28718428290

Committed 04 Jul 2026 08:17PM UTC coverage: 86.103% (-0.9%) from 87.05%

Build # 28718428290

Build Type

push

github

Committed by

web-flow

Commit Message

Feat/mo acquisition functions (#15)

Coverage Stats

270 of 363 new or added lines in 5 files covered. (74.38%)

3922 of 4555 relevant lines covered (86.1%)

1.72 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

97.67

/src/tests/test_unit/test_multi_objective.py

import unittest
import numpy as np


from smt_optim.acquisition_functions.multi_obj import init_mpi


class MockModel:
    def __init__(self, val, var):
        self.val = val
        self.var = var

    def predict_values(self, x):
        return np.array([[self.val]])

    def predict_variances(self, x):
        return np.array([[self.var]])


class MockProblem:
    def __init__(self):
        self.num_obj = 2


class MockDataset:
    def __init__(self):
        self.points = np.array([[0.5, 0.5]])

    def export_as_dict(self):
        return {
            "obj": np.array([[0.2, 0.2], [0.8, 0.8]]),
            "rscv": np.array([0.0, 0.0]),
            "fidelity": np.array([0, 0]),
            "x": np.array([[0.1], [0.9]]),
        }


class MockState:
    def __init__(self):
        self.obj_models = [MockModel(0.1, 0.0), MockModel(0.1, 0.0)]
        self.problem = MockProblem()
        self.scaled_dataset = MockDataset()


class TestMultiObjectiveFixes(unittest.TestCase):
    def test_mpi_zero_variance(self):
        state = MockState()
        # Initialize MPI
        mpi_func = init_mpi(state)
        # Point to predict
        x_test = np.array([[0.5]])
        # Since variances are 0.0, MPI should return 0.0 to prevent infinite loop
        val = mpi_func(x_test)
        self.assertEqual(val, 0.0)

    def test_ei_cf_constraint_filtering(self):
        from smt_optim.acquisition_functions.multi_obj import init_bi_obj_ei_cf

        state = MockState()
        # Create a mock dataset with one feasible and one infeasible point
        # [0.2, 0.2] has rscv = 1.0 (infeasible)
        # [0.8, 0.8] has rscv = 0.0 (feasible)
        state.scaled_dataset.export_as_dict = lambda: {
            "obj": np.array([[0.2, 0.2], [0.8, 0.8]]),
            "rscv": np.array([1.0, 0.0]),
            "fidelity": np.array([0, 0]),
            "x": np.array([[0.1], [0.9]]),
        }

        # phi is just sum of objectives
        def mock_phi(y):
            return np.sum(y, axis=-1)

        kwargs = {"phi": mock_phi, "n_accuracy": 10}

        # init_bi_obj_ei_cf should find f_min = mock_phi([0.8, 0.8]) = 1.6
        # rather than mock_phi([0.2, 0.2]) = 0.4

        # Since I can't easily extract f_min from the closure without calling it,
        # I'll call it with a point. If f_min = 1.6, and predictions are 0.1, 0.1
        # Then phi(y) = 0.2. max(f_min - phi(y), 0) = max(1.6 - 0.2, 0) = 1.4
        # If f_min was 0.4, max(0.4 - 0.2, 0) = 0.2
        ei_cf_func = init_bi_obj_ei_cf(state, kwargs)

        val = ei_cf_func(np.array([[0.5]]))
        self.assertAlmostEqual(val[0][0], 1.4)


if __name__ == "__main__":
    unittest.main()

1	import unittest	2✔
2	import numpy as np	2✔
3
4
5	from smt_optim.acquisition_functions.multi_obj import init_mpi	2✔
6
7
8	class MockModel:	2✔
9	def __init__(self, val, var):	2✔
10	self.val = val	2✔
11	self.var = var	2✔
12
13	def predict_values(self, x):	2✔
14	return np.array([[self.val]])	2✔
15
16	def predict_variances(self, x):	2✔
17	return np.array([[self.var]])	2✔
18
19
20	class MockProblem:	2✔
21	def __init__(self):	2✔
22	self.num_obj = 2	2✔
23
24
25	class MockDataset:	2✔
26	def __init__(self):	2✔
27	self.points = np.array([[0.5, 0.5]])	2✔
28
29	def export_as_dict(self):	2✔
30	return {	2✔
31	"obj": np.array([[0.2, 0.2], [0.8, 0.8]]),
32	"rscv": np.array([0.0, 0.0]),
33	"fidelity": np.array([0, 0]),
34	"x": np.array([[0.1], [0.9]]),
35	}
36
37
38	class MockState:	2✔
39	def __init__(self):	2✔
40	self.obj_models = [MockModel(0.1, 0.0), MockModel(0.1, 0.0)]	2✔
41	self.problem = MockProblem()	2✔
42	self.scaled_dataset = MockDataset()	2✔
43
44
45	class TestMultiObjectiveFixes(unittest.TestCase):	2✔
46	def test_mpi_zero_variance(self):	2✔
47	state = MockState()	2✔
48	# Initialize MPI
49	mpi_func = init_mpi(state)	2✔
50	# Point to predict
51	x_test = np.array([[0.5]])	2✔
52	# Since variances are 0.0, MPI should return 0.0 to prevent infinite loop
53	val = mpi_func(x_test)	2✔
54	self.assertEqual(val, 0.0)	2✔
55
56	def test_ei_cf_constraint_filtering(self):	2✔
57	from smt_optim.acquisition_functions.multi_obj import init_bi_obj_ei_cf	2✔
58
59	state = MockState()	2✔
60	# Create a mock dataset with one feasible and one infeasible point
61	# [0.2, 0.2] has rscv = 1.0 (infeasible)
62	# [0.8, 0.8] has rscv = 0.0 (feasible)
63	state.scaled_dataset.export_as_dict = lambda: {	2✔
64	"obj": np.array([[0.2, 0.2], [0.8, 0.8]]),
65	"rscv": np.array([1.0, 0.0]),
66	"fidelity": np.array([0, 0]),
67	"x": np.array([[0.1], [0.9]]),
68	}
69
70	# phi is just sum of objectives
71	def mock_phi(y):	2✔
72	return np.sum(y, axis=-1)	2✔
73
74	kwargs = {"phi": mock_phi, "n_accuracy": 10}	2✔
75
76	# init_bi_obj_ei_cf should find f_min = mock_phi([0.8, 0.8]) = 1.6
77	# rather than mock_phi([0.2, 0.2]) = 0.4
78
79	# Since I can't easily extract f_min from the closure without calling it,
80	# I'll call it with a point. If f_min = 1.6, and predictions are 0.1, 0.1
81	# Then phi(y) = 0.2. max(f_min - phi(y), 0) = max(1.6 - 0.2, 0) = 1.4
82	# If f_min was 0.4, max(0.4 - 0.2, 0) = 0.2
83	ei_cf_func = init_bi_obj_ei_cf(state, kwargs)	2✔
84
85	val = ei_cf_func(np.array([[0.5]]))	2✔
86	self.assertAlmostEqual(val[0][0], 1.4)	2✔
87
88
89	if __name__ == "__main__":	2✔
NEW 90	unittest.main()	×

SMTorg / smt-optim / 28718428290

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous