Skip to main content

> ML_ALGORITHM // DIFFERENTIAL-EVOLUTION-DE_v1.0

Differential Evolution (DE)

Powerful and simple population-based stochastic global optimizer for continuous spaces that perturbs vectors by adding weighted differences of population pairs.

Evolutionary Computationevolutionary-searchhigh-intrinsicsmall (<1k)
Back to All Algorithms
Computational Complexity
Training Complexity:O(generations * population_size * d)
Inference Complexity:O(1)
Hardware Profile
CPU Friendly:Yes
Requires GPU:No
Memory Footprint:low
Interpretability & Data
Interpretability Tier:high-intrinsic
Training Data Needs:small (<1k)

Interpretability Assessment

Direct vector arithmetic mutations make step sizes self-adapting to the spread of active solutions.

Suitable Tasks & Supported Modalities

Suitable Tasks:
continuous blackbox optimizationengineering parameter tuning
Supported Modalities:
tabular

Implementing Libraries

SciPySciPy Community / NumFOCUS · v1.14.1
View Spec
deap

Foundational Literature

Common Pitfalls & Warnings
  • Mutation scale factor F tuned too small (<0.4) causes population collapse into a single cluster without finding global optimum