Optimise the computation of the fluxes. 1) consider only the SED section...

Optimise the computation of the fluxes. 1) consider only the SED section corresponding to the filter when computing the best grid. 2) As filters are already normalised, there is no need to divide my its integral to get the flux.

pcigale/sed/__init__.py

@@ -330,16 +330,10 @@ class SED(object):
             # We regrid both spectrum and filter to the best wavelength grid
             # to avoid interpolating a high wavelength density curve to a low
             # density one. Also, we limit the work wavelength domain to the
-            # filter one, taking care the presence of λmin and λman in the
-            # used wavelength grid.
-            wavelength_r = utils.best_grid(wavelength, transmission[0])
-            if lambda_min not in wavelength_r:
-                wavelength_r.append(lambda_min)
-            if lambda_max not in wavelength_r:
-                wavelength_r.append(lambda_max)
-            wavelength_r.sort()
-            wavelength_r = wavelength_r[wavelength_r <= lambda_max]
-            wavelength_r = wavelength_r[wavelength_r >= lambda_min]
+            # filter one.
+
+            w = np.where((wavelength >= lambda_min)&(wavelength <= lambda_max))
+            wavelength_r = utils.best_grid(wavelength[w], transmission[0])
 
             l_lambda_r = np.interp(wavelength_r, wavelength, l_lambda)
             transmission_r = np.interp(wavelength_r, transmission[0],
@@ -350,19 +344,15 @@ class SED(object):
             else:
                 dist = 10. * parsec
 
-            # TODO: Can we avoid to normalise as the filter transmission is
-            # already normalised?
             f_lambda = utils.luminosity_to_flux(
-                (np.trapz(transmission_r * l_lambda_r, wavelength_r) /
-                 np.trapz(transmission_r, wavelength_r)),
-                dist
-            )
+                np.trapz(transmission_r * l_lambda_r, wavelength_r),
+                dist)
 
-            # Fν in W/m²/Hz. The 1.e-9 factor is because λ is in nm.
-            f_nu = lambda_eff * f_lambda * lambda_eff * 1.e-9 / c
+            # Fν in W/m²/Hz. The 1e-9 factor is because λ is in nm.
+            f_nu = lambda_eff * f_lambda * lambda_eff * 1e-9 / c
 
             # Conversion from W/m²/Hz to mJy
-            f_nu *= 1.e+29
+            f_nu *= 1e+29
 
         return f_nu