Correctly carry over likelihood profile errors.

The old code did not carry over correctly the likelihood profile errors for the source component but simply kept the original curvature errors. This gave the impression that curvature errors were similar to likelihood profile errors. This has been fixed now.

Correctly carry over likelihood profile errors.
The old code did not carry over correctly the likelihood profile errors for the source component but simply kept the original curvature errors. This gave the impression that curvature errors were similar to likelihood profile errors. This has been fixed now.
Jürgen Knödlseder
1 parent 9ec6a83a
Showing 1 changed file with 4 additions and 1 deletions
scripts/cspull.py
@@ -305,13 +305,16 @@ class cspull(ctools.cscript):
  
         # Fit model
         if self.m_profile:
-            models = obs.models()
+            models     = obs.models()
+            new_models = gammalib.GModels()
             for i in range(models.size()):
                 model_name = models[i].name()
                 like       = obsutils.cterror(obs, model_name, \
                                               log=self.m_log, \
                                               debug=self.m_debug,
                                               chatter=self.m_chatter)
+                new_models.append(like.obs().models()[i])
+            like.obs().models(new_models)
         else:
             like = obsutils.fit(obs, edisp=self.m_edisp, \
                                 log=self.m_log, \