def main():
#take 3 command line arguments for random seed number, number of dimensions, and number of gaussians
seed = int(sys.argv[1])
dims = int(sys.argv[2])
ncenters = int(sys.argv[3])
max_val = 0
np.random.seed(seed)
sog = SG.SumofGaussians(dims, ncenters)
epsilon = 1e-8
# Data
#Get Random Location
current_location = np.random.ranf(dims) * 10
next_location = 0
final_location = 0
final_val = 0
uniform = np.random.uniform(-0.01, 0.01, dims)
T = 10000000
Tmin = 0.000000000000000001
#max_val = sog.Eval(current_location)
print(" ".join(["%.8f" % (x) for x in current_location], ), end=" ")
print("%.8f" % max_val)
counter = 0
while T > Tmin or counter < 2000:
next_location = current_location + np.random.uniform(-0.01, 0.01, dims)
if (sog.Eval(next_location) > sog.Eval(current_location)):
current_location = next_location
max_val = sog.Eval(current_location)
if (sog.Eval(current_location) > sog.Eval(final_location)):
final_location = next_location
final_val = sog.Eval(current_location)
print(" ".join(["%.8f" % (x) for x in current_location], ),
end=" ")
print("%.8f" % max_val)
elif (math.exp(
(sog.Eval(next_location) - sog.Eval(current_location)) / T) >
np.random.random_sample()):
current_location = next_location
max_val = sog.Eval(current_location)
print(" ".join(["%.8f" % (x) for x in current_location], ),
end=" ")
print("%.8f" % max_val)
if (T > Tmin):
T = T - (T * 0.09)
counter += 1
print(" ".join(["%.8f" % (x) for x in final_location], ), end=" ")
print("%.8f" % final_val)
def main():
#take 3 command line arguments for random seed number, number of dimensions, and number of gaussians
seed = int(sys.argv[1])
dims = int(sys.argv[2])
ncenters = int(sys.argv[3])
max_val = 0
step_size = 0.0
inc = True
np.random.seed(seed)
sog = SG.SumofGaussians(dims, ncenters)
epsilon = 1e-8
# Data
#data_input = np.loadtxt(sys.stdin)
current_location = np.random.ranf(dims)
current_location = current_location * 10
max_val = 0
temp_val = 0
step_size = 0
inc = True
location = 0
counter = 0
temp_loc = current_location
step_size = 0.01 * sog.Deriv(current_location)
x_left = temp_loc - step_size
x_right = temp_loc + step_size
right_val = sog.Eval(x_right)
left_val = sog.Eval(x_left)
if (right_val > left_val):
current_location = x_right
while inc == True and counter < 10000:
current_location = current_location + step_size
temp_val = sog.Eval(current_location)
if "%.8f" % temp_val > "%.8f" % max_val:
max_val = temp_val
step_size = 0.01 * sog.Deriv(current_location)
print(" ".join(["%.8f" % (x) for x in current_location], ),
end=" ")
print("%.8f" % max_val)
counter += 1
else:
print(" ".join(["%.8f" % (x) for x in current_location], ),
end=" ")
print("%.8f" % max_val)
inc = False
#x_right += step_size
else:
current_location = x_left
while inc == True and counter < 10000:
current_location = current_location - step_size
temp_val = sog.Eval(current_location)
if "%.8f" % temp_val > "%.8f" % max_val:
max_val = temp_val
step_size = 0.01 * sog.Deriv(current_location)
print(" ".join(["%.8f" % (x) for x in current_location], ),
end=" ")
print("%.8f" % max_val)
counter += 1
else:
print(" ".join(["%.8f" % (x) for x in current_location], ),
end=" ")
print("%.8f" % max_val)
inc = False
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
import SumofGaussians as SG
import numpy as np, sys
import math
seed = int(sys.argv[1])
dims = int(sys.argv[2])
ncenters = int(sys.argv[3])
np.random.seed(seed)
sog = SG.SumofGaussians(dims, ncenters)
# Initialize the sog function with random position
x1 = np.random.ranf(dims) * 10
y1 = sog.Eval(x1)
x2 = np.empty(dims)
print(" ".join(["%.8f" % (x) for x in x1]), end=' ')
print("%.8f" % y1)
# Terminate at maximum of 100000 iterations
for i in range(100000):
random_val = (np.random.ranf(dims) * 0.1) - 0.05
for i in range(dims):
if 0 <= x1[i] + random_val[i] <= 10:
x2[i] = x1[i] + random_val[i]
else:
x2[i] = x1[i] - random_val[i]
y2 = sog.Eval(x2)
# This program usses Simulated Annealing to find the closest maximum to
# the starting location of the sum of gaussian function given generated based
# on the random seed number, number of dimensions, and number of centers (hills)
import SumofGaussians as SG
import numpy as np, sys
np.set_printoptions(formatter={'float': lambda x: "{0:0.8f}".format(x)
}) # set printing format
seed = int(sys.argv[1]) #random number seed
dims = int(sys.argv[2]) #number of dimensions
ncenters = int(sys.argv[3]) #number of hills
np.random.seed(seed) #seed the numpy random number generator
sog = SG.SumofGaussians(dims, ncenters) #create the sum of gaussian function
iterations = 1 # starting iteration
temperature = 1.0 # starting temperature
alpha = 0.9999 # alpha used for decreasing the temperature
start = 10.0 * np.random.ranf(dims) #starting point
# perform simulated annealing
current = start
# while the tempearture is above the minimum tolerance and the
# iterations are less than the maximum number of iterations
# keep searching
while iterations <= 100000:
#print the current location and the value of its point
print(
def __init__(self, dims, ncenters):
self.sog = SG.SumofGaussians(dims, ncenters)
self.dims = dims
self.loc = r.ranf(dims) * 10
self.eval = self.sog.Eval(self.loc)
self.delta = 1