Make xy_sample receive features, not instances

miplearn/components/component.py

@@ -3,11 +3,11 @@
 #  Released under the modified BSD license. See COPYING.md for more details.
 
 import numpy as np
-from typing import Any, List, Union, TYPE_CHECKING, Tuple, Dict
+from typing import Any, List, Union, TYPE_CHECKING, Tuple, Dict, Optional
 
 from miplearn.extractors import InstanceIterator
 from miplearn.instance import Instance
-from miplearn.types import LearningSolveStats, TrainingSample
+from miplearn.types import LearningSolveStats, TrainingSample, Features
 
 if TYPE_CHECKING:
     from miplearn.solvers.learning import LearningSolver
@@ -133,14 +133,16 @@ class Component:
 
     @staticmethod
     def xy_sample(
-        instance: Any,
-        training_sample: TrainingSample,
-    ) -> Tuple[Dict, Dict]:
+        features: Features,
+        sample: TrainingSample,
+    ) -> Optional[Tuple[Dict, Dict]]:
         """
-        Given a training sample, returns a pair of x and y dictionaries containing,
-        respectively, the matrices of ML features and the labels for the sample.
+        Given a set of features and a training sample, returns a pair of x and y
+        dictionaries containing, respectively, the matrices of ML features and the
+        labels for the sample. If the training sample does not include label
+        information, returns None.
         """
-        return {}, {}
+        return None
 
     def xy_instances(
         self,
@@ -149,8 +151,12 @@ class Component:
         x_combined: Dict = {}
         y_combined: Dict = {}
         for instance in InstanceIterator(instances):
+            assert isinstance(instance, Instance)
             for sample in instance.training_data:
-                x_sample, y_sample = self.xy_sample(instance, sample)
+                xy = self.xy_sample(instance.features, sample)
+                if xy is None:
+                    continue
+                x_sample, y_sample = xy
                 for cat in x_sample.keys():
                     if cat not in x_combined:
                         x_combined[cat] = []

miplearn/components/lazy_static.py

@@ -5,14 +5,14 @@
 import logging
 import sys
 from copy import deepcopy
-from typing import Any, Dict, Tuple
+from typing import Any, Dict, Tuple, Optional
 
 import numpy as np
 from tqdm.auto import tqdm
 
 from miplearn.classifiers.counting import CountingClassifier
 from miplearn.components.component import Component
-from miplearn.types import TrainingSample
+from miplearn.types import TrainingSample, Features
 
 logger = logging.getLogger(__name__)
 
@@ -207,15 +207,16 @@ class StaticLazyConstraintsComponent(Component):
 
     @staticmethod
     def xy_sample(
-        instance: Any,
+        features: Features,
         sample: TrainingSample,
-    ) -> Tuple[Dict, Dict]:
+    ) -> Optional[Tuple[Dict, Dict]]:
+        if "LazyStatic: Enforced" not in sample:
+            return None
         x: Dict = {}
         y: Dict = {}
-        if "LazyStatic: All" not in sample:
-            return x, y
-        for cid in sorted(sample["LazyStatic: All"]):
-            cfeatures = instance.features["Constraints"][cid]
+        for (cid, cfeatures) in features["Constraints"].items():
+            if not cfeatures["Lazy"]:
+                continue
             category = cfeatures["Category"]
             if category is None:
                 continue

miplearn/components/objective.py

@@ -19,7 +19,7 @@ from miplearn.classifiers import Regressor
 from miplearn.components.component import Component
 from miplearn.extractors import InstanceIterator
 from miplearn.instance import Instance
-from miplearn.types import MIPSolveStats, TrainingSample, LearningSolveStats
+from miplearn.types import MIPSolveStats, TrainingSample, LearningSolveStats, Features
 
 if TYPE_CHECKING:
     from miplearn.solvers.learning import LearningSolver
@@ -164,18 +164,20 @@ class ObjectiveValueComponent(Component):
 
