Archive.See: Description
| Class | Description |
|---|---|
| AbstractArchive |
This
AbstractArchive provides some common methods for Archive
s. |
| AdaptiveGridArchive |
The
AdaptiveGridArchive uses an adaptive grid in order to bound the
size of the Archive. |
| ArchiveModule |
The
ArchiveModule determines an implementation for the
Archive interface. |
| BoundedArchive |
An
Archive with bounded size. |
| CrowdingArchive | |
| DefaultArchive | |
| PopulationArchive | |
| UnboundedArchive |
An
Archive of unbounded size. |
| Enum | Description |
|---|---|
| ArchiveModule.Type |
Archive type.
|
Provides different implementations for the Archive.
The AbstractArchive provides common methods for archives to assure that no Pareto-dominated individual remains in the archive.
It should be used for all implementations of Archive.
The BoundedArchive is an abstract class which provides common methods for bound archives, i.e. archives that have a specified maximum size.
Such archives need to decide which non-dominated individuals should be dropped if the maximum size is reached.
The UnboundedArchive stores each found Individual as long as it is not dominated.
Especially for high dimensional and continuous problems, the number of non-dominated individuals can get very high.
The CrowdingArchive uses the crowding distance of NSGA2 to decide which Individuals to drop if the maximum size is reached.
The AdaptiveGridArchive uses an adaptive grid to decide which Individuals to drop.
The PopulationArchive just mirrors the non-dominated Individuals of the current population.
Compared to the bounded archives above, it should only be used for the development or comparison of optimization algorithms.
The DefaultArchive defines the implementation to use if no archive is specified using the ArchiveModule.
The default is the CrowdingArchive with a maximum size of 100 Individuals.
The ArchiveModule allows to select the implementation for the Archive.