genetic_algorithms 3.0.0

Library for solving genetic algorithm problems
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
# Examples

> Complete, runnable examples demonstrating every engine and feature.

## Overview

The `examples/` directory contains **19 runnable examples** covering 12 of the 15 engines (GP, CMA-ES, and EDA runnable patterns live in `tests/`), problem
types, and framework features. Each example demonstrates the full workflow: problem definition,
chromosome configuration, operator selection, engine setup, execution, and result interpretation.

Run any example with:

```sh
cargo run --example <name>
```

Some examples require feature flags (e.g., `--features observer-metrics`). See the table below
for details.

## Examples Catalog

| Example | Engine/Feature | Problem Type | Key Concepts |
|---------|---------------|-------------|--------------|
| `rastrigin` | Ga | Continuous optimization | Range genes, Gaussian mutation, tournament selection |
| `nsga2_zdt1` | Nsga2Ga | Bi-objective | Pareto ranking, crowding distance, multi-objective fitness |
| `island_model` | IslandGa | Parallel GA | Migration, sub-populations, coarse-grained parallelism |
| `job_scheduling` | Ga | Permutation/scheduling | Custom crossover (PMX), permutation representation |
| `feature_selection` | Ga | Binary classification | Binary genes, adaptive mutation, feature subsets |
| `niching` | Ga | Multimodal | Fitness sharing, niche radius, multimodal landscape |
| `knapsack_binary` | Ga | Binary/combinatorial | Binary chromosome, weight/value constraints |
| `nqueens_range` | Ga | Constraint satisfaction | Range genes, permutation encoding, penalty fitness |
| `onemax_binary` | Ga | Binary baseline | Binary genes, simple fitness, convergence monitoring |
| `onemax_extension` | Ga | Diversity control | Extension strategies, MassExtinction, diversity triggers |
| `nsga3_dtlz2` | Nsga3Ga | Many-objective (3+) | Reference points, Das-Dennis, DTLZ2 benchmark |
| `moead_dtlz2` | MoeaDGa | Decomposition | Tchebycheff scalarization, weight vectors, DTLZ2 |
| `spea2_zdt1` | Spea2Ga | Archive-based MO | Strength Pareto, k-NN density, ZDT1 benchmark |
| `sms_emoa_zdt1` | SmsEmoaGa | Indicator-based | Hypervolume contribution, steady-state, ZDT1 |
| `ibea_zdt1` | IbeaGa | Indicator-based | I_eps+ indicator, exponential scaling, ZDT1 |
| `aos_demo` | Ga + AOS | Adaptive operators | Operator portfolio, reward accumulation, strategy switching |
| `constrained_g1` | Ga + Constraints | Constrained optimization | Deb's feasibility rules, penalty functions, G1 problem |
| `hall_of_fame_demo` | Ga + HOF | Solution archive | Top-N tracking, deduplication, diversity filtering |
| `memetic_rastrigin` | Ga + Memetic | Local search hybrid | HillClimbing, Lamarckian mode, local refinement |

### Where to find runnable patterns for v3.0.0 features

Variable-length chromosomes and Genetic Programming are not yet covered by dedicated examples in `examples/` — runnable patterns live in the integration test suites instead, and these are the canonical references until standalone examples are added:

| Feature | Reference | Description |
|---------|-----------|-------------|
| Variable-length chromosomes (`ChromosomeLength::Variable`, `Mutation::Insertion`, `Mutation::Deletion`, `Crossover::VariableLength`, parsimony pressure) | [`tests/test_variable_length.rs`]../tests/test_variable_length.rs | End-to-end patterns for length-changing mutation, alignment strategies, and `with_length_penalty` |
| Genetic Programming (`GpGa<N>`, `GpChromosome<N>`, custom `GpNode` impls, subtree/point/hoist mutation, bloat control, serde checkpointing) | [`tests/gp.rs`]../tests/gp.rs | Symbolic regression with `MathNode`, classification trees with `BoolNode`, custom primitive sets |

The [Genetic Programming guide](gp.md) and the variable-length sections in [Chromosomes](chromosomes.md), [Mutation operators](operators/mutation.md), [Crossover operators](operators/crossover.md), and [Survivor operators](operators/survivor.md) walk through these APIs.

## Detailed Walkthroughs

The following sections walk through four representative examples in detail, showing the
complete workflow from problem definition to result interpretation.