     @staticmethod
     def xy_sample(
-        instance: Any,
+        features: Features,
         sample: TrainingSample,
-    ) -> Tuple[Dict, Dict]:
-        x: Dict = {}
-        y: Dict = {}
+    ) -> Optional[Tuple[Dict, Dict]]:
         if "Lower bound" not in sample:
-            return x, y
-        features = instance.features["Instance"]["User features"]
+            return None
+        f = features["Instance"]["User features"]
         if "LP value" in sample and sample["LP value"] is not None:
-            features += [sample["LP value"]]
-        x["Lower bound"] = [features]
-        x["Upper bound"] = [features]
-        y["Lower bound"] = [[sample["Lower bound"]]]
-        y["Upper bound"] = [[sample["Upper bound"]]]
+            f += [sample["LP value"]]
+        x = {
+            "Lower bound": [f],
+            "Upper bound": [f],
+        }
+        y = {
+            "Lower bound": [[sample["Lower bound"]]],
+            "Upper bound": [[sample["Upper bound"]]],
+        }
         return x, y

miplearn/components/primal.py

@@ -211,15 +211,15 @@ class PrimalSolutionComponent(Component):
 
     @staticmethod
     def xy_sample(
-        instance: Any,
+        features: Features,
         sample: TrainingSample,
-    ) -> Tuple[Dict, Dict]:
+    ) -> Optional[Tuple[Dict, Dict]]:
         if "Solution" not in sample:
-            return {}, {}
+            return None
         assert sample["Solution"] is not None
         return cast(
             Tuple[Dict, Dict],
-            PrimalSolutionComponent._extract(instance.features, sample),
+            PrimalSolutionComponent._extract(features, sample),
         )
 
     @staticmethod
@@ -227,7 +227,10 @@ class PrimalSolutionComponent(Component):
         features: Features,
         sample: TrainingSample,
     ) -> Dict:
-        return cast(Dict, PrimalSolutionComponent._extract(features, sample))
+        return cast(
+            Dict,
+            PrimalSolutionComponent._extract(features, sample),
+        )
 
     @staticmethod
     def _extract(

miplearn/features.py

@@ -58,6 +58,7 @@ class FeaturesExtractor:
         self,
         instance: "Instance",
     ) -> Dict[str, ConstraintFeatures]:
+        has_static_lazy = instance.has_static_lazy_constraints()
         constraints: Dict[str, ConstraintFeatures] = {}
         for cid in self.solver.get_constraint_ids():
             user_features = None
@@ -83,6 +84,10 @@ class FeaturesExtractor:
                 "Category": category,
                 "User features": user_features,
             }
+            if has_static_lazy:
+                constraints[cid]["Lazy"] = instance.is_constraint_lazy(cid)
+            else:
+                constraints[cid]["Lazy"] = False
         return constraints
 
     @staticmethod

miplearn/instance.py

@@ -101,13 +101,13 @@ class Instance(ABC):
     def get_constraint_category(self, cid: str) -> Optional[str]:
         return cid
 
-    def has_static_lazy_constraints(self):
+    def has_static_lazy_constraints(self) -> bool:
         return False
 
     def has_dynamic_lazy_constraints(self):
         return False
 
-    def is_constraint_lazy(self, cid):
+    def is_constraint_lazy(self, cid: str) -> bool:
         return False
 
     def find_violated_lazy_constraints(self, model):

miplearn/types.py

@@ -98,6 +98,7 @@ ConstraintFeatures = TypedDict(
         "Sense": str,
         "Category": Optional[Hashable],
         "User features": Optional[List[float]],
+        "Lazy": bool,
     },
     total=False,
 )

tests/components/test_component.py

@@ -7,8 +7,7 @@ from miplearn import Component, Instance
 
 
 def test_xy_instance():
-    def _xy_sample(instance, sample):
-        print(sample)
+    def _xy_sample(features, sample):
         x = {
             "s1": {
                 "category_a": [
@@ -54,8 +53,10 @@ def test_xy_instance():
     comp = Component()
     instance_1 = Mock(spec=Instance)
     instance_1.training_data = ["s1", "s2"]
+    instance_1.features = {}
     instance_2 = Mock(spec=Instance)
     instance_2.training_data = ["s3"]
+    instance_2.features = {}
     comp.xy_sample = _xy_sample
     x_expected = {
         "category_a": [

tests/components/test_lazy_static.py

@@ -9,7 +9,7 @@ from miplearn.components.lazy_static import StaticLazyConstraintsComponent
 from miplearn.instance import Instance
 from miplearn.solvers.internal import InternalSolver
 from miplearn.solvers.learning import LearningSolver
-from miplearn.types import TrainingSample
+from miplearn.types import TrainingSample, Features
 
 
 def test_usage_with_solver():
@@ -234,32 +234,35 @@ def test_fit():
 
