def generate_population(toolbox: base.Toolbox, ccl_objects: dict, options: dict) -> List[gp.PrimitiveTree]:
"""Generate initial population"""
pop = []
codes: Set[str] = set()
pbar = tqdm.tqdm(total=options['population_size'])
while len(pop) < options['population_size']:
ind = toolbox.individual()
if not check_symbol_counts(ind, options):
continue
try:
sympy_expr = generate_sympy_expr(ind, ccl_objects)
if sympy_expr.has(sympy.zoo, sympy.oo, sympy.nan, sympy.I):
continue
sympy_code = str(sympy_expr)
except RuntimeError:
continue
if options['max_constant_allowed'] is not None and not check_max_constant(sympy_expr, options):
continue
if options['unique_population']:
if sympy_code in codes:
continue
codes.add(sympy_code)
ind.sympy_code = sympy_code
pop.append(ind)
pbar.update()
pbar.close()
return pop
# ## Evaluating Fitness
# +
def eval_knapsack(individual):
weight = 0.0
value = 0.0
for package_id in individual:
weight += packages[package_id]["weight"]
value += packages[package_id]["value"]
#if len(individual) > MAX_ITEM or weight > MAX_WEIGHT:
# return 10000, 0 # Ensure overweighted bags are dominated
return weight, value
indv = toolbox.individual()
indv_fitness = eval_knapsack(indv)
print(f"The fitness (weight, monetary value) of knapsack {indv}\nis: \n{indv_fitness}")
# -
# ## Transformations
# ### Crossover
#
# Here we use a different crossover from previous notebooks:
#
# `child1` gets that `parent1` and `parent2` overlap: `child1 = parent1 & parent2`
#