revonet 0.2.1

Rust implementation of real-coded genetic algorithm for solving optimization problems and training of neural networks. The latter is also known as neuroevolution.
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

use ::*;
use ea::*;
use math::*;


/// Structure to hold results for the genetic algorithm run.
#[derive(Clone, Debug, Deserialize, Serialize)]
pub struct EAResult<T: Individual> {
    /// Array of minimal absolute values of fitness for each generation.
    pub min_fitness: Vec<f32>,
    /// Array of maximal absolute values of fitness for each generation.
    pub max_fitness: Vec<f32>,
    /// Array of average absolute values of fitness for each generation.
    pub avg_fitness: Vec<f32>,
    /// Best individual ever found during the single run.
    pub best: T,
    /// Number of function evaluations required to find the `best` individual.
    pub best_fe_count: u32,
    /// Number of function evaluations required to find the solution according to the `OptProblem::is_solution` function.
    pub first_hit_fe_count: u32,
    /// Total number of function evaluations used in the current run.
    pub fe_count: u32,
}

impl<T: Individual> EAResult<T> {
    /// Initialize empty result structure.
    pub fn new() -> EAResult<T> {
        EAResult{avg_fitness: Vec::new(),
                 min_fitness: Vec::new(),
                 max_fitness: Vec::new(),
                 best: T::new(),
                 best_fe_count: 0,
                 first_hit_fe_count: 0,
                 fe_count: 0,
                 }
    }
}

impl<T: Individual+Clone+DeserializeOwned+Serialize> Jsonable for EAResult<T> {
    type T = Self;
}

/// Structure to hold results for multipole runs of evolutionary algorithm.
#[derive(Clone, Debug, Deserialize, Serialize)]
pub struct EAResultMultiple<T: Individual> {
    /// Array of minimal absolute values of fitness for each generation.
    pub min_fitness: Vec<f32>,
    /// Array of maximal absolute values of fitness for each generation.
    pub max_fitness: Vec<f32>,
    /// Array of average absolute values of fitness for each generation.
    pub avg_fitness_mean: Vec<f32>,
    /// Array of SD for average absolute values of fitness for each generation.
    pub avg_fitness_sd: Vec<f32>,
    /// Best individual ever found during the single run.
    pub best: T,
    /// Mean number of function evaluations required to find the `best` individual.
    pub best_fe_count_mean: f32,
    /// SD for number of function evaluations required to find the `best` individual.
    pub best_fe_count_sd: f32,
    /// Mean number of function evaluations required to find the solution according to the `OptProblem::is_solution` function.
    pub first_hit_fe_count_mean: f32,
    /// SD for number of function evaluations required to find the solution according to the `OptProblem::is_solution` function.
    pub first_hit_fe_count_sd: f32,
    /// Number of runs when solution was found.
    pub success_count: u32,
    /// Total number of runs which were performed in order to compute the statistics.
    pub run_count: u32,
}

impl<T: Individual+Clone> EAResultMultiple<T> {
    pub fn new(rs: &[EAResult<T>]) -> EAResultMultiple<T> {
        let mut avg_fitness_mean: Vec<f32> = Vec::new();
        let mut avg_fitness_sd: Vec<f32> = Vec::new();
        let mut min_fitness: Vec<f32> = Vec::new();
        let mut max_fitness: Vec<f32> = Vec::new();
        let mut best_fe_count_mean = 0f32;
        let mut best_fe_count_sd = 0f32;
        let mut first_hit_fe_count_mean = 0f32;
        let mut first_hit_fe_count_sd = 0f32;
        let mut success_count = 0;
        let mut best_fitness = std::f32::MAX;
        let mut best_run_idx = std::usize::MAX;

        let run_count = rs.len();
        for k in 0..run_count {
            let cur_res = &rs[k];
            if k != 0 {
                acc(&mut avg_fitness_mean, &cur_res.avg_fitness);
                acc(&mut avg_fitness_sd, &sqr(&cur_res.avg_fitness));
                min_inplace_vv(&mut min_fitness, &cur_res.min_fitness);
                max_inplace_vv(&mut max_fitness, &cur_res.max_fitness);
            } else {
                avg_fitness_mean = cur_res.avg_fitness.clone();
                avg_fitness_sd = sqr(&cur_res.avg_fitness);
                min_fitness = cur_res.min_fitness.clone();
                max_fitness = cur_res.max_fitness.clone();
            }

            best_fe_count_mean += cur_res.best_fe_count as f32;
            best_fe_count_sd += ((cur_res.best_fe_count) * (cur_res.best_fe_count)) as f32;
            if cur_res.first_hit_fe_count > 0 {
                first_hit_fe_count_mean += cur_res.first_hit_fe_count as f32;
                first_hit_fe_count_sd += ((cur_res.first_hit_fe_count) * (cur_res.first_hit_fe_count)) as f32;
                success_count += 1;
            }

            if rs[k].best.get_fitness() < best_fitness {
                best_fitness = cur_res.best.get_fitness();
                best_run_idx =  k;
            }
        }
        mul_inplace(&mut avg_fitness_mean, 1f32/run_count as f32);
        mul_inplace(&mut avg_fitness_sd, 1f32/run_count as f32);    // compute SD as: mean square  - squared mean
        sub_inplace(&mut avg_fitness_sd, &sqr(&avg_fitness_mean));
        max_inplace_vv(&mut avg_fitness_sd, &vec![0f32; avg_fitness_mean.len()]);
        best_fe_count_mean /= run_count as f32;
        best_fe_count_sd = best_fe_count_sd / run_count as f32 - best_fe_count_mean*best_fe_count_mean;
        if success_count > 0 {
            first_hit_fe_count_mean /= success_count as f32;
            first_hit_fe_count_sd = first_hit_fe_count_sd / success_count as f32 - first_hit_fe_count_mean * first_hit_fe_count_mean;
        }

