Replace interp1d #2394

Author: jason-rpkt

docs/examples/shading/plot_partial_module_shading_simple.py

@@ -38,7 +38,7 @@
 from pvlib import pvsystem, singlediode
 import pandas as pd
 import numpy as np
-from scipy.interpolate import interp1d
+from scipy.interpolate import make_interp_spline
 import matplotlib.pyplot as plt
 
 from scipy.constants import e as qe, k as kB
@@ -178,9 +178,9 @@ def plot_curves(dfs, labels, title):
 
 
 def interpolate(df, i):
-    """convenience wrapper around scipy.interpolate.interp1d"""
-    f_interp = interp1d(np.flipud(df['i']), np.flipud(df['v']), kind='linear',
-                        fill_value='extrapolate')
+    """convenience wrapper around scipy.interpolate"""
+    f_interp = make_interp_spline(np.flipud(df['i']), np.flipud(df['v']), k=1)
+
     return f_interp(i)
 
 

pvlib/iam.py

@@ -440,7 +440,7 @@ def interp(aoi, theta_ref, iam_ref, method='linear', normalize=True):
     method : str, default 'linear'
         Specifies the interpolation method.
         Useful options are: 'linear', 'quadratic', 'cubic'.
-        See scipy.interpolate.interp1d for more options.
+        See scipy.interpolate for more options.
 
     normalize : boolean, default True
         When true, the interpolated values are divided by the interpolated
@@ -470,7 +470,7 @@ def interp(aoi, theta_ref, iam_ref, method='linear', normalize=True):
     '''
     # Contributed by Anton Driesse (@adriesse), PV Performance Labs. July, 2019
 
-    from scipy.interpolate import interp1d
+    from scipy.interpolate import CubicSpline, make_interp_spline
 
     # Scipy doesn't give the clearest feedback, so check number of points here.
     MIN_REF_VALS = {'linear': 2, 'quadratic': 3, 'cubic': 4, 1: 2, 2: 3, 3: 4}
@@ -483,10 +483,31 @@ def interp(aoi, theta_ref, iam_ref, method='linear', normalize=True):
         raise ValueError("Negative value(s) found in 'iam_ref'. "
                          "This is not physically possible.")
 
-    interpolator = interp1d(theta_ref, iam_ref, kind=method,
-                            fill_value='extrapolate')
-    aoi_input = aoi
+    theta_ref = np.asarray(theta_ref)
+    iam_ref = np.asarray(iam_ref)
+
+    if method == "linear":
+        spline = make_interp_spline(theta_ref, iam_ref, k=1)
+
+        def interpolator(x):
+            return spline(x)
+
+    elif method == "quadratic":
+        spline = make_interp_spline(theta_ref, iam_ref, k=2)
+
+        def interpolator(x):
+            return spline(x)
 
+    elif method == "cubic":
+        spline = CubicSpline(theta_ref, iam_ref, extrapolate=True)
+
+        def interpolator(x):
+            return spline(x)
+
+    else:
+        raise ValueError(f"Invalid interpolation method '{method}'.")
+
+    aoi_input = aoi
     aoi = np.asanyarray(aoi)
     aoi = np.abs(aoi)
     iam = interpolator(aoi)

pvlib/spectrum/response.py

@@ -6,7 +6,7 @@
 import numpy as np
 import pandas as pd
 import scipy.constants
-from scipy.interpolate import interp1d
+from scipy.interpolate import CubicSpline
 
 
 _PLANCK_BY_LIGHT_SPEED_OVER_ELEMENTAL_CHARGE_BY_BILLION = (
@@ -66,16 +66,16 @@ def get_example_spectral_response(wavelength=None):
     if wavelength is None:
         resolution = 5.0
         wavelength = np.arange(280, 1200 + resolution, resolution)
+    x = SR_DATA[0]
+    y = SR_DATA[1]
+    spline = CubicSpline(
+        x, y,
+        extrapolate=False
+    )
+    values = spline(wavelength)
+    values[(wavelength < x[0]) | (wavelength > x[-1])] = 0.0
 
-    interpolator = interp1d(SR_DATA[0], SR_DATA[1],
-                            kind='cubic',
-                            bounds_error=False,
-                            fill_value=0.0,
-                            copy=False,
-                            assume_sorted=True)
-
-    sr = pd.Series(data=interpolator(wavelength), index=wavelength)
-
+    sr = pd.Series(data=values, index=wavelength)
     sr.index.name = 'wavelength'
     sr.name = 'spectral_response'
 

tests/test_iam.py

@@ -213,6 +213,32 @@ def test_iam_interp():
     with pytest.raises(ValueError):
         _iam.interp(0.0, [0, 90], [1, -1])
 
+    # check linear after updating interp1d
+    theta_ref = np.array([0, 60, 90])
+    iam_ref = np.array([1.0, 0.8, 0.0])
+
+    aoi = np.array([0, 30, 60])
+    iam = _iam.interp(
+        aoi, theta_ref, iam_ref,
+        method="linear", normalize=False)
+    expected = np.array([1.0, 0.9, 0.8])
+    np.testing.assert_allclose(iam, expected)
+
+    # check quadratic
+    theta_ref = np.array([0, 30, 60, 90])
+    iam_ref = 1.0 - 1e-4 * theta_ref**2
+    aoi = np.array([15, 45, 75])
+    iam = _iam.interp(
+        aoi,
+        theta_ref,
+        iam_ref,
+        method="quadratic",
+        normalize=False
+    )
+
+    expected = 1.0 - 1e-4 * aoi**2
+    np.testing.assert_allclose(iam, expected, rtol=1e-12)
+
 
 @pytest.mark.parametrize('aoi,expected', [
     (45, 0.9975036250000002),