In my application I need to tally the boundary crossing particle. With the interesting boundary in a Z-plane I see some very strange behaviours of charged particles, electrons in particular, i.e. some times electrons crossing the boundary at exactly 90 deg angles. An inspection of the PreStepPoint()->GetMomentumDirection() and PostStepPoint()->GetMomentumDirection() revealed that the direction cosines of the electron has been reset to (1,0,0). I understand the direction of a charged particle can be changed at steps limited by boundary and I my application I do see such cases, but for what reason the direction is changed to (1,0,0) I don't know! I suppose it has to be a bug in transportation or in the multiple scattering process.
Let's please use BUGZILLA for the comments !! ---------------- Vladimir Ivantchenko wrote: Hi Fan, 1) I hope that in the last tag cand-02 in any case at the end of thacking the direction is not set to 0 but will be the same as in PreStep; 2) Because of fluctuations of the True Range in msc process and because of fluctuations of energy loss it is possible that Transportation limit the step but particle is stoped because of energy loss - random numbers can provide that. I will do a check as well of the final momenttum direction with the last version of G4. cheers, Vladimir On Fri, 21 Nov 2003, Fan Lei wrote: > > > > > > Please, check the energy in PostStep. Is it different from 0 or not? > > > Hi Vladimir > > The energy is 0. indeed! so the electron is stopped during the step. G4 > assumed it is stopped right on the boundary and for stationary electron G4 > assumes its direction cosine is (1,0,0), am I right? > > When a charged particle loss all its kinetic energy in a step, should the > ionization process define the step rather than the transportation? > > Cheers. > > Fan -------------------- Fan Lei wrote: Hi Michel, Michel Maire wrote: > > could you try the command /process/inactivate msc to see if the > problem persists ? > -- Just tried this, and the problem persists, so it may not related to msc. However, I got quite a lot of warnings: "G4ProcessTable::Insert : arguments are 0 pointer " during the run when msc is deactivated. Regards. Fan PS. Here are some of the screen printout to show the change in direction: ....... Total CPU time used for initialization: 5.14 seconds 90 (0.23596,-0.426848,0.872997) (1,0,0) compt Transportation 90 (-0.154617,-0.790555,0.592551) (1,0,0) compt Transportation 90 (0.249552,-0.568133,0.784187) (1,0,0) eIoni Transportation 90 (0.233442,0.339266,0.911265) (1,0,0) conv Transportation Graphics systems deleted. Visualization Manager deleting... Analyis Manager deleting.. ---------- The numbers are: #1: the angle between poststep particle direction and the boundary surface normal; #2: momentumdirection at PreStepPoint; #3: momentumdirection at PostStepPoint; #4: process created the particle; #5: process defined the current step
More on this .... ----------------------------------------- Henrique Araujo <h.araujo@imperial.ac.uk> wrote: Dear Fan We use 250 eV cuts for LISA inertial sensor and tune msc parameters for backscattering fraction (forced step after boundary). The results I get for Au and Pt agree well with experimental values; I think I would pick up any major problems. The effect you describe is probably very small, but I will check again. Thanks! Henrique On Fri, 21 Nov 2003, Vladimir IVANTCHENKO wrote: > > Hi Fan, > > This problem is exist in any MC. The solution : make smaller cuts, add > step limits - you will have more precise simulation. No other way in > detailed simulation. Of course, one can add a parameterisation for the > detector response... > > cheers, Vladimir > > On Fri, 21 Nov 2003, Fan Lei wrote: > > > > > Vladimir, > > > > OK, I think we understand what is happening here, I trust this > > trivial problem will be corrected in v6.0 release. > > > > On the other hand, the G4 current assumpation of charged particle stopped > > right on the boundary, rather than its real position that is slightly short > > of the boundary, can have some implications in applications where charge > > entering a volumn is collectd, such as the LISA simulations imperial college > > is performing, i.e. this could lead to a slight enhancement. Will it be > > possible for the tracking manager to relocate the PostStepPoint to where > > exactly the ionization process has determined? > > > > Cheers. > > > > Fan
As it is not the transportation that creates this momentum direction, but the energy-loss process, I have reassigned this report accordingly. Best regards, John Apostolakis
In the reference tag geant4-V06-00-ref-01 the problem is fixed: particles which are stopped keep the direction from PreStepPoint
Fixed
