Hi Chris, I think I am debugging the same problem as you are but on top of
v3. So far I found one problem related to the confusion in phi convention.
When TBLRecon::OneSeg is envoked to constructed "fake segs" from 1 real
segment and a tof hit. A check of phi_seg - phi_tof was made and only
combinations with phi_seg-phi_tof small are picked. However, I found one
case that: phi_seg = 178 while phi_tof = -182. Program failed to recognize
this. So I guess many times, right sector does not have the help from
OneSeg(). This could cause a problem in efficiency. The fix is a no
brainer.
Comments on the events you listed:
3070-12, very messy on left sector. looking at the segments, it seems to
be a ++ event. not sure if it is worthy to get it back. gets a lot of fast
fit, but no good newton.
3070-14, positive charge on the left. looks fine in nsed with v3 top. do
not know what problem you are refering to.
3070-19, X-mas tree on the right. no segments on the right reconed anyway
due to empty outer chamber. positive charge on the left very clean though.
Chi
On Mon, 16 Jun 2003, Chris Crawford wrote:
> hi,
> i did a very small comparison of v15, v18, and head in nsed, just
> events that were bad in the head version: 3070:12,14,19 .
> results: v2_15 got all of them, and v2_18 got 2 of them v2_head:
> none of them
>
> i'm still working on why the newest version chokes up; one thing i
> suspect is that the new cuts on chi^2 are invalid since the bug in chi^2
> was fixed. i don't think it's worth tagging v2_20 until we figure out
> what's going on.
>
> but this proves an important point, that we really need to work on
> debugging what we have before we add new features. and also, we need to
> make sure the code works 'better-than-before' before submitting changes.
>
> hats off to chi for finding a couple of important bugs (like factor of
> 1000 in t2d b-field).
> --chris
>
>
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:29 EST