Merge pull request #38 from Schmeitzke/main

Add Simple and Advanced Geometry Dataset Generators

reasoning_gym/geometry/__init__.py

@@ -0,0 +1,9 @@
+from .simple_geometry import SimpleGeometryConfig, SimpleGeometryDataset
+from .advanced_geometry import AdvancedGeometryConfig, AdvancedGeometryDataset
+
+__all__ = [
+    "SimpleGeometryConfig",
+    "SimpleGeometryDataset",
+    "AdvancedGeometryConfig",
+    "AdvancedGeometryDataset",
+]

reasoning_gym/geometry/advanced_geometry.py

@@ -0,0 +1,229 @@
+import random
+from dataclasses import dataclass
+from typing import Optional, List
+
+import sympy
+from sympy.geometry import Point, Triangle, Segment
+
+from ..factory import ProceduralDataset, register_dataset
+
+
+@dataclass
+class AdvancedGeometryConfig:
+    """
+    Configuration for generating advanced geometry tasks.
+    """
+    min_coord: int = -10  # Minimum x/y coordinate
+    max_coord: int = 10   # Maximum x/y coordinate
+    size: int = 50        # Number of problems to generate
+    seed: Optional[int] = None
+
+    # Probability or list of tasks we want to generate
+    # For demonstration, we have three categories:
+    task_types: List[str] = None
+
+    def __post_init__(self):
+        if self.task_types is None:
+            # Default set of advanced tasks
+            self.task_types = [
+                "orthocenter",
+                "incircle_radius",
+                "angle_measure",
+            ]
+
+    def validate(self):
+        assert self.min_coord < self.max_coord, "min_coord must be < max_coord."
+        assert self.size > 0, "Size of dataset must be positive."
+        assert len(self.task_types) > 0, "Must specify at least one task type."
+
+
+class AdvancedGeometryDataset(ProceduralDataset):
+    """
+    A dataset for advanced geometry tasks using coordinate geometry.
+    """
+
+    def __init__(self, config: AdvancedGeometryConfig):
+        self._prompt_templates = {
+            "orthocenter": [
+                "Given triangle ABC with coordinates A={A}, B={B}, and C={C}, find the coordinates of its orthocenter.",
+                "For triangle with vertices A={A}, B={B}, and C={C}, determine the orthocenter (intersection of altitudes).",
+            ],
+            "incircle_radius": [
+                "Consider triangle ABC with coordinates A={A}, B={B}, and C={C}. Compute the radius of its incircle.",
+                "Find the incircle radius of triangle ABC whose vertices are A={A}, B={B}, and C={C}.",
+            ],
+            "angle_measure": [
+                "In triangle ABC with coordinates A={A}, B={B}, and C={C}, find the measure (in degrees) of angle ABC.",
+                "Given a triangle with vertices A={A}, B={B}, C={C}, determine the angle at B in degrees.",
+            ],
+        }
+        super().__init__(config=config, seed=config.seed, size=config.size)
+
+    def __getitem__(self, idx: int) -> dict:
+        """
+        Generate a single advanced geometry item based on the config's task types.
+        """
+        rng = random.Random(self.seed + idx)
+        task_type = rng.choice(self.config.task_types)
+
+        # Randomly generate coordinates for a triangle
+        A, B, C = self._generate_non_degenerate_triangle(rng)
+
+        # Build a question and compute the solution
+        if task_type == "orthocenter":
+            question, answer, metadata = self._build_orthocenter_task(rng, A, B, C)
+        elif task_type == "incircle_radius":
+            question, answer, metadata = self._build_incircle_radius_task(rng, A, B, C)
+        elif task_type == "angle_measure":
+            question, answer, metadata = self._build_angle_measure_task(rng, A, B, C)
+        else:
+            raise ValueError(f"Unknown task_type: {task_type}")
+
+        return {
+            "question": question,
+            "answer": answer,
+            "metadata": metadata,
+        }
+
+    def _generate_non_degenerate_triangle(self, rng: random.Random):
+        """
+        Generate a random non-degenerate triangle with integer coordinates
+        in [min_coord, max_coord] x [min_coord, max_coord].
+        """
+        max_attempts = 100
+        for _ in range(max_attempts):
+            # Generate points with integer coordinates
+            points = []
+            for _ in range(3):
+                x = rng.randint(self.config.min_coord, self.config.max_coord)
+                y = rng.randint(self.config.min_coord, self.config.max_coord)
+                points.append(Point(x, y))
+
