3D graphics for alleles' distribution

compile.bash

@@ -1 +1,2 @@
-g++ starklag.cpp -o starklag -std=c++17
+g++ starklag.cpp -o starklag -std=c++17
+# g++ starklag.cpp -o starklag -std=c++17 -fsanitize=address,undefined

graphics.py

@@ -1,4 +1,6 @@
+import mpl_toolkits.mplot3d as a3
 import matplotlib.pyplot as plt
+import numpy as np
 import sys
 import os
 
@@ -9,6 +11,51 @@
         if file.endswith('.txt'):
             if file in {'debug.txt', 'dna-stats.txt'}:
                 continue
+            if 'alleles' in file: # 3D histogram
+                data = open(f"{directory}/{file}", 'r').read().strip().split('\n')
+                # Data are in the form: "allele:value,allele:value,..." for each line
+                data = [line.split(',') for line in data]
+                # Remove the last element of each line, which is an empty string
+                data = [line[:-1] for line in data]
+                # Data are in the form: [["allele:value", "allele:value", ...], ...]
+                labels: list[str] = []
+                for i in range(len(data)): # iterate over each line
+                    labels.append([pair.split(':')[0] for pair in data[i]])
+                    data[i] = [int(pair.split(':')[1]) for pair in data[i]]
+                # Add padding to the data so that it is a square matrix
+                max_len_line = 0
+                for index, line in enumerate(data):
+                    if len(line) > len(data[max_len_line]):
+                        max_len_line = index
+                # Padding with zeros may be needed on the left or on the right
+                for index, line in enumerate(data):
+                    if len(line) < len(data[max_len_line]):
+                        left_padding_needed = labels[max_len_line].index(labels[index][0])
+                        right_padding_needed = len(data[max_len_line]) - len(data[index]) - left_padding_needed
+                        data[index] = [0] * left_padding_needed + data[index] + [0] * right_padding_needed
+                # Data are in the form: [["value", "value", ...], ...]
+                fig = plt.figure()
+                ax = fig.add_subplot(111, projection='3d')
+                data_array = np.array(data)
+                z_data = [10*i for i in range(len(data_array))]
+                x_data, y_data = np.meshgrid(
+                    np.arange(data_array.shape[1]),
+                    np.arange(data_array.shape[0])
+                )
+                x_data = x_data.flatten()
+                y_data = y_data.flatten()
+                z_data = np.zeros(data_array.shape).flatten()
+                ax.bar3d(x_data, y_data, z_data, 1, 1, data_array.flatten())
+                ax.set_xticks(np.arange(data_array.shape[1]))
+                ax.set_xticklabels(labels[max_len_line], fontsize=8)
+                plt.title(file[:-4])
+                plt.xlabel('Alleles')
+                plt.ylabel('Time')
+                # plt.set_zlabel('Frequency')
+                plt.savefig(f'{directory}/{file[:-4]}.png')
+                plt.show()
+                continue
+            # 2D graphic
             ys = list(map(float, open(f'{directory}/{file}', 'r').read().strip().split('\n')))
             xs = [10 * i for i in range(len(ys))]
 

organism.cpp

@@ -2,7 +2,6 @@
 #include "dna.cpp"
 #include <random>
 #include <chrono>
-#include <thread>
 #define DEBUG 1
 
 

starklag.cpp

@@ -1,5 +1,5 @@
-#include "organism.cpp"
-#include <set>
+#include "stats.cpp"
+#include <thread>
 #ifdef _WIN32
     #include <windows.h>
 #endif
@@ -46,25 +46,6 @@
     }
 #endif
 
