fixed event-based issue. Moving particle allocation to once per simulation was causing incorrect answers. The cause was that the particle tally accumulators were not being reset to 0 each iteration. The fix was to reset the particle acccumulation variables at particle death (could instead be done in initialize_history() function, if desired).

.travis.yml

@@ -32,9 +32,9 @@ env:
 matrix:
   include:
     - python: "3.7"
-      env: OMP=y MPI=y PHDF5=y DAGMC=y EVENT_BASED=y
+      env: OMP=y MPI=n PHDF5=n DAGMC=n EVENT_BASED=y
     - python: "3.7"
-      env: OMP=y MPI=y PHDF5=y DAGMC=y EVENT_BASED=n
+      env: OMP=y MPI=n PHDF5=n DAGMC=n EVENT_BASED=n
 notifications:
   webhooks: https://coveralls.io/webhook?repo_token=$COVERALLS_REPO_TOKEN
 install:

src/particle.cpp

@@ -391,6 +391,12 @@ Particle::event_death()
   global_tally_tracklength += tally_tracklength_;
   #pragma omp atomic
   global_tally_leakage += tally_leakage_;
+
+  // Reset particle tallies once accumulated
+  tally_absorption_  = 0.0;
+  tally_collision_   = 0.0;
+  tally_tracklength_ = 0.0;
+  tally_leakage_     = 0.0;
 }