### Walkthrough 1: `rastrigin` — Standard GA on Continuous Optimization

**Source:** [`examples/rastrigin.rs`](../examples/rastrigin.rs)

This is the most common starting point for new users. It minimizes the Rastrigin function,
a classic continuous optimization benchmark with many local minima and a global minimum of
0.0 at the origin.

#### Problem Definition

The Rastrigin function is defined as:

```
f(x) = A*n + sum(x_i^2 - A*cos(2*pi*x_i))   where A = 10
```

We optimize in 5 dimensions with each variable in `[-5.12, 5.12]`.

#### Configuration Walkthrough

```rust
use genetic_algorithms::chromosomes::Range as RangeChromosome;
use genetic_algorithms::configuration::ProblemSolving;
use genetic_algorithms::ga::Ga;
use genetic_algorithms::genotypes::Range as RangeGenotype;
use genetic_algorithms::initializers::range_random_initialization;
use genetic_algorithms::operations::{Crossover, Mutation, Selection, Survivor};
use genetic_algorithms::traits::{ChromosomeT, ConfigurationT, CrossoverConfig,
    MutationConfig, SelectionConfig, StoppingConfig};
use genetic_algorithms::{CompositeObserver, LogObserver};
use std::sync::Arc;

// 5-dimensional continuous optimization in [-5.12, 5.12]
const DIMENSIONS: usize = 5;
const POP_SIZE: usize = 100;
const MAX_GENERATIONS: usize = 500;

// Fitness: Rastrigin function (minimize toward 0.0)
let fitness_fn = |dna: &[RangeGenotype<f64>]| -> f64 {
    let a = 10.0;
    let n = dna.len() as f64;
    a * n + dna.iter()
        .map(|g| g.value.powi(2) - a * (2.0 * std::f64::consts::PI * g.value).cos())
        .sum::<f64>()
};

// Allele range for each dimension
let alleles = vec![RangeGenotype::new(0, vec![(-5.12, 5.12)], 0.0_f64)];
let alleles_clone = alleles.clone();

let mut ga = Ga::new()
    .with_genes_per_chromosome(DIMENSIONS)
    .with_population_size(POP_SIZE)
    .with_initialization_fn(move |genes_per_chromosome, _, _| {
        range_random_initialization(genes_per_chromosome, Some(&alleles_clone), Some(false))
    })
    .with_fitness_fn(fitness_fn)
    .with_selection_method(Selection::Tournament)
    .with_crossover_method(Crossover::Uniform)
    .with_mutation_method(Mutation::Gaussian)
    .with_survivor_method(Survivor::Fitness)
    .with_problem_solving(ProblemSolving::Minimization)
    .with_max_generations(MAX_GENERATIONS)
    .with_observer(Arc::new(CompositeObserver::new().add(Arc::new(LogObserver))))
    .build()
    .expect("Failed to build GA configuration");
```

#### Key Lines Explained

- **Chromosome type:** `RangeChromosome<f64>` — a chromosome with `f64` genes, each bounded
  within an interval.
- **Initialization:** `range_random_initialization` randomly places each gene within its
  allele bounds `[-5.12, 5.12]`.
- **Tournament selection:** Standard pressure-based selection — each tournament pits 2 random
  individuals against each other, and the fitter wins.
- **Gaussian mutation:** Adds random Gaussian noise to each gene, appropriate for continuous
  landscapes. Configure sigma via `with_mutation_sigma(...)`.
- **Observer:** `CompositeObserver` fans out to `LogObserver` (console logging) and optionally
  `MetricsObserver` (behind the `observer-metrics` feature flag).

#### Running

```sh
cargo run --example rastrigin
```

#### Expected Output

The GA prints progress every 50 generations, showing best and average fitness:

```
== Rastrigin Continuous Optimization ==
Chromosome: 5 dimensions, Population: 100, Max generations: 500
Operators: Selection=Tournament, Crossover=Uniform, Mutation=Gaussian
-------------------------------------------------------
Generation   50: best =  14.9245, avg =  35.9282
Generation  100: best =   5.9701, avg =  19.3028
...
Generation  500: best =   0.0001, avg =   2.3400
-------------------------------------------------------
Finished. Best fitness: 0.000099
Near-optimal solution found!
```

---

### Walkthrough 2: `nsga2_zdt1` — Multi-Objective GA on ZDT1

**Source:** [`examples/nsga2_zdt1.rs`](../examples/nsga2_zdt1.rs)

This example demonstrates bi-objective optimization using NSGA-II on the ZDT1 benchmark
problem. ZDT1 has two conflicting minimization objectives defined over 30 continuous variables
in `[0, 1]`. The Pareto-optimal front is a convex curve where minimizing f1 forces f2 upward.

