Differential Evolution
Algorithms AIMech:DifferentialEvolution DRAFT
Aliases: DE
A population-based continuous optimization method that generates trial vectors using scaled differences between population members, combines mutant and target coordinates, and selects replacements through target-trial objective comparisons.
Identity Notes & Concept Scope
This record denotes differential evolution, with the 1997 DE/rand/1/bin procedure as its explicitly inspected core. Mutation/crossover variants are distinguished by the source’s DE/x/y/z notation rather than treated as identical executable rules. It differs from generic genetic algorithms, a fitted Gaussian search distribution, objective-gradient descent, and DEHB’s additional budget-specific populations and Hyperband scheduling. No normalized identity match was found across all 442 pinned AIO labels/scoped synonyms and 270 preceding native IDs/names/aliases. Full-checkout fixed-string searches included ignored and hidden files in worktree and main, returning 19 and 19 lines respectively. These were prior candidate leads, source discussions and neighboring DEHB/FABOLAS method mentions. Eight exact AIO CLASS nodes with scoped synonyms and outgoing edges and seven native neighboring mechanisms were inspected. Absence is bounded to this snapshot and semantic neighborhood.
Claims & Grounded Evidence
1 Claim
Mechanism
#method
Maintain a fixed-size population of real-valued vectors. In the inspected DE/rand/1/bin scheme, for each target select three mutually distinct current-generation population indices that also differ from the target index. Add a scaled difference of two selected vectors to the third to form a mutant. Build a trial by independently choosing mutant or target coordinates according to the crossover probability, forcing one randomly selected coordinate to come from the mutant. Evaluate the trial and use a greedy comparison against its target to choose the next-generation member. Complete all target comparisons using the old population before replacing it with the next population. Repeat until the configured stopping criterion is met. Population size, mutation scale, crossover probability and initialization are configuration choices.
⚠️ Curated Limitations & Negative Results
- The prose specifies replacement for a strictly smaller objective, whereas the displayed C-style Figure3 uses <=. The tie policy must therefore be identified when implementing or comparing versions; these source statements are not treated as identical code. A forced mutant coordinate need not change its numerical value.
- DE/rand/1/bin needs at least four population members for its distinct-index rule. Difference-vector proposals can leave the initialization range; that range is not a hard feasible-domain boundary. Bound repair, general constraint handling, noise treatment and categorical encodings are additional choices.
- The method uses function evaluations, not derivatives of the objective. Population collapse can eliminate difference-vector variation, and finite-budget greedy selection does not certify a global optimum. The source’s control-variable suggestions and comparative results are empirical, not universal guarantees or reproduced experiments.
Artificial Intelligence Ontology (AIO) Grounding
Explicit AIO Snapshot Assessment: No Match
Rationale: No inspected AIO target captures this general optimization procedure or a defining component. The assessment distinguishes evolutionary language models, learning/training tasks and predictive models from the native algorithm; it is bounded to this pinned vocabulary.
Search Scope: All 442 pinned AIO CLASS identities and scoped synonyms, all 270 preceding maintained identities, ignored/hidden checkout text, and semantic inspection of eight exact AIO targets plus seven neighboring native mechanisms.
| Examined AIO Candidate | Rejection Reason |
|---|---|
AIO:EvolutionaryLLM
|
The CLASS definition is a large language model with evolutionary optimization. Its broad related synonym evolutionary algorithms does not make a general continuous optimizer a language model or an exact vocabulary match. |
AIO:TrainingStrategy
|
The pinned definition concerns preprocessing/training of machine-learning models. General black-box continuous optimization is not inherently model training; an optional application does not establish a necessary mapping. |
AIO:MachineLearningTask
|
The target denotes building methods that learn from data. These algorithms can optimize arbitrary evaluated functions without a learning task, so shared optimization vocabulary alone is insufficient. |
AIO:MetaLearning
|
Adapting a search distribution or optimizer state does not require learning from metadata about machine-learning experiments as in the pinned definition. |
AIO:ActiveLearning
|
Candidate evaluation is not the defined task of querying labels for data points. |
AIO:ReinforcementLearning
|
A possible policy-search application does not turn the general optimizer into the pinned learning task; no environment/reward-policy interface is inherent to the full identity. |
AIO:RandomForest
|
Population members and search distributions are not a predictive decision-tree ensemble; no random-forest surrogate is required. |
AIO:EvaluationBias
|
This is a bias concept, not the optimizer. Possible evaluation or sampling problems are limitations rather than identity mappings. |
Evidence Sources & Primary Literature
1 Source| Source ID | Publication / Document | Type / Access | Reference & Checksum |
|---|---|---|---|
method-source
|
Rainer Storn, Kenneth Price
|
primary paper
full_text
|
sha256:d8bd6c0f8030…
|
Curation Provenance
Curator: Codex
Method: ai_assisted
Model: GPT-6
Created: 2026-10-11T12:18:29Z
Updated: 2026-10-11T12:18:29Z
Source File:
data/curated/DifferentialEvolution.yaml