Feature/scipy interpolation issue 261

src/grid/cubic.py

@@ -142,9 +142,10 @@ def interpolate(self, points, values, use_log=False, nu_x=0, nu_y=0, nu_z=0, met
             The interpolation of a function (or of it's derivatives) at a :math:`M` point.
 
         """
-        if method not in ["cubic", "linear", "nearest"]:
+        supported_methods = ["linear", "nearest", "slinear", "cubic", "quintic", "pchip"]
+        if method not in supported_methods:
             raise ValueError(
-                f"Argument method should be either cubic, linear, or nearest , got {method}"
+                f"Argument method should be one of {supported_methods}, got {method}"
             )
         if self.ndim != 3:
             raise NotImplementedError(
@@ -159,12 +160,21 @@ def interpolate(self, points, values, use_log=False, nu_x=0, nu_y=0, nu_z=0, met
         if use_log:
             values = np.log(values)
 
-        # Use scipy if linear and nearest is requested and raise error if it's not cubic.
-        if method in ["linear", "nearest"]:
+        # Use scipy if no derivatives are requested
+        if method in supported_methods and (nu_x == 0 and nu_y == 0 and nu_z == 0):
             x, y, z = self.get_points_along_axes()
             values = values.reshape(self.shape)
             interpolate = RegularGridInterpolator((x, y, z), values, method=method)
-            return interpolate(points)
+            interpolated = interpolate(points)
+            if use_log:
+                return np.exp(interpolated)
+            return interpolated
+
+        # At this point, derivatives are requested, which requires our custom cubic spline implementation
+        if method != "cubic":
+            raise NotImplementedError(
+                f"Computing analytical derivatives (nu_x={nu_x}, nu_y={nu_y}, nu_z={nu_z}) is only supported for the 'cubic' method."
+            )
 
         # Interpolate the Z-Axis.
         def z_spline(z, x_index, y_index, nu_z=nu_z):

src/grid/tests/test_cubic.py

@@ -119,7 +119,7 @@ def test_raise_error_when_using_interpolation(self):
                 points, values, method="not cubic"
             )
         self.assertEqual(
-            "Argument method should be either cubic, linear, or nearest , got not cubic",
+            "Argument method should be one of ['linear', 'nearest', 'slinear', 'cubic', 'quintic', 'pchip'], got not cubic",
             str(err.exception),
         )
         # Test raises error if dimension is two.
@@ -201,10 +201,10 @@ def gaussian(points):
         num_pts = 500
         random_pts = np.random.uniform(-0.9, 0.9, (num_pts, 3))
         interpolated = cubic.interpolate(random_pts, gaussian_pts, use_log=False)
-        assert_allclose(interpolated, gaussian(random_pts), rtol=1e-5, atol=1e-6)
+        assert_allclose(interpolated, gaussian(random_pts), rtol=1e-5, atol=1e-5)
 
-        interpolated = cubic.interpolate(random_pts, gaussian_pts, use_log=True)
-        assert_allclose(interpolated, gaussian(random_pts), rtol=1e-5, atol=1e-6)
+        interpolated_log = cubic.interpolate(random_pts, gaussian_pts, use_log=True)
+        assert_allclose(interpolated_log, gaussian(random_pts), rtol=1e-5, atol=1e-5)
 
     def test_interpolation_of_linear_function_using_scipy_linear_method(self):
         r"""Test interpolation of a linear function using scipy with linear method."""