+            A, B, C = points
+
+            # Calculate signed area to check for non-degeneracy
+            # Using the formula: 1/2 * |x1(y2 - y3) + x2(y3 - y1) + x3(y1 - y2)|
+            area = abs(
+                A.x * (B.y - C.y) + 
+                B.x * (C.y - A.y) + 
+                C.x * (A.y - B.y)
+            ) / 2
+
+            if area > 0:
+                return A, B, C
+
+        raise ValueError(f"Failed to generate a non-degenerate triangle after {max_attempts} attempts.")
+
+    def _build_orthocenter_task(self, rng: random.Random, A: Point, B: Point, C: Point):
+        """
+        Build a question about finding the orthocenter of triangle ABC.
+        """
+        # Convert segments to lines
+        BC_line = sympy.Line(B, C)
+        CA_line = sympy.Line(C, A)
+
+        # Calculate altitudes by creating lines perpendicular from each vertex
+        alt_A = BC_line.perpendicular_line(A)
+        alt_B = CA_line.perpendicular_line(B)
+
+        # Find orthocenter (intersection of any two altitudes, e.g. alt_A and alt_B)
+        ortho = alt_A.intersection(alt_B)[0]
+
+        x_ortho_approx = float(ortho.x.evalf())
+        y_ortho_approx = float(ortho.y.evalf())
+
+        question_template = rng.choice(self._prompt_templates["orthocenter"])
+        question = question_template.format(
+            A=(A.x, A.y), B=(B.x, B.y), C=(C.x, C.y)
+        )
+        answer_str = f"({x_ortho_approx:.3f}, {y_ortho_approx:.3f})"
+
+        metadata = {
+            "A": (A.x, A.y),
+            "B": (B.x, B.y),
+            "C": (C.x, C.y),
+            "orthocenter_exact": (str(ortho.x), str(ortho.y)),
+            "orthocenter_approx": (x_ortho_approx, y_ortho_approx),
+        }
+        return question, answer_str, metadata
+
+
+    def _build_incircle_radius_task(self, rng: random.Random, A: Point, B: Point, C: Point):
+        """
+        Build a question about finding the incircle radius of triangle ABC.
+        """
+        # Calculate side lengths
+        a = B.distance(C)
+        b = C.distance(A)
+        c = A.distance(B)
+
+        # Semi-perimeter
+        s = (a + b + c) / 2
+
+        # Area using Heron's formula
+        area = sympy.sqrt(s * (s - a) * (s - b) * (s - c))
+
+        # Radius of incircle = Area / Semi-perimeter
+        radius = area / s
+
+        # Convert to float for final answer
+        radius_approx = float(radius.evalf())
+
+        question_template = rng.choice(self._prompt_templates["incircle_radius"])
+        question = question_template.format(
+            A=(A.x, A.y), B=(B.x, B.y), C=(C.x, C.y)
+        )
+        answer_str = f"{radius_approx:.3f}"
+
+        metadata = {
+            "A": (A.x, A.y),
+            "B": (B.x, B.y),
+            "C": (C.x, C.y),
+            "incircle_radius_exact": str(radius),
+            "incircle_radius_approx": radius_approx,
+        }
+        return question, answer_str, metadata
+
+    def _build_angle_measure_task(self, rng: random.Random, A: Point, B: Point, C: Point):
+        """
+        Build a question about finding the measure of angle ABC in degrees.
+        """
+        # Angle at B means the angle ∠ABC
+        # Vector BA = A - B, BC = C - B
+        BA = A - B
+        BC = C - B
+
+        # Use vector dot product to find angle between BA and BC
+        # angle = arccos((BA · BC) / (|BA| * |BC|))
+        dot_val = BA.dot(BC)
+        mag_ba = BA.distance(Point(0, 0))
+        mag_bc = BC.distance(Point(0, 0))
+
+        # numerical check
+        if mag_ba == 0 or mag_bc == 0:
+            # degenerate, but theoretically we forced a non-degenerate triangle
+            angle_deg = 0
+        else:
+            cos_theta = dot_val / (mag_ba * mag_bc)
+            # clamp cos_theta to [-1, 1] to avoid floating rounding errors
+            cos_theta = max(-1, min(1, cos_theta))
+            angle_rad = sympy.acos(cos_theta)
+            angle_deg = float(angle_rad.evalf() * 180 / sympy.pi)
+
+        question_template = rng.choice(self._prompt_templates["angle_measure"])
+        question = question_template.format(
+            A=(A.x, A.y), B=(B.x, B.y), C=(C.x, C.y)
+        )
+
+        answer_str = f"{angle_deg:.2f}°"
+        metadata = {
+            "A": (A.x, A.y),
+            "B": (B.x, B.y),
+            "C": (C.x, C.y),
+            "angle_ABC_degrees": angle_deg,
+        }
+        return question, answer_str, metadata
+
+
+# Register the dataset
+register_dataset("advanced_geometry", AdvancedGeometryDataset, AdvancedGeometryConfig)