#### Problem Definition

```
Variables: x = (x_1, ..., x_30) in [0, 1]
f1(x) = x_1
f2(x) = g(x) * (1 - sqrt(x_1 / g(x)))
g(x)  = 1 + (9 / 29) * sum(x_2, ..., x_30)
```

#### Configuration Walkthrough

```rust
use genetic_algorithms::chromosomes::Range as RangeChromosome;
use genetic_algorithms::configuration::GaConfiguration;
use genetic_algorithms::genotypes::Range as RangeGenotype;
use genetic_algorithms::initializers::range_random_initialization;
use genetic_algorithms::nsga2::configuration::{Nsga2Configuration, ObjectiveDirection};
use genetic_algorithms::nsga2::Nsga2Ga;
use genetic_algorithms::{LogObserver, Nsga2Observer};
use std::sync::Arc;

const N_VARS: usize = 30;
const POP_SIZE: usize = 100;
const MAX_GENERATIONS: usize = 250;

let nsga2_config = Nsga2Configuration::new()
    .with_num_objectives(2)
    .with_population_size(POP_SIZE)
    .with_max_generations(MAX_GENERATIONS)
    .with_objective_directions(vec![
        ObjectiveDirection::Minimize,
        ObjectiveDirection::Minimize,
    ]);

let mut ga_config = GaConfiguration::default();
ga_config.limit_configuration.genes_per_chromosome = N_VARS;

let alleles = vec![RangeGenotype::new(0, vec![(0.0_f64, 1.0_f64)], 0.0_f64)];
let alleles_clone = alleles.clone();

let obj_f1 = |dna: &[RangeGenotype<f64>]| -> f64 { dna[0].value };
let obj_f2 = |dna: &[RangeGenotype<f64>]| -> f64 {
    let n = dna.len();
    let g = 1.0 + (9.0 / (n - 1) as f64)
        * dna[1..].iter().map(|gene| gene.value).sum::<f64>();
    g * (1.0 - (dna[0].value / g).sqrt())
};

let mut nsga2 = Nsga2Ga::<RangeChromosome<f64>>::new(nsga2_config, ga_config)
    .with_alleles(alleles)
    .with_initialization_fn(move |n, _, _| {
        range_random_initialization(n, Some(&alleles_clone), Some(true))
    })
    .with_objective_fns(vec![Box::new(obj_f1), Box::new(obj_f2)])
    .with_observer(
        Arc::new(LogObserver)
            as Arc<dyn Nsga2Observer<RangeChromosome<f64>> + Send + Sync>
    )
    .build()
    .expect("Failed to build NSGA-II");
```

#### Key Lines Explained

- **Nsga2Configuration:** Defines the multi-objective engine parameters — number of objectives,
  population size, and per-objective minimization vs. maximization.
- **GaConfiguration:** The base GA configuration sets chromosome-level parameters (gene count).
  The NSGA-II engine reads its operators from this config.
- **Objective functions:** Each objective is passed as a `Box<dyn Fn(&[Gene]) -> f64>`. The
  engine does not use the scalar `ChromosomeT::fitness()` value — only these objective
  values.
- **Observer:** `LogObserver` implements `Nsga2Observer`, logging Pareto front and crowding
  events.

#### Running

```sh
cargo run --example nsga2_zdt1
```

#### Expected Output

```
== NSGA-II ZDT1 Multi-Objective Optimization ==
Variables: 30, Population: 100, Generations: 250
...
Completed. Pareto front: 100 non-dominated solutions
Pareto front (10 points sampled from 100 non-dominated solutions):
  f1=0.0000, f2=0.3119
  f1=0.1135, f2=0.6590
  f1=0.2148, f2=0.5377
  ...
  f1=0.8932, f2=0.0456
  f1=0.9984, f2=0.0063
```

---

### Walkthrough 3: `constrained_g1` — Constraint Handling with Feasibility Rules

**Source:** [`examples/constrained_g1.rs`](../examples/constrained_g1.rs)

This example solves a constrained optimization problem (simplified G1 benchmark) using a GA
with static penalty functions. It demonstrates the constraint handling framework.

