Merge pull request #291 from marco-2023/update_scipy

Fix spherical harmonics for new SciPy sph_harm_y API

Author: marco-2023

.github/workflows/pytest.yaml

@@ -18,7 +18,7 @@ jobs:
       matrix:
         # os: ["ubuntu-latest", "windows-latest", "macos-latest"]
         os: ["ubuntu-latest", "windows-latest"]
-        py: ["3.9", "3.10", "3.11", "3.12", "3.13"]
+        py: ["3.10", "3.11", "3.12", "3.13"]
 
     steps:
       - uses: "actions/checkout@v6"

pyproject.toml

@@ -21,16 +21,15 @@ classifiers = [
     'Intended Audience :: Science/Research',
     "Intended Audience :: Education",
     "Natural Language :: English",
-    "Programming Language :: Python :: 3.9",
     "Programming Language :: Python :: 3.10",
     "Programming Language :: Python :: 3.11",
     "Programming Language :: Python :: 3.12",
     "Programming Language :: Python :: 3.13",
 ]
 dependencies = [
-    "numpy>=1.16",
+    "numpy>=1.23.5",
     "pytest>=8.0.0",
-    "scipy>=1.4",
+    "scipy>=1.15.0",
     "importlib_resources",
     "sympy",
 ]

src/grid/tests/test_atomgrid.py

@@ -29,6 +29,7 @@
     assert_raises,
 )
 from scipy.spatial.transform import Rotation as R
+from scipy.integrate import trapezoid
 
 from grid.angular import LEBEDEV_DEGREES, AngularGrid
 from grid.atomgrid import AtomGrid, _get_rgrid_size
@@ -688,7 +689,7 @@ def func(sph_points):
 
         # Test the integral of spherical average is the integral of Gaussian e^(-x^2)e^(-y^2)...
         #   from -infinity to infinity which is equal to pi^(3/2)
-        integral = 4.0 * np.pi * np.trapz(y=spherical_avg(oned_grid) * oned_grid**2.0, x=oned_grid)
+        integral = 4.0 * np.pi * trapezoid(y=spherical_avg(oned_grid) * oned_grid**2.0, x=oned_grid)
         actual_integral = np.sqrt(np.pi) ** 3.0
         assert_allclose(actual_integral, integral)
 

src/grid/utils.py

@@ -20,7 +20,7 @@
 """Utility Module."""
 
 import numpy as np
-from scipy.special import sph_harm
+from scipy.special import sph_harm_y
 
 _bragg = np.array(
     [
@@ -560,21 +560,23 @@ def generate_real_spherical_harmonics_scipy(l_max: int, theta: np.ndarray, phi:
     l_list = np.arange(l_max + 1)
     for l_val in l_list:
         # generate m=0 real spherical harmonic
-        zero_real_sph = sph_harm(0, l_val, theta, phi).real
+        zero_real_sph = sph_harm_y(l_val, 0, phi, theta).real
 
-        # generate order m=positive real spherical harmonic
-        m_list_p = np.arange(1, l_val + 1, dtype=float)
+        # generate order m=positive real spherical harmonic: sqrt(2) * (-1)^m * Re(Y_l^m)
+        m_list_p = np.arange(1, l_val + 1, dtype=int)
         pos_real_sph = (
-            sph_harm(m_list_p[:, None], l_val, theta, phi).real
-            * np.sqrt(2)
-            * (-1) ** m_list_p[:, None]  # Remove Conway phase from SciPy
+            np.sqrt(2)
+            * (-1) ** m_list_p[:, None]
+            * sph_harm_y(l_val, m_list_p[:, None], phi, theta).real
+            # Remove Conway phase from SciPy
         )
-        # generate order m=negative real spherical harmonic
-        m_list_n = np.arange(-1, -l_val - 1, -1, dtype=float)
+        # generate order m=negative real spherical harmonic: sqrt(2) * (-1)^m * Im(Y_l^m)
         neg_real_sph = (
-            sph_harm(m_list_p[:, None], l_val, theta, phi).imag
-            * np.sqrt(2)
-            * (-1) ** m_list_n[:, None]  # Remove Conway phase from SciPy
+            np.sqrt(2)
+            * (-1) ** m_list_p[:, None]
+            * sph_harm_y(
+                l_val, m_list_p[:, None], phi, theta
+            ).imag  # Remove Conway phase from SciPy
         )
 
         # Convert to horton 2 order
@@ -753,7 +755,8 @@ def generate_derivative_real_spherical_harmonics(l_max: int, theta: np.ndarray,
     sph_harm_vals = generate_real_spherical_harmonics(l_max, theta, phi)
     i_output = 0
     for l_val in l_list:
-        for m in [0, *sum([[x, -x] for x in range(1, l_val + 1)], [])]:
+        m_values = [0] + [m for x in range(1, l_val + 1) for m in (x, -x)]
+        for m in m_values:
             # Take all spherical harmonics at degree l_val
             sph_harm_degree = sph_harm_vals[(l_val) ** 2 : (l_val + 1) ** 2, :]
 
@@ -777,7 +780,7 @@ def index_m(m):
             # Compute it using SciPy, removing conway phase (-1)^m and multiply by 2^0.5.
             sph_harm_m = (
                 fac
-                * sph_harm(np.abs(float(m)) + 1, l_val, theta, phi)
+                * sph_harm_y(l_val, np.abs(int(m)) + 1, phi, theta)
                 * np.sqrt(2)
                 * (-1.0) ** float(m)
             )