NaNs in Cl fix for PolyChord

cobaya/likelihoods/base_classes/cmblikes.py

@@ -578,8 +578,13 @@ def exact_chi_sq(self, C, Chat, L):
                     (np.trace(M) - self.nmaps - np.linalg.slogdet(M)[1]))
 
     def logp(self, **data_params):
-        cls = self.provider.get_Cl(ell_factor=True)
-        return self.log_likelihood(cls, **data_params)
+        Cls = self.provider.get_Cl(ell_factor=True)
+        for Cl_key, Cl_array in Cls.items():
+            if Cl_key != 'ell':
+                if np.any(np.isnan(Cl_array)):
+                    self.log.error("nans in Cls['%s']: returning logzero and carrying on." % Cl_key)
+                    return -np.inf
+        return self.log_likelihood(Cls, **data_params)
 
     # noinspection PyUnboundLocalVariable
     def log_likelihood(self, dls, **data_params):

cobaya/likelihoods/base_classes/planck_2018_CamSpec_python.py

@@ -270,6 +270,11 @@ def chi_squared(self, CT, CTE, CEE, data_params):
 
     def logp(self, **data_params):
         Cls = self.provider.get_Cl(ell_factor=True)
+        for Cl_key, Cl_array in Cls.items():
+            if Cl_key != 'ell':
+                if np.any(np.isnan(Cl_array)):
+                    self.log.error("nans in Cls['%s']: returning logzero and carrying on." % Cl_key)
+                    return -np.inf
         return -0.5 * self.chi_squared(Cls.get('tt'), Cls.get('te'), Cls.get('ee'),
                                        data_params)
 

cobaya/likelihoods/base_classes/planck_clik.py

@@ -115,8 +115,13 @@ def get_requirements(self):
 
     def logp(self, **params_values):
         # get Cl's from the theory code
-        cl = self.provider.get_Cl(units="FIRASmuK2")
-        return self.log_likelihood(cl, **params_values)
+        Cls = self.provider.get_Cl(units="FIRASmuK2")
+        for Cl_key, Cl_array in Cls.items():
+            if Cl_key != 'ell':
+                if np.any(np.isnan(Cl_array)):
+                    self.log.error("nans in Cls['%s']: returning logzero and carrying on." % Cl_key)
+                    return -np.inf
+        return self.log_likelihood(Cls, **params_values)
 
     def log_likelihood(self, cl, **params_values):
         # fill with Cl's

cobaya/likelihoods/base_classes/planck_pliklite.py

@@ -157,5 +157,10 @@ def chi_squared(self, c_l_arr, calPlanck=1):
 
     def logp(self, **data_params):
         Cls = self.provider.get_Cl(ell_factor=True)
+        for Cl_key, Cl_array in Cls.items():
+            if Cl_key != 'ell':
+                if np.any(np.isnan(Cl_array)):
+                    self.log.error("nans in Cls['%s']: returning logzero and carrying on." % Cl_key)
+                    return -np.inf
         return -0.5 * self.get_chi_squared(0, Cls.get('tt'), Cls.get('te'), Cls.get('ee'),
                                            data_params[self.calibration_param])