 
 def test_xy_sample() -> None:
-    instance = Mock(spec=Instance)
     sample: TrainingSample = {
-        "LazyStatic: Enforced": {"c1", "c2", "c4", "c5"},
-        "LazyStatic: All": {"c1", "c2", "c3", "c4", "c5"},
+        "LazyStatic: Enforced": {"c1", "c2", "c4"},
     }
-    instance.features = {
+    features: Features = {
         "Constraints": {
             "c1": {
                 "Category": "type-a",
                 "User features": [1.0, 1.0],
+                "Lazy": True,
             },
             "c2": {
                 "Category": "type-a",
                 "User features": [1.0, 2.0],
+                "Lazy": True,
             },
             "c3": {
                 "Category": "type-a",
                 "User features": [1.0, 3.0],
+                "Lazy": True,
             },
             "c4": {
                 "Category": "type-b",
                 "User features": [1.0, 4.0, 0.0],
+                "Lazy": True,
             },
             "c5": {
                 "Category": "type-b",
                 "User features": [1.0, 5.0, 0.0],
+                "Lazy": False,
             },
         }
     }
@@ -271,7 +274,6 @@ def test_xy_sample() -> None:
         ],
         "type-b": [
             [1.0, 4.0, 0.0],
-            [1.0, 5.0, 0.0],
         ],
     }
     y_expected = {
@@ -282,9 +284,10 @@ def test_xy_sample() -> None:
         ],
         "type-b": [
             [False, True],
-            [False, True],
         ],
     }
-    x_actual, y_actual = StaticLazyConstraintsComponent.xy_sample(instance, sample)
+    xy = StaticLazyConstraintsComponent.xy_sample(features, sample)
+    assert xy is not None
+    x_actual, y_actual = xy
     assert x_actual == x_expected
     assert y_actual == y_expected

tests/components/test_objective.py

@@ -11,35 +11,10 @@ from numpy.testing import assert_array_equal
 from miplearn.instance import Instance
 from miplearn.classifiers import Regressor
 from miplearn.components.objective import ObjectiveValueComponent
-from miplearn.types import TrainingSample
+from miplearn.types import TrainingSample, Features
 from tests.fixtures.knapsack import get_test_pyomo_instances
 
 
-def test_xy_sample() -> None:
-    instance = cast(Instance, Mock(spec=Instance))
-    instance.features = {
-        "Instance": {
-            "User features": [1.0, 2.0],
-        }
-    }
-    sample: TrainingSample = {
-        "Lower bound": 1.0,
-        "Upper bound": 2.0,
-        "LP value": 3.0,
-    }
-    x_expected = {
-        "Lower bound": [[1.0, 2.0, 3.0]],
-        "Upper bound": [[1.0, 2.0, 3.0]],
-    }
-    y_expected = {
-        "Lower bound": [[1.0]],
-        "Upper bound": [[2.0]],
-    }
-    x_actual, y_actual = ObjectiveValueComponent.xy_sample(instance, sample)
-    assert x_actual == x_expected
-    assert y_actual == y_expected
-
-
 def test_x_y_predict() -> None:
     # Construct instance
     instance = cast(Instance, Mock(spec=Instance))
@@ -125,3 +100,54 @@ def test_obj_evaluate():
             "R2": -5.012843605607331,
         },
     }
+
+
+def test_xy_sample_with_lp() -> None:
+    features: Features = {
+        "Instance": {
+            "User features": [1.0, 2.0],
+        }
+    }
+    sample: TrainingSample = {
+        "Lower bound": 1.0,
+        "Upper bound": 2.0,
+        "LP value": 3.0,
+    }
+    x_expected = {
+        "Lower bound": [[1.0, 2.0, 3.0]],
+        "Upper bound": [[1.0, 2.0, 3.0]],
+    }
+    y_expected = {
+        "Lower bound": [[1.0]],
+        "Upper bound": [[2.0]],
+    }
+    xy = ObjectiveValueComponent.xy_sample(features, sample)
+    assert xy is not None
+    x_actual, y_actual = xy
+    assert x_actual == x_expected
+    assert y_actual == y_expected
+
+
+def test_xy_sample_without_lp() -> None:
+    features: Features = {
+        "Instance": {
+            "User features": [1.0, 2.0],
+        }
+    }
+    sample: TrainingSample = {
+        "Lower bound": 1.0,
+        "Upper bound": 2.0,
+    }
+    x_expected = {
+        "Lower bound": [[1.0, 2.0]],
+        "Upper bound": [[1.0, 2.0]],
+    }
+    y_expected = {
+        "Lower bound": [[1.0]],
+        "Upper bound": [[2.0]],
+    }
+    xy = ObjectiveValueComponent.xy_sample(features, sample)
+    assert xy is not None
+    x_actual, y_actual = xy
+    assert x_actual == x_expected
+    assert y_actual == y_expected

