[BUG] AGEMOEA2 division by zero error #751
Description
Checklist
- I searched existing issues
- I'm using the latest pymoo version
Bug Description
AGEMOEA 2 division by zero error when using below example code. Im using pymoo with commit id: 6e1cb88.
Minimal Code to Reproduce
class ContentStrategyProblem(ElementwiseProblem):
def __init__(self):
super().__init__(n_var=4, n_obj=2, xl=np.array([0.01, 0.01, 0.01, 0.0]), xu=np.array([1.0, 1.0, 1.0, 1.0]))
def _evaluate(self, x, out, *args, **kwargs):
source_influence, amplifier_effect, banter_strength, has_fst_intervention = x
# Objective 1: Maximize engagement (we'll use a simplified formula)
base_engagement = 1.5 if has_fst_intervention else 1
engagement_score = (source_influence * amplifier_effect * banter_strength) * base_engagement
# Objective 2: Minimize negative sentiment (hypothetical)
negative_sentiment = 0.5 * has_fst_intervention + 0.1 * (1 - source_influence)
out["F"] = [-engagement_score, negative_sentiment]
runner = RayParallelization(job_resources={})
optimization_problem = ContentStrategyProblem(
# elementwise_runner=runner
)
agemoea2_algorithm = AGEMOEA2()
termination = get_termination("time", "00:01:00")
result = minimize(
problem=optimization_problem,
algorithm=agemoea2_algorithm,
verbose=True,
seed=1
)Error Message
==========================================================
n_gen | n_eval | n_nds | eps | indicator
==========================================================
1 | 100 | 6 | - | -
2 | 200 | 11 | 0.0577279929 | ideal
3 | 300 | 6 | 0.1778856766 | ideal
4 | 400 | 8 | 0.0863211991 | ideal
5 | 500 | 9 | 0.0139915214 | ideal
6 | 600 | 4 | 0.0121387444 | ideal
7 | 700 | 3 | 0.2310262917 | ideal
8 | 800 | 2 | 0.1939060943 | ideal
9 | 900 | 5 | 0.3272611788 | ideal
10 | 1000 | 2 | 0.4087270492 | ideal
11 | 1100 | 6 | 0.1537786751 | ideal
12 | 1200 | 5 | 0.2311699925 | ideal
13 | 1300 | 8 | 0.0133638859 | ideal
14 | 1400 | 6 | 0.0330066173 | ideal
15 | 1500 | 10 | 0.5777152290 | nadir
16 | 1600 | 5 | 0.0441168778 | ideal
17 | 1700 | 7 | 0.0507478193 | f
18 | 1800 | 8 | 0.0034690261 | nadir
19 | 1900 | 7 | 0.0337110676 | ideal
20 | 2000 | 10 | 0.0058099718 | ideal
21 | 2100 | 8 | 0.0061289139 | ideal
22 | 2200 | 8 | 0.0098262262 | f
23 | 2300 | 9 | 0.0088472570 | ideal
24 | 2400 | 8 | 0.1296656299 | ideal
25 | 2500 | 2 | 0.1566509640 | ideal
26 | 2600 | 3 | 0.0279537565 | ideal
27 | 2700 | 4 | 0.0043663282 | ideal
28 | 2800 | 7 | 0.0417304699 | ideal
29 | 2900 | 8 | 0.0209455696 | ideal
30 | 3000 | 10 | 0.0219271875 | ideal
31 | 3100 | 9 | 0.0142879662 | ideal
32 | 3200 | 10 | 0.4389988711 | nadir
33 | 3300 | 6 | 0.0050860066 | ideal
34 | 3400 | 7 | 0.0395848263 | f
35 | 3500 | 7 | 0.9999775983 | nadir
36 | 3600 | 6 | 5.250872E+02 | nadir
37 | 3700 | 7 | 2.714909E+02 | nadir
38 | 3800 | 5 | 0.0026890043 | ideal
39 | 3900 | 8 | 0.9999724392 | nadir
40 | 4000 | 7 | 3.955529E+01 | nadir
41 | 4100 | 7 | 0.0028461624 | ideal
42 | 4200 | 9 | 0.0267046646 | ideal
43 | 4300 | 11 | 1.799695E+01 | nadir
44 | 4400 | 4 | 5.416172E+01 | nadir
45 | 4500 | 4 | 0.1193224803 | ideal
46 | 4600 | 8 | 0.0044991349 | ideal
47 | 4700 | 10 | 0.8980986769 | nadir
48 | 4800 | 6 | 0.0278943971 | nadir
49 | 4900 | 6 | 0.0430304756 | ideal
50 | 5000 | 7 | 0.0069831521 | ideal
51 | 5100 | 3 | 0.0136774367 | ideal
52 | 5200 | 5 | 0.0186564096 | ideal
53 | 5300 | 7 | 0.0293342425 | ideal
54 | 5400 | 8 | 0.0047865591 | ideal
55 | 5500 | 8 | 4.222404E+01 | nadir
