Hello all.
Just a brief e-mail-update from me before the weekend. Following Peter's
cleanup of the clustering/tracking-code I have redone my
efficiency-estimates and as hinted at in a previous mail the results are
now very good. I remind you that these numbers are UPPER LIMITS because
all I am testing are the trackers and not the clusterfinder (see my note
for details on the algorithm).
Please also note that this e-mail is not a polished analysis-note - it is
just a report on my progress in the spirit of keeping communications
running ;-) All the details will be filled in in a later, proper note when
more work has been done.
The results (see below) mainly show that when all events (background,
collisions, central, non-central) are included, we have the following
efficiencies for reconstruction a single introduced 4 GeV pion:
MRS angle 90 deg 40 deg
TrackFollowFinder 99.8% 99.9%
TrackStringFinder 99.3% 98.2%
The tables below show the number of events I have used, plus a breakdown
into three categories based on SumADC from Tiles ~= centrality (?). The
conclution is that the eff. for TSF goes down a few percent at high
multipicities whereas that for TFF doesn't seem to sink at all.
Before the numbers I also give you a brief "conclusion": TFF now seems to
work better than TSF, which is brilliant since it uses roughly 10% of the
CPU-time of TSF. The low efficincy (~80%) quoted in my last note was due
to the algorithm giving up at high multiplicities, and this is now
fixed. Given the numbers above and below, I'd say that we now have
tracking on finished clusters under control in TPM1, and that the things
which need further testing is weather or not we can reproduce this
when adding a geant-track at the sequence level, and seeing if we
consistently produce the same tracks with and without the added pion.
Have a nice weekend, and here are the raw numbers in a slightly unreadable
format: (ok event = event where exactly one reconstructed track overlaps
with the geant-track to more than 80%)
Efficiency-estimates @ 90 and 40 degrees
Adc-categories (from Tiles):
C1: 0 < SumAdc <= 300000
C2: 300000 < SumAdc <= 600000
C3: 600000 < SumAdc
*******************************************
Run 2336, Seq's 0,1,2,3
MRS @ 90 degrees
Total number of events used: 34635
Number of events C1: 30495
Number of events C2: 3299
Number of events C3: 832
TrackStringFinder Ratio
-----------------
Number of ok events: 34407 99.3%
Number of errors: 0
Number of ok events C1: 30333 99.5%
Number of ok events C2: 3256 98.7%
Number of ok events C3: 818 98.3%
TrackFollowFinder Ratio
-----------------
Number of ok events: 34558 99.8%
Number of errors: 0
Number of ok events C1: 30437 99.9%
Number of ok events C2: 3290 99.7%
Number of ok events C3: 831 99.9%
******************************************
Run 2513, Seq's 0,1,2,3
MRS @ 40 degrees
Total number of events used: 15097
Number of events C1: 9954
Number of events C2: 4088
Number of events C3: 1055
TrackStringFinder Ratio
-----------------
Number of ok events: 14823 98.2%
Number of errors: 0
Number of ok events C1: 9813 98.5%
Number of ok events C2: 3991 97.6%
Number of ok events C3: 1019 96.6%
TrackFollowFinder Ratio
-----------------
Number of ok events: 15086 99.9%
Number of errors: 0
Number of ok events C1: 9949 99.9%
Number of ok events C2: 4084 99.9%
Number of ok events C3: 1053 99.8%
------------------------------------------------
Bjorn H. Samset
Master-student in Heavy Ion physics
Mob: +47 92 05 19 98 Office: +47 22 85 77 62
Adr: Kri 2A709 Sognsveien 218 0864 Oslo
This archive was generated by hypermail 2b29 : Fri Oct 06 2000 - 10:11:00 EDT