Don't include intermediary features in sample; rename some keys

miplearn/components/dynamic_common.py

@@ -52,7 +52,7 @@ class DynamicConstraintsComponent(Component):
         cids: Dict[str, List[str]] = {}
         constr_categories_dict = instance.get_constraint_categories()
         constr_features_dict = instance.get_constraint_features()
-        instance_features = sample.get_vector("instance_features_user")
+        instance_features = sample.get_vector("instance_features")
         assert instance_features is not None
         for cid in self.known_cids:
             # Initialize categories

miplearn/components/objective.py

@@ -79,7 +79,7 @@ class ObjectiveValueComponent(Component):
     ) -> Tuple[Dict[str, List[List[float]]], Dict[str, List[List[float]]]]:
         lp_instance_features = sample.get_vector("lp_instance_features")
         if lp_instance_features is None:
-            lp_instance_features = sample.get_vector("instance_features_user")
+            lp_instance_features = sample.get_vector("instance_features")
         assert lp_instance_features is not None
 
         # Features

miplearn/components/primal.py

@@ -142,7 +142,7 @@ class PrimalSolutionComponent(Component):
     ) -> Tuple[Dict[Category, List[List[float]]], Dict[Category, List[List[float]]]]:
         x: Dict = {}
         y: Dict = {}
-        instance_features = sample.get_vector("instance_features_user")
+        instance_features = sample.get_vector("instance_features")
         mip_var_values = sample.get_vector("mip_var_values")
         var_features = sample.get_vector_list("lp_var_features")
         var_names = sample.get_vector("var_names")

miplearn/components/static_lazy.py

@@ -204,14 +204,14 @@ class StaticLazyConstraintsComponent(Component):
         x: Dict[str, List[List[float]]] = {}
         y: Dict[str, List[List[float]]] = {}
         cids: Dict[str, List[str]] = {}
-        instance_features = sample.get_vector("instance_features_user")
+        instance_features = sample.get_vector("instance_features")
         constr_features = sample.get_vector_list("lp_constr_features")
         constr_names = sample.get_vector("constr_names")
         constr_categories = sample.get_vector("constr_categories")
         constr_lazy = sample.get_vector("constr_lazy")
         lazy_enforced = sample.get_set("lazy_enforced")
         if constr_features is None:
-            constr_features = sample.get_vector_list("constr_features_user")
+            constr_features = sample.get_vector_list("constr_features")
 
         assert instance_features is not None
         assert constr_features is not None

miplearn/features/extractor.py

@@ -5,7 +5,7 @@
 import collections
 import numbers
 from math import log, isfinite
-from typing import TYPE_CHECKING, Dict, Optional, List, Any
+from typing import TYPE_CHECKING, Dict, Optional, List, Any, Tuple
 
