you want to add some settings, you can subclass the Optimization.Optimizer.Settings class and override this function in your optimizer to create an instance of this subclass. Note that the default implementation of this method will try to locate any subclasses of Optimization.Optimizer.Settings and use it if found. See Optimization.Optimizer.Settings for more information on implementing your own settings class.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ CreateSolution Method

Create a new solution.

public virtual Solution CreateSolution (uint idx)

⊟ Parameters

idx

Solution id.

⊟ Returns

A new solution.

⊟ Remarks

This function creates a new solution object. You can override this to create a solution of a custom subclass. Note that the default implementation will already look for subclasses of Optimization.Solution, so you generally don't need to override this function. See Optimization.Solution for more information on implementing a custom solution class.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ CreateState Method

Create state.

protected virtual State CreateState ()

⊟ Returns

New state.

⊟ Remarks

Create a new state. Override this method to create a custom state subclass.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ CurrentIteration Property

The current iteration.

public uint CurrentIteration { set; get; }

⊟ Value

The current iteration.

⊟ Remarks

Represents the current optimization iteration.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Description Property

Optimizer description.

public string Description { get; }

⊟ Value

Optimizer description.

⊟ Remarks

Get the optimizer description.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Extensions Property

Extensions.

public List<Extension> Extensions { get; }

⊟ Value

List of extensions.

⊟ Remarks

Get list of extensions associated with the optimizer.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Finished Method

Check whether the optimization is finished.

protected virtual bool Finished ()

⊟ Returns

true if the optimization is finished, false otherwise.

⊟ Remarks

Override this method to implement your own way of determining when the optimization is finished. By default, there is a maximum iteration set by the MaxIterations setting.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Fitness Property

The fitness object template.

public Fitness Fitness { get; }

⊟ Value

The fitness object.

⊟ Remarks

The fitness object template associated with the optimizer. The fitness object is a template used by the solutions do calculate the fitness value of the solution, using a mathematical expression.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ FromStorage Method

Restore optimizer from storage.

public virtual void FromStorage (Optimization.Storage.Storage storage, Optimization.Storage.Records.Optimizer optimizer)

⊟ Parameters

storage

The storage.

optimizer

The storage optimizer record.

⊟ Remarks

Restore the optimizer state from storage.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ FromStorage Method

Restore solution from storage.

public virtual void FromStorage (Optimization.Storage.Storage storage, Optimization.Storage.Records.Optimizer optimizer, Optimization.Storage.Records.Solution solution, Solution sol)

⊟ Parameters

storage

The storage.

optimizer

The optimizer storage record.

solution

The solution storage record.

sol

The solution.

⊟ Remarks

Restore a solution from storage.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ FromXml Method

Load optimizer from XML.

public virtual void FromXml (System.Xml.XmlNode node)

⊟ Parameters

node

XML node.

⊟ Remarks

Load optimizer settings, paramaters and boundaries from XML. Subclasses can override this and extract custom data from the XML. The passed node is the optimizer root node.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ GetDescription Method

Get optimizer description.

public static string GetDescription (Type type)

⊟ Parameters

type

Optimizer type.

⊟ Returns

The optimizer description, or null if there is no description available.

⊟ Remarks

Gets the optimizer description from the Optimization.Attributes.OptimizerAttribute.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ GetEnumerator Method

Get solution enumerator.

public IEnumerator<Solution> GetEnumerator ()

⊟ Returns

Solution enumerator.

⊟ Remarks

Implements an enumerator which enumerates over the solutions in the optimizer population.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ GetName Method

Get the optimizer name.

public static string GetName (Type type)

⊟ Parameters

type

Optimizer type.

⊟ Returns

The optimizer name.

⊟ Remarks

Get the optimizer name (see Optimization.Attributes.OptimizerAttribute). If the name is not set through an attribute, the class name is used.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ HasBoundary Method

Check boundary existence.

public bool HasBoundary (string name)

⊟ Parameters

name

Boundary name.

⊟ Returns

Whether the boundary exists.

⊟ Remarks

Check whether a given boundary exists.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ HasParameter Method

Check parameter existence.

public bool HasParameter (string name)

⊟ Parameters

name

Parameter name.

⊟ Returns

Whether the parameter exists.

⊟ Remarks

Check whether a given parameter exists.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ IncrementIteration Method

Increment the current iteration.

protected virtual void IncrementIteration ()

⊟ Remarks

This is usually called from Optimizer.Next to increase the current iteration count.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Initialize Method

Initialize optimizer.

public virtual void Initialize ()

⊟ Remarks

Initialize the optimizer. This creates the initial population by calling Optimizer.InitializePopulation. It furthermore tells the storage that the optimization has begun. The application should call this before starting the optimization process.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ InitializePopulation Method

Initialize solution population.

public virtual void InitializePopulation ()

⊟ Remarks

Initialize the solution population. By default, this will create a number of solutions according to the PopulationSize setting.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Log Method

Generic message logging.

public virtual void Log (string type, string str)

⊟ Parameters

type

Log type.

str

Log message.

⊟ Remarks

The default implementation will log the message in the storage object.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Log Method

Generic message logging.

public void Log (string type, string format, params object[] args)

⊟ Parameters

type

Log type.

format

Log message format.

args

Format arguments.

⊟ Remarks

The default implementation will log the message in the storage object.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Name Property

The optimizer name.

public string Name { get; }

⊟ Value

The optimizer name.

⊟ Remarks

Get the optimizer name. The name is derived from the class name by default. If you want to specify an alternative name, you can apply the Optimization.Attributes.OptimizerAttribute to your optimizer subclass.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Next Method

Create new optimization iteration.

public virtual bool Next ()

⊟ Returns

true if a new iteration was created, false if the optimization is finished.