#### Problem Definition

13 real-valued variables in `[0.0, 1.0]` with three inequality constraints:

```
g1: x[0..5] sum <= 4.0
g2: x[5..10] sum <= 3.0
g3: x[10..13] sum <= 2.0
```

Objective: minimize `sum(x)` subject to the three constraints.

#### Configuration Walkthrough

```rust
use genetic_algorithms::chromosomes::Range as RangeChromosome;
use genetic_algorithms::configuration::ProblemSolving;
use genetic_algorithms::constraints::PenaltyStrategy;
use genetic_algorithms::ga::Ga;
use genetic_algorithms::genotypes::Range as RangeGene;
use genetic_algorithms::initializers::range_random_initialization;
use genetic_algorithms::operations::{Crossover, Mutation, Selection, Survivor};
use genetic_algorithms::traits::{
    ChromosomeT, ConfigurationT, CrossoverConfig, MutationConfig,
    SelectionConfig, StoppingConfig,
};

const N_VARS: usize = 13;
const POP_SIZE: usize = 200;
const MAX_GEN: usize = 200;

let alleles = vec![RangeGene::new(0, vec![(0.0_f64, 1.0_f64)], 0.0)];
let alleles_clone = alleles.clone();

// Define inequality constraints as closure functions
let constraints = vec![
    |dna: &[RangeGene<f64>]| {
        let sum: f64 = dna[0..5].iter().map(|g| g.value).sum();
        (sum - 4.0).max(0.0)
    },
    |dna: &[RangeGene<f64>]| {
        let sum: f64 = dna[5..10].iter().map(|g| g.value).sum();
        (sum - 3.0).max(0.0)
    },
    |dna: &[RangeGene<f64>]| {
        let sum: f64 = dna[10..13].iter().map(|g| g.value).sum();
        (sum - 2.0).max(0.0)
    },
];

let fitness_fn = |dna: &[RangeGene<f64>]| -> f64 {
    dna.iter().map(|g| g.value).sum()
};

let mut ga: Ga<RangeChromosome<f64>> = Ga::new()
    .with_genes_per_chromosome(N_VARS)
    .with_population_size(POP_SIZE)
    .with_initialization_fn(move |genes_per_chromosome, _, _| {
        range_random_initialization(genes_per_chromosome, Some(&alleles_clone), Some(false))
    })
    .with_fitness_fn(fitness_fn)
    .with_selection_method(Selection::Tournament)
    .with_crossover_method(Crossover::BlendAlpha)
    .with_mutation_method(Mutation::Gaussian)
    .with_problem_solving(ProblemSolving::Minimization)
    .with_survivor_method(Survivor::Fitness)
    .with_max_generations(MAX_GEN)
    .with_constraint_fns(constraints)
    .with_penalty_strategy(PenaltyStrategy::Static { coefficient: 100.0 })
    .build()
    .expect("Failed to build GA");
```

#### Key Lines Explained

- **constraint_fns:** A `Vec` of closures, each returning a violation magnitude (0.0 if
  satisfied, positive otherwise). The GA sums all violations internally.
- **PenaltyStrategy::Static:** Applies a fixed penalty coefficient to the sum of violations,
  added to the raw fitness. Alternatives include `Dynamic` (increasing over generations),
  `Adaptive`, and `Death` (infeasible individuals are discarded).
- **Deb's feasibility rules:** When comparing two individuals, the GA prioritizes: (1)
  feasible over infeasible, (2) lower violation if both are infeasible, (3) better fitness
  if both are feasible.

#### Running

```sh
cargo run --example constrained_g1
```

#### Expected Output

```
Constrained G1 benchmark with static penalty
Variables: 13, Population: 200, Generations: 200

Constraints:
  g1: x[0..5] sum <= 4.0
  g2: x[5..10] sum <= 3.0
  g3: x[10..13] sum <= 2.0

...
Feasible solution found!
Best fitness: 6.0213
Constraint violations: g1=0.000, g2=0.000, g3=0.000
```

---

### Walkthrough 4: `aos_demo` — Adaptive Operator Selection

**Source:** [`examples/aos_demo.rs`](../examples/aos_demo.rs)

This example demonstrates Adaptive Operator Selection (AOS) using a crossover portfolio
with Probability Matching. The GA dynamically selects among 3 crossover operators based on
their recent performance.

