Created attachment 278 [details] model demonstrating the problem the trajectory filtering algorithm has some dramatic effects for very long traces with many changes: at some point it just drops almost all points. the demonstrating example is attached. consider the following queries: simulate 1 [<=50000] {e, f + 100 } // works fine, albeit trajectory appears as a solid rectangle simulate 1 [<=100000] {e, f + 100 } // no rectangle for f anymore, looks strange Some higher values for trace resolution parameter can postpone the problem, but I think it should still not drop all most of the values.