 import numpy as np
 
@@ -42,16 +42,19 @@ class FeaturesExtractor:
         # sample.put("constr_lhs", constraints.lhs)
         sample.put_vector("constr_rhs", constraints.rhs)
         sample.put_vector("constr_senses", constraints.senses)
-        self._extract_user_features_vars(instance, sample)
+        vars_features_user, var_categories = self._extract_user_features_vars(
+            instance, sample
+        )
+        sample.put_vector("var_categories", var_categories)
         self._extract_user_features_constrs(instance, sample)
         self._extract_user_features_instance(instance, sample)
-        self._extract_var_features_AlvLouWeh2017(sample)
+        alw17 = self._extract_var_features_AlvLouWeh2017(sample)
         sample.put_vector_list(
             "var_features",
             self._combine(
                 [
-                    sample.get_vector_list("var_features_AlvLouWeh2017"),
-                    sample.get_vector_list("var_features_user"),
+                    alw17,
+                    vars_features_user,
                     sample.get_vector("var_lower_bounds"),
                     sample.get_vector("var_obj_coeffs"),
                     sample.get_vector("var_upper_bounds"),
@@ -80,12 +83,12 @@ class FeaturesExtractor:
         sample.put_vector("lp_constr_sa_rhs_down", constraints.sa_rhs_down)
         sample.put_vector("lp_constr_sa_rhs_up", constraints.sa_rhs_up)
         sample.put_vector("lp_constr_slacks", constraints.slacks)
-        self._extract_var_features_AlvLouWeh2017(sample, prefix="lp_")
+        alw17 = self._extract_var_features_AlvLouWeh2017(sample)
         sample.put_vector_list(
             "lp_var_features",
             self._combine(
                 [
-                    sample.get_vector_list("lp_var_features_AlvLouWeh2017"),
+                    alw17,
                     sample.get_vector("lp_var_reduced_costs"),
                     sample.get_vector("lp_var_sa_lb_down"),
                     sample.get_vector("lp_var_sa_lb_up"),
@@ -105,7 +108,7 @@ class FeaturesExtractor:
             "lp_constr_features",
             self._combine(
                 [
-                    sample.get_vector_list("constr_features_user"),
+                    sample.get_vector_list("constr_features"),
                     sample.get_vector("lp_constr_dual_values"),
                     sample.get_vector("lp_constr_sa_rhs_down"),
                     sample.get_vector("lp_constr_sa_rhs_up"),
@@ -113,11 +116,11 @@ class FeaturesExtractor:
                 ],
             ),
         )
-        instance_features_user = sample.get_vector("instance_features_user")
-        assert instance_features_user is not None
+        instance_features = sample.get_vector("instance_features")
+        assert instance_features is not None
         sample.put_vector(
             "lp_instance_features",
-            instance_features_user
+            instance_features
             + [
                 sample.get_scalar("lp_value"),
                 sample.get_scalar("lp_wallclock_time"),
@@ -138,7 +141,7 @@ class FeaturesExtractor:
         self,
         instance: "Instance",
         sample: Sample,
-    ) -> None:
+    ) -> Tuple[List, List]:
         categories: List[Optional[str]] = []
         user_features: List[Optional[List[float]]] = []
         var_features_dict = instance.get_variable_features()
@@ -174,8 +177,7 @@ class FeaturesExtractor:
                     )
                 user_features_i = list(user_features_i)
             user_features.append(user_features_i)
-        sample.put_vector("var_categories", categories)
-        sample.put_vector_list("var_features_user", user_features)
+        return user_features, categories
 
     def _extract_user_features_constrs(
         self,
@@ -224,7 +226,7 @@ class FeaturesExtractor:
                 lazy.append(instance.is_constraint_lazy(cname))
             else:
                 lazy.append(False)
-        sample.put_vector_list("constr_features_user", user_features)
+        sample.put_vector_list("constr_features", user_features)
         sample.put_vector("constr_lazy", lazy)
         sample.put_vector("constr_categories", categories)
 
@@ -247,16 +249,12 @@ class FeaturesExtractor:
             )
         constr_lazy = sample.get_vector("constr_lazy")
         assert constr_lazy is not None
-        sample.put_vector("instance_features_user", user_features)
+        sample.put_vector("instance_features", user_features)
         sample.put_scalar("static_lazy_count", sum(constr_lazy))
 
     # Alvarez, A. M., Louveaux, Q., & Wehenkel, L. (2017). A machine learning-based
     # approximation of strong branching. INFORMS Journal on Computing, 29(1), 185-195.
-    def _extract_var_features_AlvLouWeh2017(
-        self,
-        sample: Sample,
-        prefix: str = "",
-    ) -> None:
+    def _extract_var_features_AlvLouWeh2017(self, sample: Sample) -> List:
         obj_coeffs = sample.get_vector("var_obj_coeffs")
         obj_sa_down = sample.get_vector("lp_var_sa_obj_down")
         obj_sa_up = sample.get_vector("lp_var_sa_obj_up")
@@ -328,7 +326,7 @@ class FeaturesExtractor:
             for v in f:
                 assert isfinite(v), f"non-finite elements detected: {f}"
             features.append(f)
-        sample.put_vector_list(f"{prefix}var_features_AlvLouWeh2017", features)
+        return features
 
     def _combine(
         self,

tests/components/test_dynamic_lazy.py

@@ -25,13 +25,13 @@ def training_instances() -> List[Instance]:
         MemorySample(
             {
                 "lazy_enforced": {"c1", "c2"},
-                "instance_features_user": [5.0],
+                "instance_features": [5.0],
             },
         ),
         MemorySample(
             {
                 "lazy_enforced": {"c2", "c3"},
-                "instance_features_user": [5.0],
+                "instance_features": [5.0],
             },
         ),
     ]
@@ -56,7 +56,7 @@ def training_instances() -> List[Instance]:
         MemorySample(
             {
                 "lazy_enforced": {"c3", "c4"},
-                "instance_features_user": [8.0],
+                "instance_features": [8.0],
             },
         )
     ]