#### Problem Definition

Minimize the sum of 8 integer gene values (each in `[0, 100]`). This is a deliberately
simple problem where different crossover operators may perform differently.

#### Configuration Walkthrough

```rust
use genetic_algorithms::chromosomes::Range as RangeChromosome;
use genetic_algorithms::configuration::ProblemSolving;
use genetic_algorithms::ga::Ga;
use genetic_algorithms::genotypes::Range as RangeGene;
use genetic_algorithms::initializers::range_random_initialization;
use genetic_algorithms::operations::{Crossover, Mutation, Selection, Survivor};
use genetic_algorithms::traits::{
    ChromosomeT, ConfigurationT, MutationConfig, SelectionConfig, StoppingConfig,
};

let n: i32 = 8;
let alleles = vec![RangeGene::new(0, vec![(0_i32, 100_i32)], 0)];
let alleles_clone = alleles.clone();

let mut ga: Ga<RangeChromosome<i32>> = Ga::new()
    .with_genes_per_chromosome(n.try_into().unwrap())
    .with_population_size(50)
    .with_max_generations(100)
    .with_initialization_fn(move |genes_per_chromosome, _, _| {
        range_random_initialization(genes_per_chromosome, Some(&alleles_clone), Some(false))
    })
    .with_fitness_fn(|dna: &[RangeGene<i32>]| {
        dna.iter().map(|g| g.value() as f64).sum()
    })
    .with_selection_method(Selection::Tournament)
    // AOS crossover portfolio with 3 operators
    .with_crossover_portfolio(vec![
        Crossover::Uniform,
        Crossover::SinglePoint,
        Crossover::BlendAlpha,
    ])
    .with_mutation_method(Mutation::Swap)
    .with_mutation_probability_max(0.2)
    .with_aos_strategy(genetic_algorithms::aos::AosStrategy::pm_default())
    .with_reward_window(50)
    .with_problem_solving(ProblemSolving::Minimization)
    .with_survivor_method(Survivor::Fitness)
    .with_alleles(alleles)
    .build()
    .expect("Failed to build GA with AOS configuration");
```

#### Key Lines Explained

- **crossover_portfolio:** Instead of a single crossover operator, this provides a `Vec` of
  3 candidates. The AOS system selects among them each generation.
- **AosStrategy::pm_default():** Probability Matching with default parameters (credit-based
  selection). The GA tracks recent fitness improvements from each operator and adjusts
  selection probabilities accordingly.
- **reward_window:** Number of generations used for the exploration-exploitation trade-off.
  During the first half of the window, all operators are explored equally.
- **Mutation::Swap:** Simple swap mutation on the integer gene values, appropriate for the
  permutation-like representation of the gene order.

#### Running

```sh
cargo run --example aos_demo
```

#### Expected Output

```
=== AOS Demo: Crossover Portfolio with Probability Matching ===

Optimization complete!
Best fitness: 246.0000
Best chromosome (first 4 of 8 genes): [23, 17, 35, 42]

=== AOS Demo Complete ===
The GA dynamically selected among Uniform, SinglePoint, and BlendAlpha crossover
operators based on recent fitness improvement (Probability Matching).
```

## Running an Example

All examples are run via `cargo run --example <name>`. For example:

```sh
# Standard GA on continuous optimization
cargo run --example rastrigin

# Multi-objective optimization with NSGA-II
cargo run --example nsga2_zdt1

# Many-objective optimization with NSGA-III (requires 3+ objectives)
cargo run --example nsga3_dtlz2

# Constrained optimization
cargo run --example constrained_g1

# Adaptive operator selection
cargo run --example aos_demo

# Memetic algorithm (GA + local search)
cargo run --example memetic_rastrigin
```

Some examples use optional feature flags:

```sh
# Observer-metrics integration
cargo run --example rastrigin --features observer-metrics

# Benchmark functions (behind feature flag)
cargo run --example nsga3_dtlz2 --features benchmarks
```

Add `--release` for better performance on longer runs:

```sh
cargo run --example rastrigin --release
```

## Related

- [Engines Overview]engines.md — All 12 optimization engines and when to use each
- [Getting Started]getting-started.md — Quick-start guide
- [Configuration Reference]configuration.md — Full builder options and parameter tables
- [README Examples Table]../README.md — Example catalog in the project README
- [Source: examples/]../examples/ — All 19 example source files