[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: too many cdcTruthHits
Beni,
This is an example of boundary-crossing chatter that sometimes happens
in geant3. For a detailed explanation of why it happens, ask me
sometimes. The main thing is that it was understood by the authors, and
accepted as a consequence of having a small tracking error epsilon,
compared with the scale of the experiment. Our experimental scale is
10m, and our epsil is 10 microns, a span of 1e6. Rounding errors in
single-precision can be on this same order, especially when non-trivial
rotations are involved, as in the CDC stereo layers. There is nothing
wrong with these. It is just that the tracking takes an extra
micro-step in trying to get from one side of a boundary to the other.
I agree that it is undesirable that extra truth points be generated by
this tracking chatter. I will circulate a fix later today that will
eliminate extra truth points caused by this effect. When you see it
checked in, please run a test and confirm that the problem is resolved.
Richard Jones
Beni Zihlmann wrote:
> Hi Richard,
>
> Mark found that quite often he sees more cdcTruthHis than CDC layers.
> While I was looking into this I found some weired behavior I do not
> quite understand. Maybe you know immediately whats going on.
> So I every time hdgeant gets into GUSTEP.F I print out the following
> line:
> GUSTEP: "volume name" x , y , z
> Then I call savehits.F regardless of it is sensitive volume or not. In
> savehits.F
> the variable INWVOL is checked and if it is ==1 (enter new volume) I
> print
> the following line:
> Enter volume: "volume name" x, y, z
> and if it is ==2 (leaving volume) I print
> Leaving volume: "volume name" x, y, z
>
>
> So I get the following output I do not understand:
>
> ....
> GUSTEP: CDCO -12.68686 55.57017 149.4882 Leaving
> volume: CDCO -12.68686 55.57017 149.4882 3.4471179E-09
> GUSTEP: LASS -12.68686 55.57017 149.4882 Enter volume:
> LASS -12.68686 55.57017 149.4882 3.4471179E-09
> GUSTEP: LASS -13.48804 56.87979 151.7202 GUSTEP: LASS
> -14.31226 58.17541 153.9520 GUSTEP: LASS
> -15.15931 59.45658 156.1835 GUSTEP: LASS
> -16.02898 60.72287 158.4149 GUSTEP: LASS
> -16.92102 61.97383 160.6460 GUSTEP: LASS
> -17.39989 62.62656 161.8214 Leaving volume: LASS
> -17.39989 62.62656 161.8214 3.9509893E-09
> GUSTEP: LASS -17.39989 62.62656 161.8214 Enter volume:
> LASS -17.39989 62.62656 161.8214 3.9509893E-09
> GUSTEP: LASS -17.40071 62.62768 161.8234 Leaving
> volume: LASS -17.40071 62.62768 161.8234 3.9510719E-09
> GUSTEP: BM01 -17.40071 62.62768 161.8234 Enter volume:
> BM01 -17.40071 62.62768 161.8234 3.9510719E-09
> GUSTEP: BM01 -18.67320 64.34534 164.9268 Leaving
> volume: BM01 -18.67320 64.34534 164.9268 4.0778918E-09
> GUSTEP: BM02 -18.67320 64.34534 164.9268 Enter volume:
> BM02 -18.67320 64.34534 164.9268 4.0778918E-09
> GUSTEP: BM02 -19.94047 66.05596 168.0175 Leaving
> volume: BM02 -19.94047 66.05596 168.0175 4.2041912E-09
> GUSTEP: BM03 -19.94047 66.05596 168.0175 .....
>
> The problem is that hdgeant leave the volume LASS and enters it again
> right away.
> From the coordinates I gather that this happens close to the border of
> the volume.
> Do you have any explanation why this happens? This happens also quite
> frequent
> in the CDC straws where Mark found it in the firs place causing
> multiple cdcTruthHits.
>
> cheers,
> Beni
>
>
>
>
>
>
>