def kakak():
ga_tsp = GA_TSP(func=cal_total_weight,
n_dim=flow_nums,
size_pop=50,
max_iter=200,
prob_mut=0.06)
best_goods, best_value = ga_tsp.run()
print(best_goods, best_value)
point2(ga_tsp)
def _GA_start(flows_len):
if(flows_len%2!=0):
flows_len=flows_len-1
len=flows_len
ga_tsp = GA_TSP(func=_GA_Fit, n_dim=len, size_pop=len, max_iter=200,prob_mut=0.06)
best_goods, best_value = ga_tsp.run()
print(best_goods, best_value)
#point2(ga_tsp)
return best_goods
def GA_go(self):
ga_tsp = GA_TSP(func=self.cal_total_weight,
points=self.flow,
pop=50,
max_iter=100,
Pm=0.01)
best_goods, best_value = ga_tsp.fit()
print(best_goods, best_value)
Y_history = pd.DataFrame(ga_tsp.FitV_history)
fig, ax = plt.subplots(2, 1)
ax[0].plot(Y_history.index, Y_history.values, '.', color='red')
Y_history.min(axis=1).cummin().plot(kind='line')
plt.show()
def kaka(self):
ga_tsp = GA_TSP(func=cal_total_weight,
n_dim=flow_nums,
size_pop=flow_nums,
max_iter=500,
prob_mut=0.01)
best_goods, best_value = ga_tsp.run()
print(best_goods, best_value)
fit_history = pd.DataFrame(ga_tsp.all_history_Y)
fig, ax = plt.subplots(2, 1)
ax[0].plot(fit_history.index, fit_history.values, '.', color='red')
plt_min = fit_history.min(axis=1)
ax[1].plot(plt_min.index, plt_min, Label='min')
ax[1].plot(plt_min.index, plt_min.cummin())
# fit_history.max(axis=1).cummin().plot(kind='line')
plt.show()
def Get_Data(_num_points=20):
num_points = _num_points
points_coordinate = np.random.rand(num_points,
2) * 10 # generate coordinate of points
distance_matrix = spatial.distance.cdist(points_coordinate,
points_coordinate,
metric='euclidean')
def cal_total_distance(routine):
'''The objective function. input routine, return total distance.
cal_total_distance(np.arange(num_points))
'''
num_points, = routine.shape
return sum([
distance_matrix[routine[i % num_points],
routine[(i + 1) % num_points]]
for i in range(num_points)
])
# %% do GA
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=num_points,
size_pop=50,
max_iter=500,
prob_mut=1)
best_points, best_distance = ga_tsp.run()
return points_coordinate.tolist(), best_points, best_distance
def find_path(dist_mat, num_points, max_iter=1000):
distance_matrix = dist_mat
def cal_total_distance(routine):
num_points, = routine.shape
return sum([
distance_matrix[routine[i % num_points],
routine[(i + 1) % num_points]]
for i in range(num_points)
])
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=num_points,
size_pop=200,
max_iter=max_iter,
prob_mut=0.4)
return ga_tsp
# 目标函数，输入路径 返回总距离
# 使用方式：compute_distance(np.arange(num_points))
def compute_distance(routine):
num_points, = routine.shape
#print(routine)
#print(routine.shape)
# 求和
return sum([
distance_matrix[routine[i % num_points], routine[(i + 1) % num_points]]
for i in range(num_points)
])
# 遗传算法
from sko.GA import GA_TSP
# prob_mut 种群的新成员由变异而非交叉得来的概率
ga_tsp = GA_TSP(func=compute_distance,
n_dim=num_points,
size_pop=50,
max_iter=500,
prob_mut=0.2)
best_points, best_distance = ga_tsp.run()
# 画图
fig, ax = plt.subplots(1, 2)
best_points_ = np.concatenate([best_points, [best_points[0]]])
best_points_coordinate = points_coordinate[best_points_, :]
ax[0].plot(best_points_coordinate[:, 0], best_points_coordinate[:, 1], 'o-r')
ax[1].plot(ga_tsp.generation_best_Y)
plt.show()
distanceMatrix = spatial.distance.cdist(pointsCoordinate,
pointsCoordinate,
metric='euclidean')
def getTotalDistance(routine):
num_points, = routine.shape
return sum([
distanceMatrix[routine[i % num_points], routine[(i + 1) % num_points]]
for i in range(num_points)
])
ga_tsp = GA_TSP(func=getTotalDistance,
n_dim=NUM_POINT,
size_pop=50,
max_iter=500,
prob_mut=1)
best_points, best_distance = ga_tsp.run()
print('\n遗传旅行商\n')
print(best_points, '\nbest Distance:', best_distance)
# fig, ax = plt.subplots(1, 2)
# best_points_ = np.concatenate([best_points, [best_points[0]]])
# best_points_coordinate = pointsCoordinate[best_points_, :]
# ax[0].plot(best_points_coordinate[:, 0], best_points_coordinate[:, 1], 'o-r')
# ax[1].plot(ga_tsp.generation_best_Y)
# plt.show()
'''
粒子群算法
'''
'''The objective function. input routine, return total distance.
