Correct Graph Coloring Difficulty (#318)

* correct gcolor difficulty

* refactor test

reasoning_gym/algorithmic/graph_color.py

@@ -3,7 +3,7 @@ from dataclasses import dataclass
 from random import Random
 from typing import Any, Optional
 
-from ..coaching import AttributeType, BaseCurriculum, RangeAttributeDefinition
+from ..coaching import AttributeType, BaseCurriculum, RangeAttributeDefinition, ScalarAttributeDefinition
 from ..factory import ProceduralDataset, register_dataset
 
 
@@ -155,11 +155,10 @@ def greedy_graph_coloring(puzzle):
 class GraphColorConfig:
     """Configuration for GraphColor puzzle generation"""
 
-    min_num_colors: int = 3
+    num_colors: int = 3
-    max_num_colors: int = 3
     min_num_vertices: int = 10
     max_num_vertices: int = 10
-    edge_probability: float = 0.4
+    edge_probability: float = 0.1
     seed: Optional[int] = None
     size: int = 500
 
@@ -188,7 +187,7 @@ class GraphColorDataset(ProceduralDataset):
         puzzle = None
         solution = None
         num_vertices = rng.randint(self.config.min_num_vertices, self.config.max_num_vertices)
-        num_colors = rng.randint(self.config.min_num_colors, self.config.max_num_colors)
+        num_colors = self.config.num_colors
         while solution is None:
             puzzle = generate_graph_coloring_puzzle(
                 rng=rng,
@@ -197,6 +196,8 @@ class GraphColorDataset(ProceduralDataset):
                 num_colors=num_colors,
             )
             solution = greedy_graph_coloring(puzzle)
+            if not solution:
+                num_vertices = rng.randint(self.config.min_num_vertices, self.config.max_num_vertices)
 
         edges = str(puzzle["edges"])
         question = f"""Please provide a coloring for this graph such that every vertex is not connected to a vertex of the same color. The graph has these properties:
@@ -253,23 +254,22 @@ class GraphColorCurriculum(BaseCurriculum):
         self._define_attributes(
             RangeAttributeDefinition(
                 name="num_vertices",
-                levels=[10, 20, 30, 40],
+                levels=[10, 20, 25, 50],
                 default_level=0,
                 description="Number of vertices in the graph",
-                attr_type=AttributeType.APPEND,
+                attr_type=AttributeType.STATIC,
                 min_value=10,
                 lower_field_name="min_num_vertices",
                 upper_field_name="max_num_vertices",
             ),
-            RangeAttributeDefinition(
+            ScalarAttributeDefinition(
                 name="num_colors",
-                levels=[3, 4, 5, 6],
+                field_name="num_colors",
+                levels=[5, 4, 3],
                 default_level=0,
                 description="Number of colors in the graph",
-                attr_type=AttributeType.APPEND,
+                attr_type=AttributeType.STATIC,
-                min_value=2,
+                min_value=3,
-                lower_field_name="min_num_colors",
-                upper_field_name="max_num_colors",
             ),
         )
 

tests/test_graph_color.py

@@ -12,8 +12,7 @@ def test_graph_color():
         size=10,
         min_num_vertices=10,
         max_num_vertices=10,
-        min_num_colors=4,
+        num_colors=4,
-        max_num_colors=4,
         edge_probability=0.4,
     )
     dataset = GraphColorDataset(config)
@@ -35,9 +34,8 @@ def test_graph_color():
         size=1,
         min_num_vertices=10,
         max_num_vertices=10,
-        min_num_colors=3,
+        num_colors=3,
-        max_num_colors=3,
+        edge_probability=0.1,
-        edge_probability=0.3,
     )
     dataset = GraphColorDataset(config)
 
@@ -49,11 +47,10 @@ def test_graph_color():
     config = GraphColorConfig(
         seed=42,
         size=1,
-        min_num_vertices=40,
+        min_num_vertices=15,
-        max_num_vertices=40,
+        max_num_vertices=15,
-        min_num_colors=4,
+        num_colors=3,
-        max_num_colors=4,
+        edge_probability=0.1,
-        edge_probability=0.2,
     )
     dataset = GraphColorDataset(config)
 
@@ -67,8 +64,7 @@ def test_graph_color():
         size=1,
         min_num_vertices=50,
         max_num_vertices=50,
-        min_num_colors=3,
+        num_colors=3,
-        max_num_colors=3,
         edge_probability=0.1,
     )
     dataset = GraphColorDataset(config)
@@ -87,38 +83,15 @@ def test_graph_color_curriculum():
     assert base_cfg.size == 150
     assert base_cfg.seed == 1
     assert base_cfg.min_num_vertices == base_cfg.max_num_vertices == 10
-    assert base_cfg.min_num_colors == base_cfg.max_num_colors == 3
+    assert base_cfg.num_colors == base_cfg.num_colors == 5
 
     curriculum.increment_attr_level("num_vertices")
     cfg = curriculum.generate_configuration(base_value)
-    assert cfg.min_num_vertices == 10
+    assert cfg.min_num_vertices == 20
-    assert cfg.max_num_vertices == 20
 
     curriculum.increment_attr_level("num_colors")
     cfg = curriculum.generate_configuration(base_value)
-    assert cfg.min_num_colors == 3
+    assert cfg.num_colors == 4
-    assert cfg.max_num_colors == 4
-
-    curriculum.increment_attr_level("num_vertices")
     curriculum.increment_attr_level("num_colors")
     cfg = curriculum.generate_configuration(base_value)
-    assert cfg.min_num_vertices == 10
+    assert cfg.num_colors == 3
-    assert cfg.max_num_vertices == 30
-    assert cfg.min_num_colors == 3
-    assert cfg.max_num_colors == 5
-
-    curriculum.increment_attr_level("num_vertices")
-    curriculum.increment_attr_level("num_colors")
-    cfg = curriculum.generate_configuration(base_value)
-    assert cfg.min_num_vertices == 10
-    assert cfg.max_num_vertices == 40
-    assert cfg.min_num_colors == 3
-    assert cfg.max_num_colors == 6
-
-    curriculum.decrement_attr_level("num_vertices")
-    curriculum.decrement_attr_level("num_colors")
-    cfg = curriculum.generate_configuration(base_value)
-    assert cfg.min_num_vertices == 10
-    assert cfg.max_num_vertices == 30
-    assert cfg.min_num_colors == 3
-    assert cfg.max_num_colors == 5