        EAResultMultiple::<T>{
            avg_fitness_mean: avg_fitness_mean,
            avg_fitness_sd: avg_fitness_sd,
            min_fitness: min_fitness,
            max_fitness: max_fitness,
            // best: ea::Individual::clone(&rs[best_run_idx].best),
            best: rs[best_run_idx].best.clone(),
            best_fe_count_mean: best_fe_count_mean,
            best_fe_count_sd: best_fe_count_sd,
            first_hit_fe_count_mean: first_hit_fe_count_mean,
            first_hit_fe_count_sd: first_hit_fe_count_sd,
            success_count: success_count,
            run_count: run_count as u32,
        }
    }
}

impl<T: Individual+Clone+DeserializeOwned+Serialize> Jsonable for EAResultMultiple<T> {
    type T = Self;
}


//============================================================================================

#[cfg(test)]
mod test {
    #[allow(unused_imports)]
    use ea::*;
    use ga::*;
    use problem::*;
    use result::*;
    use settings::*;

    #[test]
    fn test_json_earesult() {
        let pop_size = 10u32;
        let problem_dim = 5u32;
        let problem = SphereProblem{};

        let gen_count = 10u32;
        let settings = EASettings::new(pop_size, gen_count, problem_dim);
        let mut ga: GA<SphereProblem> = GA::new(&problem);
        let res = ga.run(settings).expect("Error during GA run");

        let filename = "test_json_earesult.json";
        res.to_json(&filename);

        let res2: EAResult<RealCodedIndividual> = EAResult::from_json(&filename);
        assert!(res.best_fe_count == res2.best_fe_count);
        assert!(res.fe_count == res2.fe_count);
        assert!(res.first_hit_fe_count == res2.first_hit_fe_count);
        assert!(res.best.fitness == res2.best.fitness);
    }

    #[test]
    fn test_json_earesult_mult() {
        let pop_size = 10u32;
        let problem_dim = 5u32;
        let problem = SphereProblem{};

        let gen_count = 10u32;
        let ress = (0..3).into_iter()
            .map(|_ | {
                let settings = EASettings::new(pop_size, gen_count, problem_dim);
                let mut ga: GA<SphereProblem> = GA::new(&problem);
                ga.run(settings).expect("Error during GA run").clone()
            })
            .collect::<Vec<EAResult<RealCodedIndividual>>>();

        let filename = "test_json_earesult_mult.json";
        let res = EAResultMultiple::new(&ress);
        res.to_json(&filename);

        let res2: EAResultMultiple<RealCodedIndividual> = EAResultMultiple::from_json(&filename);
        assert!(res.run_count == res2.run_count);
        assert!(res.success_count == res2.success_count);
        assert!(res.first_hit_fe_count_mean == res2.first_hit_fe_count_mean);
        assert!(res.first_hit_fe_count_sd == res2.first_hit_fe_count_sd);
        assert!(res.best_fe_count_mean == res2.best_fe_count_mean);
        assert!(res.best_fe_count_sd == res2.best_fe_count_sd);
        assert!(res.best.fitness == res2.best.fitness);
        assert!(res.best.genes == res2.best.genes);
        assert!(res.min_fitness == res2.min_fitness);
        assert!(res.max_fitness == res2.max_fitness);
        assert!(res.avg_fitness_mean == res2.avg_fitness_mean);
        assert!(res.avg_fitness_sd == res2.avg_fitness_sd);
    }

    #[test]
    fn test_earesult_mult() {
        let pop_size = 10u32;
        let problem_dim = 5u32;
        let problem = SphereProblem{};

        let gen_count = 10u32;
        let ress = (0..3).into_iter()
            .map(|_ | {
                let settings = EASettings::new(pop_size, gen_count, problem_dim);
                let mut ga: GA<SphereProblem> = GA::new(&problem);
                ga.run(settings).expect("Error during GA run").clone()
            })
            .collect::<Vec<EAResult<RealCodedIndividual>>>();

        let filename = "test_json_earesult_mult.json";
        let res = EAResultMultiple::new(&ress);
        res.to_json(&filename);

        assert!(res.run_count == 3);
        assert!(res.min_fitness.len() == gen_count as usize);
        assert!(res.min_fitness.iter().zip(res.max_fitness.iter()).all(|(&min_f, &max_f)| min_f <= max_f));
        assert!(res.min_fitness.iter().zip(res.avg_fitness_mean.iter()).all(|(&min_f, &avg_f)| min_f <= avg_f));
        assert!(res.avg_fitness_sd.iter().all(|&f| f >= 0f32));
        assert!(res.min_fitness[(gen_count-1) as usize] == res.best.fitness);
    }
}