cal_total_distance(np.arange(num_points))
'''
num_points, = routine.shape
return sum([
distance_matrix[routine[i % num_points], routine[(i + 1) % num_points]]
for i in range(num_points)
])
from sko.GA import GA_TSP
# 3 * size_pop * max_iter
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=num_points,
size_pop=100,
max_iter=2400,
prob_mut=0.07)
ga_tsp.register(operator_name='crossover', operator=operators.crossover_ox)
best_points, best_distance = ga_tsp.run()
X = data_cities[best_points, 1]
Y = data_cities[best_points, 2]
plt.figure(figsize=(12, 8))
# plt.title('TSP Graph')
plt.title('GA: ' + str(best_distance))
plt.xlabel('X')
plt.ylabel('Y')
plt.scatter(X, Y, s=140)
def cal_total_distance(routine):
'''The objective function. input routine, return total distance.
cal_total_distance(np.arange(num_points))
'''
num_points, = routine.shape
return sum([
distance_matrix[routine[i % num_points], routine[(i + 1) % num_points]]
for i in range(num_points)
])
# %% do GA
from sko.GA import GA_TSP
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=num_points,
size_pop=300,
max_iter=800,
Pm=0.3)
best_points, best_distance = ga_tsp.run()
# %% plot
fig, ax = plt.subplots(1, 1)
best_points_ = np.concatenate([best_points, [best_points[0]]])
best_points_coordinate = points_coordinate[best_points_, :]
ax.plot(best_points_coordinate[:, 0], best_points_coordinate[:, 1], 'o-r')
plt.show()
def runGA():
ga_tsp = GA_TSP(func=cal_total_distance, n_dim=num_points, size_pop=8, max_iter=10, prob_mut=0.1)
return ga_tsp.run()
# demo: GA for TSP with UDF(user defined functions)
import numpy as np
# step1: randomly generate the data
from sko.demo_func import function_for_TSP
num_points, points_coordinate, distance_matrix, cal_total_distance = function_for_TSP(
num_points=15)
# %%step2: DO GA with UDF
from sko.GA import GA_TSP
from sko.operators import ranking, selection, crossover, mutation
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=num_points,
size_pop=50,
max_iter=100,
prob_mut=1)
ga_tsp.register('selection', selection.selection_tournament, tourn_size=3). \
register('mutation', mutation.mutation_reverse)
best_points, best_distance = ga_tsp.run()
print('best routine:', best_points, 'best_distance:', best_distance)
# %%
# step3: plot
import matplotlib.pyplot as plt
fig, ax = plt.subplots(1, 1)
best_points_ = np.concatenate([best_points, [best_points[0]]])
best_points_coordinate = points_coordinate[best_points_, :]
ax.plot(best_points_coordinate[:, 0], best_points_coordinate[:, 1], 'o-r')
plt.show()
# 距离矩阵
distance_matrix = spatial.distance.cdist(points_coordinate,
points_coordinate,
metric='euclidean')
# 计算总距离
def cal_total_distance(routine):
num_points, = routine.shape
return sum([
distance_matrix[routine[i % num_points], routine[(i + 1) % num_points]]
for i in range(num_points)
])
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=num_points,
size_pop=50,
max_iter=500,
prob_mut=1)
best_points, best_distance = ga_tsp.run()
print('best_points:', best_points)
print('best_distance:', best_distance)
fig, ax = plt.subplots(1, 2)
# 连接两个矩阵
best_points_ = np.concatenate([best_points, [best_points[0]]])
best_points_coordinate = points_coordinate[best_points_, :]
ax[0].plot(best_points_coordinate[:, 0], best_points_coordinate[:, 1], 'o-r')
ax[1].plot(ga_tsp.generation_best_Y)
plt.show()
np.random.seed(2021)
start = pd.Timestamp.now()
dim = 10 # initial number of breweries from which to start
incr = 5 # increment by this number of breweries
best_distance = 0
# Find out optimal number dims (breweries)
while best_distance < MAX_DISTANCE:
dim += incr
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=dim,
size_pop=num_points,
max_iter=200,
prob_mut=0.7)
best_points, best_distance = ga_tsp.run()
# Retrain final result with more iterations
dim -= 5
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=dim,
size_pop=num_points,
max_iter=500,
prob_mut=0.7)
best_points, best_distance = ga_tsp.run()
# reorder the whole route so that start and end is at HOME (index 0)
ls = list(best_points)
