Skip to content

README.md

Latest commit

 

History

History
122 lines (94 loc) · 4.01 KB

README.md

File metadata and controls

122 lines (94 loc) · 4.01 KB

01_basic_usage - Getting Started

This folder provides introductory examples to demonstrate the basic usage of EvoLib. They illustrate how populations are created, mutated, and evaluated

Learning Goals

  • Load a population from a YAML config.
  • Inspect individuals and population status.
  • Define and use a custom fitness function in a run
  • Understand manual mutation vs. automated runs.

Prerequisites

None. This is the ideal entry point for new users.

Configuration

All examples load their initial population from a YAML config file.
A minimal example (population.yaml) is provided here for illustration:

parent_pool_size: 2
offspring_pool_size: 4
max_generations: 10
max_indiv_age: 0
num_elites: 0

evolution:
  strategy: mu_plus_lambda

modules:
  test-vector:
    type: vector
    initializer: random_vector
    dim: 1
    bounds: [-1.0, 1.0]

    mutation:
      strategy: "constant"
      strength: 0.01
      probability: 1.0

Files & Expected Output

Running these scripts will print the population status to the console:

Note: Due to random initialization and stochastic operators (mutation, crossover, selection), the exact numerical values in your output will vary between runs. What matters is the overall pattern: population statistics are printed generation by generation, and fitness values should gradually improve as evolution progresses.

01_getting_started.py:

Minimal example - load a population from a YAML config and print its status.

$ python 01_getting_started.py
Population: Gen:   0 Fit: 0.00000000
Best Indiv age: 0
Max Generation: 0
Number of Indivs: 10
Number of Elites: 0
Population fitness: 0.000
Worst Indiv: 0.000

02_mutation.py:

Shows how to create a single individual, apply mutation manually, and inspect parameter changes.
⚠️ Illustration only: in real experiments, mutation is applied automatically inside the evolution loop.

$ python 02_mutation.py 
Before mutation: {'test-vector': 'Vector=[-0.419, 0.37, 0.944, 0.284] | Global mutation_strength=0.0100'}
After mutation:  {'test-vector': 'Vector=[-0.405, 0.368, 0.932, 0.284] | Global mutation_strength=0.0100'}

03_population_mutation.py:

Applies mutation across all individuals of the initialized population.
⚠️ Illustration only: in practice, this is handled automatically by pop.run() or run_one_generation().

$ python 03_population_mutation.py
Before mutation:
  Indiv 0: {'test-vector': 'Vector=[-0.495, 0.138, 0.069, -0.931] | Global mutation_strength=0.0100'}
  Indiv 1: {'test-vector': 'Vector=[0.962, -0.135, -0.303, -0.281] | Global mutation_strength=0.0100'}
  [...]
After mutation:
  Indiv 0: {'test-vector': 'Vector=[-0.482, 0.147, 0.08, -0.938] | Global mutation_strength=0.0100'}
  Indiv 1: {'test-vector': 'Vector=[0.977, -0.121, -0.315, -0.286] | Global mutation_strength=0.0100'}
  [...]

04_fitness.py:

Demonstrates how to define and register a custom fitness function.
Uses resume_or_create(...) to create or resume a run and executes the evolution loop with pop.run().
This example shows the canonical workflow for actual experiments.

$ python 04_fitness.py 
start: strategy=EvolutionStrategy.MU_PLUS_LAMBDA, parents(mu)=2, offspring(lambda)=4, max_gen=10
Population: Gen:   1 Fit: 0.01941376
Population: Gen:   2 Fit: 0.01743742
[...]
Population: Gen:  10 Fit: 0.00638989

See Also

Notes

These examples are minimal and for illustration purposes. They are meant to illustrate concepts step by step:

  • How EvoLib represents populations and individuals
  • How mutation affects parameters
  • How to attach a fitness function and run evolution For practical experiments, always prefer the higher-level API (resume_or_create, pop.run()).