Simulated Annealing in Jupyter Notebook

Similar with particle swarm optimization, simulated annealing (SA) is a fast optimization method. It mimics the process of the cooling system in die casting. The cooling system in die casting should be slow to avoid the internal energy trapped in the cast that reduce the quality. The SA code can be found in the internet, e.g. the panda site.

SA use some exponential calculation, so we need to import “math” library in the beginning. Open your Jupyter Notebook with the proper environment. Use Anaconda Navigator to get the environment and open the terminal. After finding your working direktory, type “Jupyter Notebook” to open the IDE in your browser automatically. Copy and paste the SA code at the end of this post to show how the SA work. You can change the objective function according your case study.

Matplotlib library was used to create the performance chart. Best performance was at 8.1 after 258 iteration. Try the code below in your Jupyter Notebook or previous method, e.g. by console, IDLE, etc.

import time
import random
import math
import numpy as np
import matplotlib.pyplot as plt
# Customization section:
initial_temperature = 100
cooling = 0.8  # cooling coefficient
number_variables = 2 
upper_bounds = [3, 3]   
lower_bounds = [-3, -3]  
computing_time = 1 # second(s)

def objective_function(X):
    value = 3*(1-x)**2*math.exp(-x**2 - (y+1)**2) - 10*(x/5 - x**3 - y**5)*math.exp(-x**2 - y**2) -1/3*math.exp(-(x+1)**2 - y**2)
    return value

# Simulated Annealing Algorithm:
for v in range(number_variables):
    initial_solution[v] = random.uniform(lower_bounds[v],upper_bounds[v])

current_solution = initial_solution
best_solution = initial_solution
n = 1  # no of solutions accepted
best_fitness = objective_function(best_solution)
current_temperature = initial_temperature # current temperature
start = time.time()
no_attempts = 100 # number of attempts in each level of temperature
record_best_fitness =[]

for i in range(9999999):
    for j in range(no_attempts):

        for k in range(number_variables):
            current_solution[k] = best_solution[k] + 0.1*(random.uniform(lower_bounds[k],upper_bounds[k]))
            current_solution[k] = max(min(current_solution[k],upper_bounds[k]),lower_bounds[k])  # repair the solution respecting the bounds

        current_fitness = objective_function(current_solution)
        E = abs(current_fitness - best_fitness)
        if i == 0 and j == 0:
            EA = E

        if current_fitness < best_fitness:
            p = math.exp(-E/(EA*current_temperature))
            # make a decision to accept the worse solution or not
            if random.random()<p:
                accept = True # this worse solution is accepted
                accept = False # this worse solution is not accepted
            accept = True # accept better solution
        if accept==True:
            best_solution = current_solution # update the best solution
            best_fitness = objective_function(best_solution)
            n = n + 1 # count the solutions accepted
            EA = (EA *(n-1) + E)/n # update EA

    print('interation: {}, best_solution: {}, best_fitness: {}'.format(i, best_solution, best_fitness))
    # Cooling the temperature
    current_temperature = current_temperature*cooling
    # Stop by computing time
    end = time.time()
    if end-start >= computing_time:

Tinggalkan Balasan

Isikan data di bawah atau klik salah satu ikon untuk log in:


You are commenting using your account. Logout /  Ubah )

Foto Google

You are commenting using your Google account. Logout /  Ubah )

Gambar Twitter

You are commenting using your Twitter account. Logout /  Ubah )

Foto Facebook

You are commenting using your Facebook account. Logout /  Ubah )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.