-std::ofstream speed_stats("speed-stats.txt");
-std::ofstream strength_stats("strength-stats.txt");
-std::ofstream fertility_stats("fertility-stats.txt");
-std::ofstream nature_stats("nature-stats.txt");
-std::ofstream lifespan_stats("lifespan-stats.txt");
-std::ofstream attack_stats("attack-stats.txt");
-std::ofstream defense_stats("defense-stats.txt");
-std::ofstream vision_stats("vision-stats.txt");
-std::unordered_map<Gene, std::ofstream*> gene_stats = {
-    {Gene::SPEED, &speed_stats},
-    {Gene::STRENGTH, &strength_stats},
-    {Gene::FERTILITY, &fertility_stats},
-    {Gene::NATURE, &nature_stats},
-    {Gene::LIFESPAN, &lifespan_stats},
-    {Gene::ATTACK, &attack_stats},
-    {Gene::DEFENSE, &defense_stats},
-    {Gene::VISION, &vision_stats}
-};
-
 
 bool isDead(Organism* organism) {
     for (std::vector<Organism*>::iterator it = Organism::dead_organisms.begin(); it != Organism::dead_organisms.end(); it++) {
@@ -230,52 +211,8 @@ int main() {
         }
     });
 
-    std::ofstream dna_stats("dna-stats.txt");
-    std::ofstream organisms_count("organisms-count.txt");
     for (int _ = 0; !quit; _++) {
-        organisms_count << Organism::organisms.size() << std::endl;
-        // I use map<Gene, ...> and not vector<...> just for the purpose of being explicit
-        std::unordered_map<Gene, int> sum;
-        std::unordered_map<Gene, std::vector<int>> values_found; // values_found[gene] = all values found for that gene
-        std::unordered_map<Gene, std::unordered_map<int, int>> values; // values[gene][allele] = counter of that allele
-        for (Gene gene : genes) {
-            sum.insert({gene, 0});
-            values_found.insert({gene, std::vector<int>()});
-            values.insert({gene, std::unordered_map<int, int>()});
-        }
-        for (Organism* organism : Organism::organisms) {
-            DNA* dna = organism->dna;
-            for (Allele* allele : dna->alleles) {
-                if (std::find(values_found[allele->name].begin(), values_found[allele->name].end(), allele->value) == values_found[allele->name].end()) {
-                    values[allele->name][allele->value] = 1;
-                    values_found[allele->name].push_back(allele->value);
-                } else {
-                    values[allele->name][allele->value]++;
-                }
-                sum[allele->name] += allele->value;
-            }
-        }
-        std::unordered_map<Gene, float> averages;
-        for (Gene gene : genes) {
-            averages.insert({gene, (float)sum[gene]/(float)Organism::organisms.size()});
-        }
-        // Output the statistics to dna-stats.txt
-        dna_stats << "== DNA stats at " << _ << " decaframes from start ==\n\n";
-        dna_stats << "Allele's distribution:\n";
-        for (Gene gene : genes) {
-            dna_stats << "- " << gene_to_string[gene] << " (average: " << averages[gene] << ")\n";
-            *gene_stats[gene] << averages[gene] << std::endl;
-            std::sort(values_found[gene].begin(), values_found[gene].end());
-            for (int value : values_found[gene]) {
-                if (std::find(possible_random_allele_values[gene].begin(), possible_random_allele_values[gene].end(), value) != possible_random_allele_values[gene].end()) {
-                    dna_stats << "\t" << allele_to_string[gene][value] << " : " << values[gene][value] << "\n";
-                } else {
-                    dna_stats << "\t" << value << " : " << values[gene][value] << "\n";
-                }
-            }
-        }
-        dna_stats << std::endl;
-
+        dumpStats(_);
         for (int i = 0; i < 10; i++) {
             while (paused) {
                 std::this_thread::sleep_for(std::chrono::milliseconds(10));

stats.cpp

@@ -0,0 +1,91 @@
+#include "organism.cpp"
+
+
+std::ofstream dna_stats("dna-stats.txt");
+std::ofstream organisms_count("organisms-count.txt");
+
+std::ofstream speed_stats("speed-stats.txt");
+std::ofstream strength_stats("strength-stats.txt");
+std::ofstream fertility_stats("fertility-stats.txt");
+std::ofstream nature_stats("nature-stats.txt");
+std::ofstream lifespan_stats("lifespan-stats.txt");
+std::ofstream attack_stats("attack-stats.txt");
+std::ofstream defense_stats("defense-stats.txt");
+std::ofstream vision_stats("vision-stats.txt");
+std::unordered_map<Gene, std::ofstream*> gene_stats = {
+    {Gene::SPEED, &speed_stats},
+    {Gene::STRENGTH, &strength_stats},
+    {Gene::FERTILITY, &fertility_stats},
+    {Gene::NATURE, &nature_stats},
+    {Gene::LIFESPAN, &lifespan_stats},
+    {Gene::ATTACK, &attack_stats},
+    {Gene::DEFENSE, &defense_stats},
+    {Gene::VISION, &vision_stats}
+};
+std::ofstream speed_alleles_stats("speed-alleles-stats.txt");
+std::ofstream strength_alleles_stats("strength-alleles-stats.txt");
+std::ofstream fertility_alleles_stats("fertility-alleles-stats.txt");
+std::ofstream nature_alleles_stats("nature-alleles-stats.txt");
+std::ofstream lifespan_alleles_stats("lifespan-alleles-stats.txt");
+std::ofstream attack_alleles_stats("attack-alleles-stats.txt");
+std::ofstream defense_alleles_stats("defense-alleles-stats.txt");
+std::ofstream vision_alleles_stats("vision-alleles-stats.txt");
+std::unordered_map<Gene, std::ofstream*> alleles_stats = {
+    {Gene::SPEED, &speed_alleles_stats},
+    {Gene::STRENGTH, &strength_alleles_stats},
+    {Gene::FERTILITY, &fertility_alleles_stats},
+    {Gene::NATURE, &nature_alleles_stats},
+    {Gene::LIFESPAN, &lifespan_alleles_stats},
+    {Gene::ATTACK, &attack_alleles_stats},
+    {Gene::DEFENSE, &defense_alleles_stats},
+    {Gene::VISION, &vision_alleles_stats}
+};
+
+
+void dumpStats(int _) {
+    organisms_count << Organism::organisms.size() << std::endl;
+    // I use map<Gene, ...> and not vector<...> just for the purpose of being explicit
+    std::unordered_map<Gene, int> sum;
+    std::unordered_map<Gene, std::vector<int>> values_found; // values_found[gene] = all values found for that gene
+    std::unordered_map<Gene, std::unordered_map<int, int>> values; // values[gene][allele] = counter of that allele
+    for (Gene gene : genes) {
+        sum.insert({gene, 0});
+        values_found.insert({gene, std::vector<int>()});
+        values.insert({gene, std::unordered_map<int, int>()});
+    }
+    for (Organism* organism : Organism::organisms) {
+        DNA* dna = organism->dna;
+        for (Allele* allele : dna->alleles) {
+            if (std::find(values_found[allele->name].begin(), values_found[allele->name].end(), allele->value) == values_found[allele->name].end()) {
+                values[allele->name][allele->value] = 1;
+                values_found[allele->name].push_back(allele->value);
+            } else {
+                values[allele->name][allele->value]++;
+            }
+            sum[allele->name] += allele->value;
+        }
+    }
+    std::unordered_map<Gene, float> averages;
+    for (Gene gene : genes) {
+        averages.insert({gene, (float)sum[gene]/(float)Organism::organisms.size()});
+    }
+    // Output the statistics to dna-stats.txt
+    dna_stats << "== DNA stats at " << _ << " decaframes from start ==\n\n";
+    dna_stats << "Allele's distribution:\n";
+    for (Gene gene : genes) {
+        dna_stats << "- " << gene_to_string[gene] << " (average: " << averages[gene] << ")\n";
+        *gene_stats[gene] << averages[gene] << std::endl;
+        std::sort(values_found[gene].begin(), values_found[gene].end());
+        for (int value : values_found[gene]) {
+            if (std::find(possible_random_allele_values[gene].begin(), possible_random_allele_values[gene].end(), value) != possible_random_allele_values[gene].end()) {
+                dna_stats << "\t" << allele_to_string[gene][value] << " : " << values[gene][value] << "\n";
+                *alleles_stats[gene] << allele_to_string[gene][value] << ":" << values[gene][value] << ",";
+            } else {
+                dna_stats << "\t" << value << " : " << values[gene][value] << "\n";
+                *alleles_stats[gene] << value << ":" << values[gene][value] << ",";
+            }
+        }
+        *alleles_stats[gene] << std::endl;
+    }
+    dna_stats << std::endl;
+}