From febc80c641d63407b64d7f648b6661b91c3a74ed Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Chlo=C3=A9=20Muller?= <chloe.muller@student-cs.fr>
Date: Wed, 29 Jan 2020 14:33:50 +0100
Subject: [PATCH] Ajout de la fonction de comparaison selon la selection
utilisee
---
algogenetique.py | 58 ++++++++++++++++++++++-----
individu.py | 8 ++--
population.py | 102 ++++++++++++++++++++---------------------------
3 files changed, 96 insertions(+), 72 deletions(-)
diff --git a/algogenetique.py b/algogenetique.py
index 138d8a1..3122ce1 100644
--- a/algogenetique.py
+++ b/algogenetique.py
@@ -4,11 +4,12 @@ import numpy
import RotTable
from individu import Individu
from population import Population, afficher
-import croisement
+from croisement import *
from Traj3D import *
from random import random
import matplotlib.pyplot as plt
import time
+from copy import deepcopy
# def main(N,tmax,pmutation, proportion,brin="plasmid_8k.fasta"):
# '''lineList = [line.rstrip('\n') for line in open(brin)]
@@ -29,22 +30,22 @@ import time
# plt.plot([i for i in range(tmax)], L, label = str(pmutation))
# return(best)
-def main(N,tmax,pmutation, proportion):
+def main(N,tmax,pmutation, proportion, indice_selection, population_initiale, enfant = croisement_un_point):
L=[]
lineList = [line.rstrip('\n') for line in open("plasmid_8k.fasta")]
brin = ''.join(lineList[1:])
- People=Population(N)
+ People=deepcopy(population_initiale)
# S1=[]
for individu in People.indiv:
- individu.evaluate(brin)
+ individu.evaluate()
# S1.append(int(individu.score))
# maximum=int(max(S1))
for i in range(tmax):
- #print(i)
+ print(i)
mini=People.indiv[0].score
best=People.indiv[0]
- People.reproduction(p = proportion, proba_mutation= pmutation)
+ People.reproduction(p = proportion, proba_mutation= pmutation, selection = indice_selection, enfant = enfant)
for individu in People.indiv:
if individu.score<mini:
best=individu
@@ -52,7 +53,8 @@ def main(N,tmax,pmutation, proportion):
L.append(mini)
# plt.subplot(221)
- plt.plot([i for i in range(tmax)], L, label = str(pmutation))
+ liste_selections = ["selection_p_best", "selection_duel_pondere", "selection_duel", "selection_par_rang", "selection_proportionnelle"]
+ plt.plot([j for j in range(tmax)], L, label = liste_selections[indice_selection])
# plt.subplot(223)
@@ -79,6 +81,7 @@ def main(N,tmax,pmutation, proportion):
def test_mutation():
+ start_time = time.time()
plt.figure()
for i in range(1,5):
print("\n \n", i)
@@ -91,5 +94,42 @@ def test_mutation():
plt.show()
-start_time = time.time()
-test_mutation()
+def comparaison_selections():
+ liste_selections = ["selection_p_best", "selection_duel_pondere", "selection_duel", "selection_par_rang", "selection_proportionnelle"]
+ liste_time = []
+ plt.figure()
+ People = Population(100)
+ for individu in People.indiv:
+ individu.evaluate()
+ S2=[individu.score for individu in People.indiv]
+ plt.hist(S2, range = (0,int(max(S2)+10)), bins = 20, color = 'blue')
+ plt.show()
+ plt.figure()
+ for i in range(5):
+ print("\n", liste_selections[i], "\n")
+ start_time = time.time()
+ best = main(100, 35, 0.001, 50, i, deepcopy(People))[0]
+ liste_time.append((liste_selections[i], time.time() - start_time, best.score))
+ plt.legend()
+ plt.xlabel("Nombre de générations")
+ plt.ylabel("Score du meilleur individu")
+ plt.title("Comparaison en fonction de la méthode de sélection")
+ print(numpy.array(liste_time))
+ plt.show()
+
+# def comparaisons_croisements():
+# liste_croisements = ["croisement_un_point", "croisement_deux_points"]
+
+
+
+
+# test_mutation()
+
+comparaison_selections()
+
+
+# [['selection_p_best' '85.1275908946991']
+# ['selection_duel_pondere' '85.47507500648499']
+# ['selection_duel' '88.86001086235046']
+# ['selection_par_rang' '87.76551222801208']
+# ['selection_proportionnelle' '87.30216789245605']]
\ No newline at end of file
diff --git a/individu.py b/individu.py
index 08744e8..2047877 100644
--- a/individu.py
+++ b/individu.py
@@ -84,10 +84,10 @@ class Individu():
# print(individu1.table.rot_table)
# individu1.mutation()
-table = RotTable()
-test = Individu(table)
-test.evaluate("AAAGGATCTTCTTGAGATCCTTTTTTTCTGCGCGTAATCTGCTGCCAGTAAACGAAAAAACCGCCTGGGGAGGCGGTTTAGTCGAA")
-print(test.score)
+# table = RotTable()
+# test = Individu(table)
+# test.evaluate("AAAGGATCTTCTTGAGATCCTTTTTTTCTGCGCGTAATCTGCTGCCAGTAAACGAAAAAACCGCCTGGGGAGGCGGTTTAGTCGAA")
+# print(test.score)
# qqun=Individu(RotTable())
diff --git a/population.py b/population.py
index 7d25073..43e3762 100644
--- a/population.py
+++ b/population.py
@@ -22,18 +22,8 @@ class Population:
