Use explicit int64 sampling for integer parameter bounds

Fixes #623

Co-authored-by: Codex <noreply@openai.com>

Author: AHMETHAKANBEZIR1

bayes_opt/parameter.py

@@ -278,7 +278,9 @@ def random_sample(
             The samples.
         """
         random_state = ensure_rng(random_state)
-        return random_state.randint(self.bounds[0], self.bounds[1] + 1, n_samples).astype(float)
+        return random_state.randint(self.bounds[0], self.bounds[1] + 1, n_samples, dtype=np.int64).astype(
+            float
+        )
 
     def to_float(self, value: int | float) -> float:
         """Convert a parameter value to a float.

tests/test_parameter.py

@@ -87,6 +87,20 @@ def target_func(**kwargs):
     assert p1.kernel_transform(np.array([1.3, 3.6, 7.2])) == pytest.approx(np.array([1, 4, 7]))
 
 
+@pytest.mark.parametrize("bounds", [(3_000_000_000, 3_000_000_010), (-3_000_000_010, -3_000_000_000), (0, 5)])
+def test_int_random_sample_large_bounds(bounds):
+    parameter = IntParameter("x", bounds)
+    samples = parameter.random_sample(100, random_state=np.random.RandomState(42))
+    repeated = parameter.random_sample(100, random_state=np.random.RandomState(42))
+    assert samples.dtype == np.dtype(float)
+    np.testing.assert_array_equal(samples, repeated)
+    assert np.all(samples >= bounds[0])
+    assert np.all(samples <= bounds[1])
+    assert np.all(samples == np.floor(samples))
+    assert bounds[0] in samples
+    assert bounds[1] in samples
+
+
 def test_cat_parameters():
     fruit_ratings = {"apple": 1.0, "banana": 2.0, "mango": 5.0, "honeydew melon": -10.0, "strawberry": np.pi}