56 | 5600 | 8 | 0.9401523120 | nadir
57 | 5700 | 11 | 0.0361796481 | ideal
58 | 5800 | 12 | 0.0029586325 | ideal
59 | 5900 | 6 | 0.0025969506 | ideal
60 | 6000 | 10 | 0.0032709755 | ideal
61 | 6100 | 8 | 1.9331520083 | nadir
62 | 6200 | 8 | 4.981988E+02 | nadir
63 | 6300 | 9 | 0.9999017070 | nadir
64 | 6400 | 2 | 0.2041693016 | ideal
65 | 6500 | 4 | 0.3043771353 | f
66 | 6600 | 3 | 0.2624973961 | f
67 | 6700 | 5 | 0.9999999998 | nadir
68 | 6800 | 6 | 5.9538163183 | nadir
69 | 6900 | 7 | 7.8606280362 | nadir
70 | 7000 | 10 | 0.0281788241 | nadir
71 | 7100 | 9 | 0.0031317177 | f
72 | 7200 | 8 | 0.8343467910 | nadir
73 | 7300 | 6 | 1.158527E+01 | nadir
---------------------------------------------------------------------------
ZeroDivisionError Traceback (most recent call last)
/tmp/ipykernel_34500/3814179118.py in <cell line: 0>()
23
24 termination = get_termination("time", "00:01:00")
---> 25 result = minimize(
26 problem=optimization_problem,
27 algorithm=agemoea2_algorithm,
/usr/local/lib/python3.11/dist-packages/pymoo/optimize.py in minimize(problem, algorithm, termination, copy_algorithm, copy_termination, **kwargs)
65
66 # actually execute the algorithm
---> 67 res = algorithm.run()
68
69 # store the deep copied algorithm in the result object
/usr/local/lib/python3.11/dist-packages/pymoo/core/algorithm.py in run(self)
136 def run(self):
137 while self.has_next():
--> 138 self.next()
139 return self.result()
140
/usr/local/lib/python3.11/dist-packages/pymoo/core/algorithm.py in next(self)
157 if infills is not None:
158 self.evaluator.eval(self.problem, infills, algorithm=self)
--> 159 self.advance(infills=infills)
160
161 # if the algorithm does not follow the infill-advance scheme just call advance
/usr/local/lib/python3.11/dist-packages/pymoo/core/algorithm.py in advance(self, infills, **kwargs)
223
224 # call the implementation of the advance method - if the infill is not None
--> 225 val = self._advance(infills=infills, **kwargs)
226
227 # always advance to the next iteration - except if the algorithm returns False
/usr/local/lib/python3.11/dist-packages/pymoo/algorithms/base/genetic.py in _advance(self, infills, **kwargs)
107
108 # execute the survival to find the fittest solutions
--> 109 self.pop = self.survival.do(self.problem, pop, n_survive=self.pop_size, algorithm=self, **kwargs)
/usr/local/lib/python3.11/dist-packages/pymoo/core/survival.py in do(self, problem, pop, n_survive, return_indices, *args, **kwargs)
53
54 else:
---> 55 survivors = self._do(problem, pop, *args, n_survive=n_survive, **kwargs)
56
57 if return_indices:
/usr/local/lib/python3.11/dist-packages/pymoo/algorithms/moo/age.py in _do(self, problem, pop, n_survive, *args, **kwargs)
126
127 # Calculate the crowding distance of the first front as well as p and the normalization constants
--> 128 crowd_dist[front_no == 0], p, normalization = self.survival_score(front1, ideal_point)
129 for i in range(1, max_f_no): # skip first front since it is normalized by survival_score
130 front = F[front_no == i, :]
/usr/local/lib/python3.11/dist-packages/pymoo/algorithms/moo/age.py in survival_score(self, front, ideal_point)
171 nn[nn < 1e-8] = 1
172
--> 173 distances = self.pairwise_distances(front, p)
174 distances[distances < 1e-8] = 1e-8 # Replace very small entries to prevent division by zero
175
ZeroDivisionError: division by zeroPyMoo & Python Version
pymoo @ git+https://github.com/anyoptimization/pymoo@main & python3.11