def selection_p_best(self,p=None):
if p==None:
p=(self.n)//2
-
- def tri_rapide_aux(tableau,debut,fin):
- if debut < fin-1:
- positionPivot=partitionner(tableau,debut,fin)
- tri_rapide_aux(tableau,debut,positionPivot)
- tri_rapide_aux(tableau,positionPivot+1,fin)
-
- def tri_rapide(tableau):
- tri_rapide_aux(tableau,0,len(tableau))
-
liste_individus=self.indiv
- tri_rapide(liste_individus)
+ liste_individus.sort(key = lambda individu : individu.score)
individus_selectionnes = [element for element in liste_individus[:p]]
self = self.modifier_population(individus_selectionnes)
@@ -41,21 +31,23 @@ class Population:
def selection_duel_pondere(self,p=None):
if p == None :
p = (self.n)//2
- newself=[]
- vu=set()
+ meilleur = self.indiv[0]
+ for individu in self.indiv :
+ if meilleur.score > individu.score:
+ meilleur = individu
+ newself=[meilleur]
m=randrange(0,self.n)
t=randrange(0,self.n) #méthode des duels pondérée: si x=10 et y=1, y a une chance sur 11 de passer
+ non_vu = [i for i in range(0, self.n)]
while len(newself)<p:
- while m in vu:
- m=randrange(0,self.n)
- while t in vu:
- t=randrange(0,self.n)
+ m = choice(non_vu)
+ non_vu.remove(m)
+ t = choice(non_vu)
+ non_vu.remove(t)
x=self.indiv[m]
y=self.indiv[t]
- vu.add(t)
- vu.add(m)
- p=random()
- if p>x.score/(x.score+y.score):
+ proba=random()
+ if proba<x.score/(x.score+y.score):
newself.append(y)
else:
newself.append(x)
@@ -68,10 +60,8 @@ class Population:
meilleur = self.indiv[0]
for individu in self.indiv :
if meilleur.score > individu.score:
- #print("meilleur, individu: ", meilleur.score, individu.score)
meilleur = individu
- newself = [meilleur]
- vu=set()
+ newself = [meilleur]
t=randrange(0,self.n)
m=randrange(0,self.n)
non_vu = [i for i in range(0, self.n)]
@@ -95,35 +85,11 @@ class Population:
if p == None :
p = (self.n)//2
liste_individus = self.indiv
- n = self.n
-
- def echanger(tableau, i, j):
- tableau[i], tableau[j] = tableau[j], tableau[i]
-
- def partitionner(tableau,debut,fin):
- echanger(tableau,debut,randint(debut,fin-1))
- partition=debut
- for i in range(debut+1,fin):
- # if tableau[i] < tableau[debut]:
- if tableau[i].score<tableau[debut].score:
- partition+=1
- echanger(tableau,i,partition)
- echanger(tableau,debut,partition)
- return partition
-
- def tri_rapide_aux(tableau,debut,fin):
- if debut < fin-1:
- positionPivot=partitionner(tableau,debut,fin)
- tri_rapide_aux(tableau,debut,positionPivot)
- tri_rapide_aux(tableau,positionPivot+1,fin)
-
- def tri_rapide(tableau):
- tri_rapide_aux(tableau,0,len(tableau))
-
- tri_rapide(liste_individus)
- individus_selectionnes = []
+ n = self.n
+ liste_individus.sort(key = lambda individu : individu.score, reverse = True)
+ individus_selectionnes = [liste_individus[-1]]
- for _ in range(p):
+ for _ in range(p-1):
curseur = random()*n*(n+1)/2
# print("curseur", curseur)
j = 1
@@ -135,26 +101,33 @@ class Population:
self = self.modifier_population(individus_selectionnes)
- def selection_proportionelle(self,p= None):
+ def selection_proportionnelle(self,p= None):
if p == None :
p = (self.n)//2
- newself=[]
+ meilleur = self.indiv[0]
+ for individu in self.indiv :
+ if meilleur.score > individu.score:
+ meilleur = individu
+ newself = [meilleur]
somme=0
for indiv in self.indiv:
- somme=somme+indiv.score
+ somme+=1/indiv.score
while len(newself)<p:
m=m=randrange(0, self.n)
x=self.indiv[m]
- p=random()
- if p<=x.score/somme:
+ proba=random()
+ if proba<=x.score/somme:
newself.append(x)
self = self.modifier_population(newself)
def reproduction(self,proba_mutation = None, selection=None,enfant=croisement_un_point, p = None):
+ liste_selections = [self.selection_p_best, self.selection_duel_pondere, self.selection_duel, self.selection_par_rang, self.selection_proportionnelle]
if proba_mutation == None :
proba_mutation = 0.001
if selection == None :
selection = self.selection_duel
+ else :
+ selection = liste_selections[selection]
if p == None :
p = (self.n)//2
vieille_taille = self.n
@@ -179,7 +152,7 @@ class Population:
def afficher(popu):
for individu in popu.indiv :
print("\n individu \n")
- print(individu.table.rot_table)
+ # print(individu.table.rot_table)
print ("score", individu.score)
def test():
@@ -194,10 +167,21 @@ def test():
#test()
+def test2():
+ popu = Population(10)
+ for individu in popu.indiv :
+ lineList = [line.rstrip('\n') for line in open("plasmid_8k.fasta")]
+ brin = ''.join(lineList[1:])
+ individu.evaluate()
+ print("\n \n POPULATION INITIALE \n \n")
+ afficher(popu)
+ popu.selection_proportionnelle()
+ print("\n\nAPRES SELECTION \n\n")
+ afficher(popu)
-
+# test2()
--
GitLab