def franken30(n, my_func, lb, ub, dimension, iteration):
X = np.array(
[src.solution(my_func, dimension, (lb, ub)) for i in range(n)])
[Xi.initRandom() for Xi in X]
for it in range(iteration):
X1 = src.op.op_pso(X, src.op.select_random, src.op.mut_pso,
src.op.crx_exponential)
X = src.op.replace_if_best(X, X1)
X = src.op.drop_probability(X)
return X
def ge():
X = np.array([src.solution(my_func, dimension, bounds) for i in range(n)])
[Xi.initRandom() for Xi in X]
src.solution.updateHistory(X)
for it in range(iteration):
X1 = src.op.op_pso(X, src.op.select_random, src.op.mut_pso,
src.op.crx_exponential)
X = src.op.replace_if_best(X, X1)
X = src.op.drop_probability(X)
src.solution.updateHistory(X)
return X
def new_franken10(n, my_func, lb, ub, dimension, iteration):
X = np.array(
[src.solution(my_func, dimension, (lb, ub)) for i in range(n)])
[Xi.initRandom() for Xi in X]
for it in range(iteration):
X1 = src.op.op_de(X, src.op.select_random, src.op.mut_pso,
src.op.crx_blend)
X = src.op.replace_if_best(X, X1)
X1 = src.op.op_pso(X, src.op.select_random, src.op.mut_pso,
src.op.crx_blend)
X = src.op.replace_if_best(X, X1)
return X
def pso(n, my_func, lb, ub, dimension, iteration):
#instantiate solutions
X = np.array(
[src.solution(my_func, dimension, (lb, ub)) for i in range(n)])
#initialize solutions
[Xi.initRandom() for Xi in X]
for it in range(iteration):
#1. Select individuals for modification in this round
# none - select all. Alternative (bee algorythm) is to select only solutions drawn with fitness-dependant probability
#2. de_operator = create an alternative set of solutions X1 using mutation+crossover
X1 = src.op.op_pso(X, src.op.select_random, src.op.mut_pso,
src.op.crx_exponential)
#3. Select individual for the next generation <- accept all
X = X1
return X
def pso():
#instantiate solutions
X = np.array([src.solution(my_func, dimension, bounds) for i in range(n)])
#initialize solutions
[Xi.initRandom() for Xi in X]
#just so we can have some animations
src.solution.updateHistory(X) # it is not necessary for the grammar
for it in range(iteration):
#1. Select individuals for modification in this round
# none - select all. Alternative (bee algorythm) is to select only solutions drawn with fitness-dependant probability
#2. de_operator = create an alternative set of solutions X1 using mutation+crossover
X1 = src.op.op_pso(X, src.op.select_random, src.op.mut_pso,
src.op.crx_exponential)
#3. Select individual for the next generation <- accept all
X = X1
src.solution.updateHistory(X)
return X
def test_case_one_element():
assert solution([0]) == []
def test_case_empth():
assert solution([]) == []
def test_case_sample():
assert solution([9, 1, 4, 9, 0, 4, 8, 9, 0,
1]) == [1, 3, 2, 3, 0, 0, 0, 0, 0]