tests/components/test_primal.py

@@ -1,22 +1,22 @@
 #  MIPLearn: Extensible Framework for Learning-Enhanced Mixed-Integer Optimization
 #  Copyright (C) 2020, UChicago Argonne, LLC. All rights reserved.
 #  Released under the modified BSD license. See COPYING.md for more details.
-from typing import cast, List
-from unittest.mock import Mock, call
+
+from typing import cast
+from unittest.mock import Mock
 
 import numpy as np
 from numpy.testing import assert_array_equal
 
 from miplearn import Classifier
-from miplearn.classifiers.threshold import Threshold, MinPrecisionThreshold
+from miplearn.classifiers.threshold import Threshold
 from miplearn.components.primal import PrimalSolutionComponent
 from miplearn.instance import Instance
-from miplearn.types import TrainingSample
+from miplearn.types import TrainingSample, Features
 
 
 def test_xy_sample_with_lp_solution() -> None:
-    instance = cast(Instance, Mock(spec=Instance))
-    instance.features = {
+    features: Features = {
         "Variables": {
             "x": {
                 0: {
@@ -56,34 +56,28 @@ def test_xy_sample_with_lp_solution() -> None:
         },
     }
     x_expected = {
-        "default": np.array(
-            [
-                [0.0, 0.0, 0.1],
-                [1.0, 0.0, 0.1],
-                [1.0, 1.0, 0.1],
-            ]
-        )
+        "default": [
+            [0.0, 0.0, 0.1],
+            [1.0, 0.0, 0.1],
+            [1.0, 1.0, 0.1],
+        ]
     }
     y_expected = {
-        "default": np.array(
-            [
-                [True, False],
-                [False, True],
-                [True, False],
-            ]
-        )
+        "default": [
+            [True, False],
+            [False, True],
+            [True, False],
+        ]
     }
-    x_actual, y_actual = PrimalSolutionComponent.xy_sample(instance, sample)
-    assert len(x_actual.keys()) == 1
-    assert len(y_actual.keys()) == 1
-    assert_array_equal(x_actual["default"], x_expected["default"])
-    assert_array_equal(y_actual["default"], y_expected["default"])
+    xy = PrimalSolutionComponent.xy_sample(features, sample)
+    assert xy is not None
+    x_actual, y_actual = xy
+    assert x_actual == x_expected
+    assert y_actual == y_expected
 
 
 def test_xy_sample_without_lp_solution() -> None:
-    comp = PrimalSolutionComponent()
-    instance = cast(Instance, Mock(spec=Instance))
-    instance.features = {
+    features: Features = {
         "Variables": {
             "x": {
                 0: {
@@ -115,28 +109,24 @@ def test_xy_sample_without_lp_solution() -> None:
         },
     }
     x_expected = {
-        "default": np.array(
-            [
-                [0.0, 0.0],
-                [1.0, 0.0],
-                [1.0, 1.0],
-            ]
-        )
+        "default": [
+            [0.0, 0.0],
+            [1.0, 0.0],
+            [1.0, 1.0],
+        ]
     }
     y_expected = {
-        "default": np.array(
-            [
-                [True, False],
-                [False, True],
-                [True, False],
-            ]
-        )
+        "default": [
+            [True, False],
+            [False, True],
+            [True, False],
+        ]
     }
-    x_actual, y_actual = comp.xy_sample(instance, sample)
-    assert len(x_actual.keys()) == 1
-    assert len(y_actual.keys()) == 1
-    assert_array_equal(x_actual["default"], x_expected["default"])
-    assert_array_equal(y_actual["default"], y_expected["default"])
+    xy = PrimalSolutionComponent.xy_sample(features, sample)
+    assert xy is not None
+    x_actual, y_actual = xy
+    assert x_actual == x_expected
+    assert y_actual == y_expected
 
 
 def test_predict() -> None:

tests/test_features.py

@@ -44,6 +44,7 @@ def test_knapsack() -> None:
             },
             "Sense": "<",
             "RHS": 67.0,
+            "Lazy": False,
             "Category": "eq_capacity",
             "User features": [0.0],
         }