use faster method as long as only one wt type is used (even if type!=0)

py_wake/tests/test_deficit_models/test_noj.py

 import pytest
-
-import matplotlib.pyplot as plt
 import numpy as np
 from py_wake import NOJ
-from py_wake.examples.data.iea37 import iea37_path
-from py_wake.examples.data.iea37._iea37 import IEA37_WindTurbines, IEA37Site
 from py_wake.site._site import UniformSite
 from py_wake.tests import npt
 from py_wake.flow_map import HorizontalGrid
@@ -72,7 +68,7 @@ def test_NOJ_6_turbines_in_row():
 
     site = UniformSite([1], 0.1)
     wake_model = NOJ(site, NibeA0)
-    WS_eff_ilk = wake_model.calc_wt_interaction(x, y, [50] * n_wt, [0.0] * n_wt, 0.0, 11.0)[0]
+    WS_eff_ilk = wake_model.calc_wt_interaction(x, y, [50] * n_wt, [0] * n_wt, 0.0, 11.0)[0]
     np.testing.assert_array_almost_equal(
         WS_eff_ilk[1:, 0, 0], 11 - np.sqrt(np.cumsum(((11 * 2 / 3 * 20**2)**2) / (20 + 8 * np.arange(1, 6))**4)))
 

py_wake/wind_turbines.py

@@ -118,20 +118,15 @@ class WindTurbines():
 
     def _ct_power(self, ws_i, type_i=0):
         ws_i = np.asarray(ws_i)
-        if np.any(type_i != 0):
-            CT = np.zeros_like(ws_i, dtype=np.float)
-            P = np.zeros_like(ws_i, dtype=np.float)
-            type_i = (np.zeros(ws_i.shape[0]) + type_i).astype(int)
-            # TODO: check if faster to calculate both in same line
+        t = np.unique(type_i)  # .astype(int)
+        if len(t) > 1:
+            if type_i.shape != ws_i.shape:
+                type_i = (np.zeros(ws_i.shape[0]) + type_i).astype(int)
             CT = np.array([self.ct_funcs[t](ws) for t, ws in zip(type_i, ws_i)])
             P = np.array([self.power_funcs[t](ws) for t, ws in zip(type_i, ws_i)])
-#             for t in np.unique(type_i).astype(int):
-#                 m = type_i == t
-#                 CT[m] = self.ct_funcs[t](ws_i[m])
-#                 P[m] = self.power_funcs[t](ws_i[m])
             return CT, P
         else:
-            return self.ct_funcs[0](ws_i), self.power_funcs[0](ws_i)
+            return self.ct_funcs[t[0]](ws_i), self.power_funcs[t[0]](ws_i)
 
     def set_gradient_funcs(self, power_grad_funcs, ct_grad_funcs):
         def add_grad(f_lst, df_lst):