Preserve fractional cyclic coordinates for integer input

patsy/mgcv_cubic_splines.py

@@ -75,6 +75,8 @@ def _map_cyclic(x, lbound, ubound):
         )
 
     x = np.copy(x)
+    if np.issubdtype(x.dtype, np.integer):
+        x = x.astype(float)
     x[x > ubound] = lbound + (x[x > ubound] - ubound) % (ubound - lbound)
     x[x < lbound] = ubound - (lbound - x[x < lbound]) % (ubound - lbound)
 
@@ -98,6 +100,26 @@ def test__map_cyclic_errors():
     pytest.raises(ValueError, _map_cyclic, x, 4.5, 4.5)
 
 
+def test__map_cyclic_integer():
+    x = np.arange(-1, 6)
+    original = x.copy()
+    mapped = _map_cyclic(x, 0.1, 3.4)
+    np.testing.assert_allclose(mapped, [2.3, 3.3, 1.0, 2.0, 3.0, 0.7, 1.7])
+    np.testing.assert_array_equal(x, original)
+
+
+def test__map_cyclic_float_dtype():
+    for dtype in [np.float32, np.float64]:
+        x = np.arange(-1, 6, dtype=dtype)
+        original = x.copy()
+        mapped = _map_cyclic(x, 0.1, 3.4)
+        assert mapped.dtype == x.dtype
+        np.testing.assert_allclose(
+            mapped, [2.3, 3.3, 1.0, 2.0, 3.0, 0.7, 1.7], rtol=2 * np.finfo(dtype).eps
+        )
+        np.testing.assert_array_equal(x, original)
+
+
 def _get_cyclic_f(knots):
     """Returns mapping of cyclic cubic spline values to 2nd derivatives.
 

patsy/test_highlevel.py

@@ -997,3 +997,41 @@ def test_NAActioon_pandas_string_index():
     dm2 = dmatrix(di, data2, 0, return_type="dataframe")
     assert np.allclose(dm2, [[1.0, 6.0, 4.0], [1.0, 7.0, 5.0]])
     assert list(dm2.columns) == ["Intercept", "x", "z"]
+
+
+@pytest.mark.parametrize(
+    "dtype", [np.int32, np.int64, np.uint64, np.float32, np.float64]
+)
+@pytest.mark.parametrize("constraints", [None, "center"])
+def test_cc_integer_input(dtype, constraints):
+    from patsy import cc
+
+    x = np.arange(6, dtype=dtype)
+    original = x.copy()
+    kwargs = dict(
+        knots=[1.0, 2.0, 3.0],
+        lower_bound=0.1,
+        upper_bound=3.4,
+        constraints=constraints,
+    )
+    np.testing.assert_allclose(
+        cc(x, **kwargs), cc(x.astype(float), **kwargs), rtol=1e-6, atol=1e-6
+    )
+    formula = (
+        "0 + cc(x, knots=[1., 2., 3.], lower_bound=0.1, upper_bound=3.4, "
+        "constraints=%r)" % constraints
+    )
+    np.testing.assert_allclose(
+        dmatrix(formula, {"x": x}),
+        dmatrix(formula, {"x": x.astype(float)}),
+        rtol=1e-6,
+        atol=1e-6,
+    )
+    training = dmatrix(formula, {"x": np.array([1.0, 2.0, 3.0])})
+    np.testing.assert_allclose(
+        build_design_matrices([training.design_info], {"x": x})[0],
+        build_design_matrices([training.design_info], {"x": x.astype(float)})[0],
+        rtol=1e-6,
+        atol=1e-6,
+    )
+    np.testing.assert_array_equal(x, original)