[Python] handle equality constraints in a Differential Evolution

I'm implementing a Differential Evolution (DE) algorithm to solve an optimization problem that includes both inequality and equality constraints.

import numpy as np
import funciones

# Differential Evolution algorithm parameters
NP = 50 # Population size
F = 0.3 # Mutation factor
CR = 0.7 # Crossover rate
GEN = 500 # Number of generations

# Function to check if a solution is feasible
def is_feasible(x, problem_id=4):
return all(funciones.restrictions(x, problem_id) <= 1)

# Population initialization
def initialize_population(D, lower_limits, upper_limits):
.
.
.
return population



# Mutation of the population
def mutation(population, lower_limits, upper_limits):
.
.
.
return apply_limits(mutated_vectors, lower_limits, upper_limits)

# Crossover of the population
def crossover(population, mutated_vectors):
.
.
.
return trial

# Selection with constraint handling
def selection(population, trial, problem_id=4):
.
.
.
return new_population

# Main function to run the Differential Evolution algorithm
def run_de(problem_id=4):
np.random.seed(15)
D, lower_limits, upper_limits = funciones.get_problem_dimensions(problem_id)
population = initialize_population(D, lower_limits, upper_limits)
for gen in range(GEN):
mutated = mutation(population, lower_limits, upper_limits)
crossed = crossover(population, mutated)
population = selection(population, crossed, problem_id)
feasible_solutions = [population for i in range(NP) if is_feasible(population, problem_id)]
if feasible_solutions:
values = [funciones.objective_function(x, problem_id) for x in feasible_solutions]
best_index = np.argmin(values)
best_solution = feasible_solutions[best_index]
return best_solution, funciones.objective_function(best_solution, problem_id)
else:
return None, None

best_solution, best_value = run_de(problem_id=4)

if best_solution is not None:
print("Best solution found:", best_solution)
print("Objective value:", best_value)
else:
print("No feasible solutions found")



My problem is handling the equalities. I tried adding them like this, but when i try the code, it give me "No feasible solutions found":

elif problema_id == 4:

g = np.zeros(5)
g[0] = -x[3] + x[2] - 0.55
g[1] = -x[2] + x[3] - 0.55
h3 = 1000 * np.sin(-x[2] - 0.25) + 1000 * np.sin(-x[3] - 0.25) + 894.8 - x[0]
g[2] = np.abs(h3)
h4 = 1000 * np.sin(x[2] - 0.25) + 1000 * np.sin(x[2] - x[3] - 0.25) + 894.8 - x[1]
g[3] = np.abs(h4)
h5 = 1000 * np.sin(x[3] - 0.25) + 1000 * np.sin(x[3] - x[2] - 0.25) + 1294.8
g[4] = np.abs(h5)

return g


However, I'm not sure if I'm handling these equalities correctly, since even when I change the parameters, I always get the error

Continue reading...