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sko.demo_func import function_for_TSP
num_points, points_coordinate, distance_matrix, cal_total_distance = function_for_TSP(
num_points=50)
# %%
from sko.GA import GA_TSP
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=num_points,
pop=500,
max_iter=2000,
Pm=0.3)
best_points, best_distance = ga_tsp.fit()
# %%
fig, ax = plt.subplots(1, 1)
best_points_ = np.concatenate([best_points, [best_points[0]]])
best_points_coordinate = points_coordinate[best_points_, :]
ax.plot(best_points_coordinate[:, 0], best_points_coordinate[:, 1], 'o-r')
plt.show()
#%%
# import matplotlib.pyplot as plt
#
# Y_history = ga_tsp.all_history_Y
'''
problem_set = 'kroC100'
size_pop = 30
prob_mut = 0.5
setting = '%s_A2' % problem_set
start = time.process_time()
nodeMap, nodeDis = readData('tsp_data/%s.tsp' % problem_set)
opt = readOpt('tsp_data/%s.opt.tour' % problem_set)
opt_distance = calculateTotalDis(opt)
opt = np.concatenate([opt, [opt[0]]])
fig, ax = plt.subplots(2, 2)
ga_tsp = GA_TSP(func=calculateTotalDis,
n_dim=len(nodeDis),
size_pop=size_pop,
max_iter=2000,
prob_mut=prob_mut)
best_points, best_distance = ga_tsp.run(1)
best_points_ = np.concatenate([best_points, [best_points[0]]])
ax[0][0].plot([nodeMap[p][0] for p in best_points_],
[nodeMap[p][1] for p in best_points_], 'o-r')
ax[0][0].set_title('GA Initial solution')
for itr in range(100):
best_points, best_distance = ga_tsp.run(100)
best_points_ = np.concatenate([best_points, [best_points[0]]])
# ax[itr].plot([nodeMap[p][0] for p in best_points_], [nodeMap[p][1] for p in best_points_], 'o-r')
print('itr', itr * 100, '. distance %.2f' % best_distance,
'dif to opt %.2f' % (best_distance - opt_distance))
if best_distance < opt_distance + 0.01:
break
cal_total_distance(np.arange(num_points))
'''
num_points, = routine.shape
return sum([
distance_matrix[routine[i % num_points], routine[(i + 1) % num_points]]
for i in range(num_points)
])
# %% do GA
from sko.GA import GA_TSP
start = time.time()
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=distance_matrix.shape[0],
size_pop=50,
max_iter=500,
prob_mut=0.2)
best_points, best_distance = ga_tsp.run()
end = time.time()
print(best_points)
print(best_distance)
print("程序的运行时间是：%s" % (end - start))
np.save('../data/all_feature/all_feature_sequence.npy', best_points)
# 读入feature.name文件 162
f_name = []
with open('../data/feature.name', 'r') as f_name_file:
for line in f_name_file:
f_name.append(line.strip())
c += 1
else:
capacity = max(capacity, c + 1)
c = 0
capacity = max(capacity, c + 1)
return capacity - max_capacity
# %%
from sko.GA import GA_TSP
ga_tsp = GA_TSP(
func=cal_total_distance,
n_dim=num_customers,
size_pop=50,
max_iter=500,
prob_mut=1,
)
# The index of customers range from 1 to num_customers:
ga_tsp.Chrom = np.concatenate([
np.zeros(shape=(ga_tsp.size_pop, num_vehicle - 1), dtype=np.int),
ga_tsp.Chrom + 1
],
axis=1)
ga_tsp.has_constraint = True
ga_tsp.constraint_ueq = [constraint_capacity]
best_points, best_distance = ga_tsp.run()
# %%
# demo: GA for TSP with UDF(user defined functions)
import numpy as np
# step1: randomly generate the data
from sko.demo_func import function_for_TSP
num_points, points_coordinate, distance_matrix, cal_total_distance = function_for_TSP(
num_points=15)
# step2: DO GA with UDF
from sko.GA import GA_TSP
from sko.GA import selection_tournament, mutation_TSP_1
ga_tsp = GA_TSP(func=cal_total_distance,
n_dim=num_points,
size_pop=500,
max_iter=800,
Pm=0.2)
ga_tsp.register('selection', selection_tournament, tourn_size=3). \
register('mutation', mutation_TSP_1)
best_points, best_distance = ga_tsp.fit()
print('best routine:', best_points, 'best_distance:', best_distance)
# %%
# step3: plot
import matplotlib.pyplot as plt
fig, ax = plt.subplots(1, 1)
best_points_ = np.concatenate([best_points, [best_points[0]]])
best_points_coordinate = points_coordinate[best_points_, :]
ax.plot(best_points_coordinate[:, 0], best_points_coordinate[:, 1], 'o-r')
plt.show()