⊟ Remarks

Call this function when all the solutions in the current iteration have been processed and their fitnesses updated. This method will in turn:
Update the current best solution
Save the current iteration in the storage
Increment the current iteration (Optimizer.IncreaseIteration)
Check if the optimization is finished (Optimizer.Finished)
Create a new population (Optimizer.Update)

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Parameter Method

Retrieve parameter by name.

public Parameter Parameter (string name)

⊟ Parameters

name

Parameter name.

⊟ Returns

The parameter.

⊟ Remarks

Get a parameter by name.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Parameters Property

Get parameters.

public List<Parameter> Parameters { get; }

⊟ Value

List of parameters.

⊟ Remarks

Get the list of parameters that serve as a template for parameters in the solutions.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Population Property

Current population of solutions.

public List<Solution> Population { get; }

⊟ Value

List of solutions.

⊟ Remarks

Get the current population of solutions.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Remove Method

Remove solution from the population.

public virtual void Remove (Solution solution)

⊟ Parameters

solution

Solution.

⊟ Remarks

Remove a solution from the current population.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ SaveIteration Method

Save iteration to storage.

protected virtual void SaveIteration ()

⊟ Remarks

Save the current iteration to storage.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Setup Method

Setup the optimizer.

protected virtual void Setup ()

⊟ Remarks

Setup the optimizer. This is called at the end of the Optimizer.Initialize method and at the end of the Optimizer.FromStorage method. Override this to setup anything after the initialization of the optimizer is complete.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ State Property

Optimizer state.

public State State { get; }

⊟ Value

Optimizer state.

⊟ Remarks

Get the optimizer state. The state contains certain state information that is shared with the solutions, such as a random number generator and the optimizer settings.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Storage Property

Storage object.

public Optimization.Storage.Storage Storage { set; get; }

⊟ Value

Storage object.

⊟ Remarks

Get or set the storage object used for storing the results. A storage object will be created by default when the optimizer is created, but if necessary you can assign a different one after the optimizer is constructed.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ System.Collections.IEnumerable.GetEnumerator Method

Get solution enumerator.

IEnumerator System.Collections.IEnumerable.GetEnumerator ()

⊟ Returns

Solution enumerator.

⊟ Remarks

Get enumerator which enumerates over the solutions in the current population.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Update Method

Update the current population.

public virtual void Update ()

⊟ Remarks

Update the current solution population using the fitnesses of each solution. This is usually called from the Optimizer.Next method and should not be needed to be called by the user.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ Update Method

Update the solution.

public virtual void Update (Solution solution)

⊟ Parameters

solution

Solution to update.

⊟ Remarks

Convenient function to update a single solution. The default implementation of Optimizer.Update will call this function for each solution in the population. If this does not make sense for your particular optimization, you should override Optimizer.Update.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

⊟ UpdateBest Method

Update the best solution so far.

protected virtual void UpdateBest ()

⊟ Remarks

Update the best solution by looking at the fitnesses of all the solutions in the current population. This function is called from Optimizer.UpdateBest and should not have to be called by the user.

⊟ Requirements

Namespace: Optimization
Assembly: Optimization (in Optimization.dll)

	Add (Solution) Add a new solution to the optimizer.
	AddExtension (Extension) Add extension.
	Boundary (string) : Boundary Retrieve boundary by name.
	CreateSolution (uint) : Solution Create a new solution.
	FromStorage (Optimization.Storage.Storage, Optimization.Storage.Records.Optimizer) Restore optimizer from storage.
	FromStorage (Optimization.Storage.Storage, Optimization.Storage.Records.Optimizer, Optimization.Storage.Records.Solution, Solution) Restore solution from storage.
	FromXml (System.Xml.XmlNode) Load optimizer from XML.
static	GetDescription (Type) : string Get optimizer description.
	GetEnumerator () : IEnumerator<Solution> Get solution enumerator.
static	GetName (Type) : string Get the optimizer name.
	HasBoundary (string) : bool Check boundary existence.
	HasParameter (string) : bool Check parameter existence.
	Initialize () Initialize optimizer.
	InitializePopulation () Initialize solution population.
	Log (string, string) Generic message logging.
	Log (string, string, params object[]) Generic message logging.
	Next () : bool Create new optimization iteration.
	Parameter (string) : Parameter Retrieve parameter by name.
	Remove (Solution) Remove solution from the population.
	Update () Update the current population.
	Update (Solution) Update the solution.

	AddBoundary (Boundary) Add boundary.
	AddParameter (Parameter) Add parameter.
	CreateSettings () : Optimizer.Settings Create settings object.
	CreateState () : State Create state.
	Finished () : bool Check whether the optimization is finished.
	IncrementIteration () Increment the current iteration.
	SaveIteration () Save iteration to storage.
	Setup () Setup the optimizer.
	UpdateBest () Update the best solution so far.

[read-only]	Best	Solution . Get the best best solution so far.
[read-only]	Boundaries	List<Boundary> . Get list of boundaries.
[read-only]	Configuration	Optimizer.Settings . Get optimizer configuration.
	CurrentIteration	uint . The current iteration.
[read-only]	Description	string . Optimizer description.
[read-only]	Extensions	List<Extension> . Extensions.
[read-only]	Fitness	Fitness . The fitness object template.
[read-only]	Name	string . The optimizer name.
[read-only]	Parameters	List<Parameter> . Get parameters.
[read-only]	Population	List<Solution> . Current population of solutions.
[read-only]	State	State . Optimizer state.
	Storage	Optimization.Storage.Storage . Storage object.

Optimizer Class