darwin

README

darwin

Darwin is a flexible evolutionary computation framework supporting both Genetic Algorithms (GA) and Genetic Programming (GP). It features an extensible architecture with the Evolvable interface, channel-based evolution engine, and async metrics streaming.

Clone and Build

git clone https://github.com/bxrne/darwin.git
cd darwin
go mod tidy
go build ./cmd/darwin

Usage

Basic Run

./darwin

Configuration

Create a custom config file:

[evolution]
population_size = 500
crossover_point_count = 1
crossover_rate = 0.9
mutation_rate = 0.05
generations = 50
elitism_percentage = 0.1
seed = 42

[bitstring_individual]
enabled = true
genome_size = 200

[tree_individual]
enabled = false
max_depth = 1
min_depth = 0
function_set = ["add"]
terminal_set = ["x"]

Parameter	Description	Default
population_size	Number of individuals in population	500
crossover_point_count	Number of crossover points	1
crossover_rate	Probability of crossover (0.0-1.0)	0.9
mutation_rate	Probability of mutation (0.0-1.0)	0.05
generations	Number of evolution generations	50
elitism_percentage	Percentage of best individuals preserved	0.1
seed	Random seed for reproducibility	42

Individual Types

Darwin supports different individual representations:

Bitstring Individuals ([bitstring_individual])

• enabled: Enable bitstring genome evolution
• genome_size: Length of binary genome

Tree Individuals ([tree_individual])

• enabled: Enable tree-based genetic programming
• max_depth: Maximum tree depth
• min_depth: Minimum tree depth
• function_set: Available functions (e.g., ["add", "subtract", "multiply", "divide"])
• terminal_set: Terminal values/variables

Predefined Configurations

The project includes several predefined configurations for different use cases:

• config/small.toml: Quick testing with bitstring individuals (100 pop, 10 gen)
• config/medium.toml: Balanced performance with bitstring individuals (500 pop, 50 gen)
• config/large.toml: Comprehensive evolution with bitstring individuals (2000 pop, 200 gen)
• config/default.toml: Genetic programming with tree individuals

Features

Genetic Programming Support

Darwin includes support for Genetic Programming (GP) with tree-based individuals. Configure [tree_individual] section to enable GP for problems like symbolic regression:

[tree_individual]
enabled = true
max_depth = 3
function_set = ["add", "subtract", "multiply", "divide"]
terminal_set = ["x", "y", "1.0", "2.0"]

Selection Methods

• Roulette Selection: Fitness-proportional selection (default)
• Tournament Selection: Tournament-based selection available

Extensible Architecture

Implement the Evolvable interface to create custom individual types:

type Evolvable interface {
    CalculateFitness()
    Mutate(rate float64)
    GetFitness() float64
    Max(i2 Evolvable) Evolvable
    MultiPointCrossover(i2 Evolvable, crossoverPointCount int) (Evolvable, Evolvable)
}

Async Metrics Streaming

Evolution runs with channel-based communication and provides real-time metrics streaming for monitoring progress.

Benchmarking

Darwin includes comprehensive benchmarking capabilities for performance analysis.

Running Benchmarks

# Run all evolution benchmarks
go test -bench=BenchmarkEvolution ./cmd/darwin -benchmem

# Run specific benchmark sizes
go test -bench=BenchmarkEvolution_Small ./cmd/darwin -benchmem
go test -bench=BenchmarkEvolution_Medium ./cmd/darwin -benchmem
go test -bench=BenchmarkEvolution_Large ./cmd/darwin -benchmem

Benchmark Results

Example output:

BenchmarkEvolution_Small-16    477    2484647 ns/op    1086608 B/op    5698 allocs/op
Config: Population=100, GenomeSize=64, Generations=10, Seed=42
Run 1: Best=0.900, Avg=0.837
Memory: Used=1085744 bytes, TotalAlloc=1085744 bytes

Performance Profiling

# CPU profiling
go test -bench=BenchmarkEvolution ./cmd/darwin -cpuprofile=cpu.prof
go tool pprof cpu.prof

# Memory profiling
go test -bench=BenchmarkEvolution ./cmd/darwin -memprofile=mem.prof
go tool pprof mem.prof

Architecture

Darwin uses a channel-based evolution engine for concurrent processing and thread-safe random number generation. The async metrics streaming allows real-time monitoring of evolution progress.

Testing

Run All Tests

go test ./...

Test Coverage

go test -cover ./...
go test -coverprofile=coverage.out ./...
go tool cover -html=coverage.out