reasoning_gym/geometry/simple_geometry.py

@@ -0,0 +1,140 @@
+import random
+from dataclasses import dataclass
+from typing import Optional
+
+from ..factory import ProceduralDataset, register_dataset
+
+@dataclass
+class SimpleGeometryConfig:
+    """
+    Configuration for generating basic geometry (angle-finding) tasks.
+    Produces a random convex polygon with N sides, random angles
+    for the first (N-1) sides, and asks the solver to find the last angle.
+    """
+
+    min_sides: int = 3          # Minimum number of sides (e.g. triangle)
+    max_sides: int = 6          # Maximum number of sides (e.g. hexagon)
+    min_angle: int = 10         # Minimum angle (in degrees) for each of the first (N-1) angles
+    max_angle: int = 170        # Maximum angle (in degrees) for each of the first (N-1) angles
+    seed: Optional[int] = None  # Random seed
+    size: int = 100             # Number of geometry tasks to generate
+
+    def validate(self) -> None:
+        """
+        Validate configuration parameters.
+        """
+        assert self.min_sides >= 3, "min_sides must be at least 3 (triangle)."
+        assert self.max_sides >= self.min_sides, "max_sides must be >= min_sides."
+        assert 0 < self.min_angle < 180, "min_angle must be in (0, 180)."
+        assert self.max_angle <= 179, "max_angle should be less than 180."
+        assert self.max_angle >= self.min_angle, "max_angle must be >= min_angle."
+
+
+class SimpleGeometryDataset(ProceduralDataset):
+    """
+    A dataset for simple polygon angle-finding tasks.
+    We randomly choose the number of sides N within [min_sides, max_sides].
+    We then generate (N-1) random angles (in degrees), ensuring their sum is
+    strictly less than the total sum for an (N)-sided convex polygon (which is 180*(N-2)).
+    The question asks for the missing angle; the answer is computed by subtracting the
+    sum of known angles from 180*(N-2).
+    """
+
+    def __init__(self, config: SimpleGeometryConfig):
+        self._prompt_templates = [
+            (
+                "Given a convex polygon with {n_sides} sides, its first {n_minus_1} interior angles "
+                "are: {angle_list}. What is the measure of the remaining interior angle (in degrees)?"
+            ),
+            (
+                "A convex polygon has {n_sides} sides. The measures of "
+                "the first {n_minus_1} interior angles are: {angle_list}. "
+                "Find the measure of the last interior angle."
+            ),
+            (
+                "Consider a convex {n_sides}-gon whose first {n_minus_1} interior angles "
+                "are: {angle_list}. Determine the measure of the remaining angle."
+            ),
+        ]
+        super().__init__(config=config, seed=config.seed, size=config.size)
+
+    def __getitem__(self, idx: int) -> dict:
+        """
+        Generate a single geometry angle-finding item.
+
+        Returns:
+            A dict with:
+                - question: str
+                - answer: str (the missing angle, as an integer or float in degrees)
+                - metadata: dict (n_sides, angles, sum_of_known, missing_angle, etc.)
+        """
+        rng = random.Random(self.seed + idx)
+
+        # Randomly pick the number of sides
+        n_sides = rng.randint(self.config.min_sides, self.config.max_sides)
+
+        # Total interior angle sum for a convex n_sides-gon
+        total_sum = 180 * (n_sides - 2)
+
+        # Generate (n_sides - 1) random angles, ensuring their sum < total_sum
+        known_angles = self._generate_valid_angles(rng, n_sides, total_sum)
+
+        # Missing angle
+        missing_angle = total_sum - sum(known_angles)
+
+        # Build the question string
+        angle_list_str = ", ".join(f"{a:.1f}°" for a in known_angles)
+        prompt = rng.choice(self._prompt_templates).format(
+            n_sides=n_sides,
+            n_minus_1=n_sides - 1,
+            angle_list=angle_list_str
+        )
+
+        # Round the missing angle to one decimal place or integer if it is very close to an integer
+        # so that the answer remains consistent and clean
+        missing_angle_rounded = round(missing_angle, 1)
+        if abs(missing_angle_rounded - round(missing_angle_rounded)) < 1e-6:
+            # If it is effectively an integer, keep it as int
+            missing_angle_rounded = int(missing_angle_rounded)
+
+        answer_str = str(missing_angle_rounded)
+
+        return {
+            "question": prompt,
+            "answer": answer_str,
+            "metadata": {
+                "n_sides": n_sides,
+                "known_angles": known_angles,
+                "sum_of_known_angles": sum(known_angles),
+                "missing_angle_raw": missing_angle,
+                "missing_angle_rounded": missing_angle_rounded,
+                "total_interior_sum": total_sum,
+            },
+        }
+
+    def _generate_valid_angles(self, rng: random.Random, n_sides: int, total_sum: int):
+        """
+        Generate (n_sides - 1) random angles in [min_angle, max_angle],
+        ensuring the sum is strictly less than total_sum to keep a valid missing angle.
+        We keep retrying until we find a valid set or reach a max attempt limit.
+        """
+        max_attempts = 100
+        for _ in range(max_attempts):
+            angles = []
+            # We choose angles one by one
+            for _ in range(n_sides - 1):
+                angle = rng.randint(self.config.min_angle, self.config.max_angle)
+                angles.append(float(angle))
+
+            # Check if the sum is strictly less than total_sum
+            if sum(angles) < total_sum:
+                return angles
+
+        # If we fail after max_attempts, raise an error
+        raise ValueError(
+            f"Could not generate valid angles for an {n_sides}-gon "
+            f"with total sum {total_sum} within {max_attempts} attempts."
+        )
+
+# Register the dataset so it can be accessed similarly to the others
+register_dataset("simple_geometry", SimpleGeometryDataset, SimpleGeometryConfig)