add random paren grouping

reasoning_gym/arithmetic/decimal_arithmetic.py

@@ -25,10 +25,33 @@ class DecimalArithmeticDatasetConfig:
         ), "precision must be 2 or more higher than max_num_decimal_places"
 
 
+def build_grouped_expression(operands, operators, rng):
+    """
+    Recursively build an arithmetic expression string from operands and operators,
+    inserting parentheses at random.
+
+    The expression is built by choosing a random split among the operands;
+    the operator at that split becomes the “root” of the subexpression.
+    With 50% chance, the resulting combination is wrapped in parentheses.
+    """
+    if len(operands) == 1:
+        return operands[0]
+    # Randomly choose a split point (1 <= split < len(operands)).
+    split = rng.randint(1, len(operands) - 1)
+    left_expr = build_grouped_expression(operands[:split], operators[: split - 1], rng)
+    right_expr = build_grouped_expression(operands[split:], operators[split:], rng)
+    # The operator at position (split - 1) is the one combining the two groups.
+    expr = left_expr + operators[split - 1] + right_expr
+    # Randomly decide to add parentheses around this subexpression.
+    if rng.choice([True, False]):
+        expr = "(" + expr + ")"
+    return expr
+
+
 def generate_arithmetic_problem(rng, min_num_decimal_places, max_num_decimal_places, terms=2, operations=None):
     """
     Generates a simple arithmetic problem with decimal numbers (as a string) formatted
-    to a specific number of decimal places.
+    to a specific number of decimal places, with random parenthesis grouping.
 
     Parameters:
         rng: Random number generator.
@@ -43,10 +66,11 @@ def generate_arithmetic_problem(rng, min_num_decimal_places, max_num_decimal_pla
     if operations is None:
         operations = ["+", "-", "*", "/"]
 
-    tokens = []
+    operands = []
-    # Build the expression by alternating numbers and operators.
+    operators = []
+
     for i in range(terms):
-        # Choose a number of decimal places for this term.
+        # Choose a random number of decimal places for this term.
         ndp = rng.randint(min_num_decimal_places, max_num_decimal_places)
         max_integer_part = 10  # Maximum whole number before the decimal
         max_value = max_integer_part * (10**ndp)
@@ -57,12 +81,13 @@ def generate_arithmetic_problem(rng, min_num_decimal_places, max_num_decimal_pla
         num = num.quantize(Decimal(quantize_str), rounding=ROUND_HALF_UP)
         # Format the number as a string with exactly ndp decimals.
         num_str = f"{num:.{ndp}f}"
-        tokens.append(num_str)
+        operands.append(num_str)
         if i < terms - 1:
             op = rng.choice(operations)
-            tokens.append(op)
+            operators.append(op)
 
-    problem_str = "".join(tokens) + " = ?"
+    expr = build_grouped_expression(operands, operators, rng)
+    problem_str = expr + " = ?"
     return problem_str
 
 
@@ -105,8 +130,6 @@ def _eval_ast(node) -> Decimal:
         else:
             raise ValueError(f"Unsupported unary operator: {node.op}")
     elif isinstance(node, ast.Constant):  # For Python 3.8+
-        # Although ast converts numeric literals to floats,
-        # converting via str helps us get a Decimal with the intended value.
         return Decimal(str(node.value))
     elif isinstance(node, ast.Num):  # For older Python versions
         return Decimal(str(node.n))
@@ -144,7 +167,7 @@ class DecimalArithmeticDataset(ProceduralDataset):
         expr = problem_str.replace(" = ?", "").strip()
         answer = evaluate_expression(expr)
 
-        problem_str = problem_str = (
+        problem_str = (
             f"Please solve this problem to a maximum of {str(self.config.precision)} significant digits, rounding up from the half. Only reply with the final value.\n"
             + problem_str
         )
@@ -157,8 +180,6 @@ class DecimalArithmeticDataset(ProceduralDataset):
         Instead of requiring exact equality, we allow an error up to one unit in the
         least significant digit as determined by the level of precision (max_num_decimal_places).
 
-        For example, if max_num_decimal_places is 6, then an error of up to 1e-6 is accepted.
-
         Returns:
             float: 1.0 if the user's answer is within tolerance; otherwise, 0.01.
         """
@@ -170,7 +191,6 @@ class DecimalArithmeticDataset(ProceduralDataset):
             correct_ans = entry["answer"]
 
             # Determine tolerance based on the desired precision.
-            # Here, we allow a difference of 1 in the last decimal place.
             precision = self.config.max_num_decimal_places
             tol = Decimal(10) ** (-precision)
             if abs(user_ans - correct_ans) <= tol: