project setup

.gitignore

@@ -1,11 +1,17 @@
 # Files
+.coverage
 
 # Folders
+.vscode/
 __pycache__/
-docs/build/
 
-.vscode/
+dist/
+*.egg-info/
+docs/build/
 
 outputs/
 data/
 tmp/
+
+# Version file autogenerated by setuptools_scm
+vopy/version.py

environment.yml

@@ -5,13 +5,11 @@ channels:
 - gpytorch
 dependencies:
 - python>=3.10
-- gpytorch
 - botorch>=0.10.0,<0.12.0
 - cvxpy
 - scikit-learn
 - scipy
 - matplotlib
-- numpy
 - jupyter
 - pytest
 - pytest-cov
@@ -20,6 +18,5 @@ dependencies:
 - pydocstyle
 - black
 - bandit
-- sphinx_rtd_theme
 - pip:
     - -r docs/requirements.txt

pyproject.toml

@@ -1,3 +1,39 @@
+[project]
+name = "VOPy"
+# version = "0.0.1"
+dynamic = ["version"]
+authors = [
+  { name="Y. Cahit Yıldırım", email="[email protected]" },
+]
+description = "A framework for black-box vector optimization"
+readme = "README.md"
+license = {file="LICENSE"}
+requires-python = ">=3.10"
+dependencies = [
+    "botorch>=0.10.0,<0.12.0",
+    "cvxpy",
+    "scikit-learn",
+    "scipy",
+    "matplotlib",
+]
+
+[project.urls]
+Homepage = "https://github.com/Bilkent-CYBORG/VOPy"
+Documentation = "https://vopy.readthedocs.io/en/latest/"
+Issues = "https://github.com/Bilkent-CYBORG/VOPy/issues"
+
+[build-system]
+requires = ["setuptools>=64", "setuptools-scm>=8"]
+build-backend = "setuptools.build_meta"
+
+[tool.setuptools]
+packages = ["vopy"]
+
+[tool.setuptools_scm]
+# Should be present even if its empty.
+local_scheme = "node-and-date"
+write_to = "./vopy/version.py"
+
 [tool.pytest.ini_options]
 filterwarnings = [
     "ignore::UserWarning",

requirements.txt

@@ -1,4 +1,3 @@
-gpytorch
 botorch>=0.10.0,<0.12.0
 cvxpy
 scikit-learn
@@ -13,6 +12,5 @@ flake8-print
 pydocstyle
 black
 bandit
-sphinx_rtd_theme
 
 -r docs/requirements.txt

test/models/test_gpytorch.py

@@ -17,6 +17,8 @@
     get_gpytorch_model_w_known_hyperparams,
 )
 
+from vopy.utils import set_seed
+from vopy.utils.seed import SEED
 from vopy.maximization_problem import Problem
 
 
@@ -25,6 +27,7 @@ class TestMultitaskExactGPModel(unittest.TestCase):
 
     def setUp(self):
         """Set up sample data and model for tests."""
+        set_seed(SEED)
         self.X = torch.randn(10, 2)
         self.Y = torch.randn(10, 2)
         self.likelihood = MultitaskGaussianLikelihood(num_tasks=self.Y.shape[-1])
@@ -41,6 +44,7 @@ class TestBatchIndependentExactGPModel(unittest.TestCase):
 
     def setUp(self):
         """Set up sample data and model for tests."""
+        set_seed(SEED)
         self.X = torch.randn(10, 2)
         self.Y = torch.randn(10, 2)
         self.likelihood = GaussianLikelihood()
@@ -70,6 +74,7 @@ class TestGPyTorchMultioutputExactModel(unittest.TestCase):
     def setUp(self):
         """Set up model instance with test data."""
         # Use the testable subclass to avoid abstract class issues
+        set_seed(SEED)
         self.model = TestableGPyTorchMultioutputExactModel(
             input_dim=2, output_dim=2, noise_var=0.1, model_kind=MultitaskExactGPModel
         )
@@ -113,6 +118,7 @@ class TestCorrelatedExactGPyTorchModel(unittest.TestCase):
 
     def setUp(self):
         """Set up correlated multitask model."""
+        set_seed(SEED)
         self.model = CorrelatedExactGPyTorchModel(input_dim=2, output_dim=2, noise_var=0.1)
         self.X = torch.randn(5, 2)
         self.Y = torch.randn(5, 2)
@@ -122,14 +128,20 @@ def setUp(self):
     def test_predict(self):
         """Test predict method."""
         means, variances = self.model.predict(self.X)
-        self.assertGreaterEqual(np.min(variances), 0, "Negative variance.")
+        for i in range(len(self.X)):
+            with self.subTest(i=i):
+                try:
+                    np.linalg.cholesky(variances[i].reshape(self.Y.shape[1], self.Y.shape[1]))
+                except np.linalg.LinAlgError:
+                    self.fail("Covariance matrix is not positive definite.")
 
 
 class TestIndependentExactGPyTorchModel(unittest.TestCase):
     """Tests for the IndependentExactGPyTorchModel class."""
 
     def setUp(self):
         """Set up independent multitask model."""
+        set_seed(SEED)
         self.model = IndependentExactGPyTorchModel(input_dim=2, output_dim=2, noise_var=0.1)
         self.X = torch.randn(5, 2)
         self.Y = torch.randn(5, 2)
@@ -147,6 +159,8 @@ class TestGetGPyTorchModelWithKnownHyperparams(unittest.TestCase):
 
     def test_model_with_given_data(self):
         """Test if the model is created correctly when X and Y data are provided."""
+        set_seed(SEED)
+
         self.input_dim = 2
         self.output_dim = 2
         self.noise_var = 0.1
@@ -187,6 +201,7 @@ class TestSingleTaskGP(unittest.TestCase):
 
     def setUp(self):
         """Set up single-task GP model."""
+        set_seed(SEED)
         self.X = torch.randn(10, 2)
         self.Y = torch.randn(10)
         self.likelihood = GaussianLikelihood()
@@ -203,6 +218,7 @@ class TestGPyTorchModelListExactModel(unittest.TestCase):
 
     def setUp(self):
         """Set up multi-output GP model list."""
+        set_seed(SEED)
         self.input_dim = 2
         self.output_dim = 2
         self.model = GPyTorchModelListExactModel(
@@ -246,6 +262,7 @@ class TestGetGPyTorchModelListWithKnownHyperparams(unittest.TestCase):
 
     def setUp(self):
         """Set up problem instance with test data."""
+        set_seed(SEED)
         self.input_dim = 2
         self.output_dim = 2
         self.noise_var = 0.1