fix: Avoid division by zero error in Local Group mass function analys…

…is if no host halos were found If no host halos meet the criteria for the Local Group mass function analysis then the host halo weight will be zero. In such cases we now report an impossible log-likelihood instead of attempting to normalize by this weight.
galacticusorg · Jul 28, 2021 · 03e574d · 03e574d
1 parent fa11391
commit 03e574d
Showing 1 changed file with 12 additions and 10 deletions.
diff --git a/source/galacticus.output.analyses.Local_Group_mass_functions.F90 b/source/galacticus.output.analyses.Local_Group_mass_functions.F90
@@ -550,17 +550,19 @@ subroutine localGroupMassFunctionFinalizeAnalysis(self)
     ! Normalize the mass function.
     weight           = sum(massFunctionCentrals)
     weightVariance   = sum(covarianceCentrals  )
-    self%massFunction=+self%massFunction/weight
-    self%covariance  =+self%covariance  /weight**2
-    do    i=1_c_size_t,size(self%massFunction)
-       do j=1_c_size_t,size(self%massFunction)
-          self%covariance(i,j)=+self%covariance  (i,j)    &
-               &               +self%massFunction(i  )    &
-               &               *self%massFunction(  j)    &
-               &               *weightVariance            &
-               &               /weight                **2
+    if (weight > 0.0d0) then
+       self%massFunction=+self%massFunction/weight
+       self%covariance  =+self%covariance  /weight**2
+       do    i=1_c_size_t,size(self%massFunction)
+          do j=1_c_size_t,size(self%massFunction)
+             self%covariance(i,j)=+self%covariance  (i,j)    &
+                  &               +self%massFunction(i  )    &
+                  &               *self%massFunction(  j)    &
+                  &               *weightVariance            &
+                  &               /weight                **2
+          end do
        end do
-    end do
+    end if
     return
   end subroutine localGroupMassFunctionFinalizeAnalysis