Representations

Representations define genotype domains, validation, repair behavior, random initialization, and concise summaries used in logs/reports.

Contract: Representation<G>.

1) Available Representation Types

Type	Java genotype class	Plugin class	Typical use
`bitstring`	`BitString`	`BitStringRepresentationPlugin`	binary optimization
`int-vector`	`IntVector`	`IntVectorRepresentationPlugin`	bounded integer variables
`categorical-vector`	`CategoricalVector`	`CategoricalVectorRepresentationPlugin`	finite symbols
`mixed-discrete-vector`	`MixedDiscreteVector`	`MixedDiscreteVectorRepresentationPlugin`	multiple categorical arities
`real-vector`	`RealVector`	`RealVectorRepresentationPlugin`	continuous spaces
`mixed-real-discrete-vector`	`MixedRealDiscreteVector`	`MixedRealDiscreteVectorRepresentationPlugin`	mixed-variable problems
`permutation-vector`	`PermutationVector`	`PermutationVectorRepresentationPlugin`	ranking and routing
`variable-length-vector`	`VariableLengthVector<Integer>`	`VariableLengthVectorRepresentationPlugin`	structured variable-length token sequences

2) Parameters by Representation

2.1 `bitstring`

Param	Type	Default
`length`	int	`64`

2.2 `int-vector`

Param	Type	Default
`length`	int	`32`
`min`	int	`0`
`max`	int	`10`

2.3 `categorical-vector`

Param	Type	Default
`length`	int	`16`
`symbols`	string[]	`[A, B, C]`

2.4 `mixed-discrete-vector`

Two configuration styles:

explicit cardinalities:

representation:
  type: mixed-discrete-vector
  cardinalities: [2, 5, 3, 9]

uniform fallback cardinality:

representation:
  type: mixed-discrete-vector
  length: 8
  cardinality: 4

2.5 `real-vector`

Param	Type	Default
`length`	int	`16`
`lower`	double	`-5.0`
`upper`	double	`5.0`

2.6 `mixed-real-discrete-vector`

Param	Type	Default
`realDimensions`	int	`8`
`cardinalities`	int[]	`[4, 4]`
`lower`	double	`-5.0`
`upper`	double	`5.0`

2.7 `permutation-vector`

Param	Type	Default
`size`	int	`20`
`length`	int	alias for `size`

2.8 `variable-length-vector`

Param	Type	Default
`minLength`	int	`2`
`maxLength`	int	`16`
`maxToken`	int	`64`

3) Domain Semantics and Repair

Each representation guarantees:

random(rng) returns valid genotype in the declared domain.
isValid(genotype) enforces domain constraints.
repair(genotype) maps invalid input into domain.

Examples:

real-vector: clips each dimension into [lower, upper].
int-vector: clamps each value into [min, max].
permutation-vector: invalid permutation repairs to identity fallback.
mixed-discrete-vector: values are wrapped via modulo cardinality.
variable-length-vector: pads/truncates to length bounds and normalizes tokens.

4) Distance Utilities

edaf-representations also includes distance helpers:

HammingDistance
EuclideanDistance
KendallTauDistance

These are useful for metrics, niching, or custom diagnostics.

5) Compatibility with Model Families

Validator-enforced mapping:

Discrete representations -> discrete models/algorithms
Continuous representations -> continuous models/algorithms
Permutation representation -> permutation models/algorithms

See Configuration Reference for exact compatibility lists.

6) Example Config Snippets

Bitstring

representation:
  type: bitstring
  length: 128

Real vector

representation:
  type: real-vector
  length: 30
  lower: -30.0
  upper: 30.0

Permutation

representation:
  type: permutation-vector
  size: 50

Mixed real/discrete

representation:
  type: mixed-real-discrete-vector
  realDimensions: 10
  cardinalities: [3, 4, 7]
  lower: -2.0
  upper: 2.0

7) Adding a New Representation

Implement Representation<YourGenotype>.
Implement RepresentationPlugin<YourGenotype>.
Register plugin in:

src/main/resources/META-INF/services/com.knezevic.edaf.v3.core.plugins.RepresentationPlugin

Use type: your-representation-type in config.

Detailed walkthrough: Extending the Framework.

Visual Summary

flowchart LR
    A["EDAF"] --> B["representations"]
    B --> C["Configure"]
    B --> D["Execute"]
    B --> E["Inspect"]
    E --> F["Iterate"]

Estimation of Distribution Algorithms Framework
Copyright (c) 2026 Dr. Karlo Knezevic
Licensed under the Apache License, Version 2.0.