tests/components/test_primal.py

@@ -25,7 +25,7 @@ def sample() -> Sample:
             "var_names": ["x[0]", "x[1]", "x[2]", "x[3]"],
             "var_categories": ["default", None, "default", "default"],
             "mip_var_values": [0.0, 1.0, 1.0, 0.0],
-            "instance_features_user": [5.0],
+            "instance_features": [5.0],
             "var_features": [
                 [0.0, 0.0],
                 None,

tests/components/test_static_lazy.py

@@ -33,7 +33,7 @@ def sample() -> Sample:
             ],
             "constr_lazy": [True, True, True, True, False],
             "constr_names": ["c1", "c2", "c3", "c4", "c5"],
-            "instance_features_user": [5.0],
+            "instance_features": [5.0],
             "lazy_enforced": {"c1", "c2", "c4"},
             "lp_constr_features": [
                 [1.0, 1.0],

tests/features/test_extractor.py

@@ -35,20 +35,6 @@ def test_knapsack() -> None:
         sample.get_vector("var_categories"),
         ["default", "default", "default", "default", None],
     )
-    assert_equals(
-        sample.get_vector_list("var_features_user"),
-        [[23.0, 505.0], [26.0, 352.0], [20.0, 458.0], [18.0, 220.0], None],
-    )
-    assert_equals(
-        sample.get_vector_list("var_features_AlvLouWeh2017"),
-        [
-            [1.0, 0.32899, 0.0],
-            [1.0, 0.229316, 0.0],
-            [1.0, 0.298371, 0.0],
-            [1.0, 0.143322, 0.0],
-            [0.0, 0.0, 0.0],
-        ],
-    )
     assert sample.get_vector_list("var_features") is not None
     assert_equals(sample.get_vector("constr_names"), ["eq_capacity"])
     # assert_equals(
@@ -65,10 +51,10 @@ def test_knapsack() -> None:
     # )
     assert_equals(sample.get_vector("constr_rhs"), [0.0])
     assert_equals(sample.get_vector("constr_senses"), ["="])
-    assert_equals(sample.get_vector("constr_features_user"), [None])
+    assert_equals(sample.get_vector("constr_features"), [None])
     assert_equals(sample.get_vector("constr_categories"), ["eq_capacity"])
     assert_equals(sample.get_vector("constr_lazy"), [False])
-    assert_equals(sample.get_vector("instance_features_user"), [67.0, 21.75])
+    assert_equals(sample.get_vector("instance_features"), [67.0, 21.75])
     assert_equals(sample.get_scalar("static_lazy_count"), 0)
 
     # after-lp
@@ -104,16 +90,6 @@ def test_knapsack() -> None:
     )
     assert_equals(sample.get_vector("lp_var_sa_ub_up"), [2.043478, inf, 2.2, inf, 69.0])
     assert_equals(sample.get_vector("lp_var_values"), [1.0, 0.923077, 1.0, 0.0, 67.0])
-    assert_equals(
-        sample.get_vector_list("lp_var_features_AlvLouWeh2017"),
-        [
-            [1.0, 0.32899, 0.0, 0.0, 1.0, 1.0, 5.265874, 46.051702],
-            [1.0, 0.229316, 0.0, 0.076923, 1.0, 1.0, 3.532875, 5.388476],
-            [1.0, 0.298371, 0.0, 0.0, 1.0, 1.0, 5.232342, 46.051702],
-            [1.0, 0.143322, 0.0, 0.0, 1.0, -1.0, 46.051702, 3.16515],
-            [0.0, 0.0, 0.0, 0.0, 1.0, -1.0, 0.0, 0.0],
-        ],
-    )
     assert sample.get_vector_list("lp_var_features") is not None
     assert_equals(sample.get_vector("lp_constr_basis_status"), ["N"])
     assert_equals(sample.get_vector("lp_constr_dual_values"), [13.538462])

tests/features/test_sample.py

@@ -26,7 +26,7 @@ def _test_sample(sample: Sample) -> None:
     _assert_roundtrip_scalar(sample, 1.0)
 
     # Vector
-    _assert_roundtrip_vector(sample, ["A", "BB", "CCC", "こんにちは", None])
+    _assert_roundtrip_vector(sample, ["A", "BB", "CCC", None])
     _assert_roundtrip_vector(sample, [True, True, False])
     _assert_roundtrip_vector(sample, [1, 2, 3])
     _assert_roundtrip_vector(sample, [1.0, 2.0, 3.0])