Module ep 

Evolutionary Programming (Fogel-style).

Classical EP differs from ES in the details:

• No crossover. Each parent produces exactly one offspring by Gaussian mutation.
• Self-adaptive σ. Each individual carries its own σ, updated by the log-normal rule σ' = σ · exp(τ · N(0, 1)). This is the same mechanism and ordering as the multi-parent ES variants: σ is perturbed first, and the updated σ’ drives that individual’s gene mutation. Survivor σ are inherited, not reset.
• q-tournament survivor selection on the (μ + μ) pool. Each individual plays q random opponents; the μ individuals with the highest win-counts survive. This diverges from truncation selection — EP gives weaker individuals a stochastic chance to survive.

Reference

• Fogel (1994), An introduction to simulated evolutionary optimization.

Structs

EpConfig

Static configuration for an EvolutionaryProgramming run.

EpState

Generation-to-generation state for EvolutionaryProgramming.

EvolutionaryProgramming

Classical Fogel EP.