Acceptances of vector mesons
Mass Resolutions for different configurations
Pattern Recognition Studies
Acceptance
Acceptance of vector mesons in each muon arm: Below are listed
the number of vector mesons that would be accepted into the
muon arms in 1 RHIC year (3000 hours) at blue-book luminosity
running (2x10**26 cm-2s-1).
| Meson |
North |
South |
North+South |
| phi |
4x10**4 |
1x10**4 |
. |
| J/psi |
6.7x10**5 |
6.4x10**5 |
. |
| psi' |
1.2x10**4 |
1.1x10**4 |
. |
| Upsilon |
382 |
331 |
293 |
For phis, the North arm has a significantly higher acceptance,
and a somewhat better signal:background because of a somewhat
better mass resolution. The south arm also appeared to get
somewhat more dimuons per event, presumably because the backplate
of the magnet is 10 cm shorter (less absorber material before
muon identifier), but different muID cuts might take care of this
difference.
For higher mass vector mesons, the acceptance difference between
North and South arms is quite small. And although the identified
muons may or may not be larger in the south arm, the occupancy
in the muon tracking chambers is better in the south arm because
of the smaller rapidity acceptance.
I looked at acceptance of vector mesons for all 8 octants compared to
every other octant and 4 contiguous octants. Below are the
acceptances relative to the all-8-octants scenario:
| Meson |
Full Acc. |
Every Other Oct. |
4 Contiguous |
| upsilon |
1.0 |
0.42 |
0.03 |
| J/psi |
1.0 |
0.25 |
0.09 |
| phi |
1.0 |
0.08 |
0.06 |
Shown below are plots of the pt and rapidity of the accepted
vector mesons for the above configurations.
Rapidity of accepted phis.
Ratio of phi rapidity spectra from 4-octant configurations to
8-octant configuration.
Pt of accepted phis.
Rapidity of accepted J/psis.
Ratio of accepted J/psis rapidity spectra in 4 octant configurations to
8 octant configuration.
Pt of accepted J/psis.
Ratio of accepted J/psis Pt spectra in 4 octant configurations to 8 octant
configuration.
Rapidity of accepted upsilons.
Ratio of accepted upsilon rapidity spectra in 4 octant configurations to 8 octant
configuration.
Pt of accepted upsilons.
Ratio of pt spectra from 4-octant configurations to 8-octant
configuration.
Mass Resolutions
Mass resolutions of vector mesons for various muon tracking
arm configurations. Unless stated otherwise, all configurations
have anode wires present. A configuration like 1-2-3 implies
1 gap at station 1, 2 gaps at station 2 and 3 gaps at station
3. Fit errors returned by ROOT are in () after the resolution.
Note that J/psis, especially have larger error bars partially
because a gaussian fit is being used on a non-gaussian peak.
| Configuration |
Upsilon, S |
Upsilon, N |
Psi, S |
Psi, N |
| baseline |
240 |
190 |
115 (20) |
106 (17) |
| baseline, no anodes |
. |
. |
114 |
105 |
| 2-2-2 |
270 |
194 |
124 (14) |
107 (10) |
| 2-2-2, no anodes |
265 |
. |
103 (11) |
120 (24) |
| 2-2-1 |
. |
. |
116 (9) |
118 (12) |
| 1-2-1 |
290 |
210 |
124 (25) |
130 (18) |
| 2-1-1 |
. |
200 |
. |
. |
| 1-1-1 |
330 (60) |
. |
160 (50) |
126 (25) |
Pattern Recognition
***Note: I found a problem in the software that was counting Monte Carlo
tracks that caused tracks to be lost sometimes and more often in higher
occupancy areas. So, I decided to rerun the pattern recognition studies
with the code again after I fixed this bug, and I am in the process of
updating the information below to reflect the new results. I will
put a line of *** below the area that I have updated.***
I did some quick studies of pattern recognition with and without anodes
and with 3 and 2 gaps/station. I did not run a lot of tracks because
of time constraints and a limited number of HIJING files on disk--
I looked at 30 events in each arm which give HIJING tracks plus
signals mixed in each event. I ran the baseline configuration which
is 3 gaps at stations 1 and 2 and 2 gaps at station 3 and anodes in
all gaps, and verified that I still got reasonable efficiency with
the current pattern recognition parameters. Mixing J/psis with
HIJING events and having 100% chamber efficiency gave a track finding
efficiency in both the North and South arms of:
Baseline Efficiency = 86%.
I believe this is a
lower limit (in this environment) on the efficiency since work still
needs to be done tuning the cut parameters that I have added and the
code is still evolving. However, this efficiency does not take into
account chamber inefficiencies, electronics, higher occupancies, which
I will try to address further down in this note.
I then left the pattern recognition cut parameters the same and
ran the baseline without anodes and saw a decrease in efficiency to
~60%. I then widened some pattern recognition parameters (primarily
changes in cuts on station 3 and station 2-3 stubs were required) and
was able to bring the efficiency up close to the baseline efficiency.
Baseline Without Anodes Efficiency (wider cuts) = 82%.
********************************************************************
To get a gross idea of how efficiency might be affected when
electronics and chamber inefficiencies are taken into account,
I then ran the Baseline configuration and Baseline Without Anodes
with only 2 gaps in each station. I left the pattern recognition
parameters the same even though they would obviously have
to be adjusted for the poorer projection errors that you would get
when you have fewer points in a station stub. However, I think
this is still informative, because it indicates whether the
pattern recognition parameters
would have to be changed significantly to pick up all tracks
or not. What I found was that the Baseline Configuration with
2 gaps/station had a reduced efficiency (as expected when keeping
the patt. rec. parameters the same) and it was more reduced in
the North arm than the South Arm. This is probably a combination
of the fact that the North arm has the higher occupancy and (maybe
more important) the fact that the North arm stations are further
apart so projection errors would likely be larger when the station
vector resolutions degrade. When I ran the Baseline Without Anodes
and 2 gaps, I found a substantially larger reduction in the
efficiency. How much of these efficiencies can be gotten back
I'm not sure yet, but I will try to make some more runs with
windows opened up and see if more tracks are found and/or if the
pattern recognition becomes swamped with potential tracks.
The efficiencies I saw were:
Baseline With 2 Gaps per Station Efficiency = 50% in North, 80% in South.
Baseline Without Anodes and With 2 Gaps per Station Eff. = 22% in both arms
By making some changes in the pattern recognition cuts, I was able to improve
the above efficiency (with the same cuts on both runs) to 80% in both North and
South for the 2 Gaps and anodes, and 44% for 2 gaps and no anodes. It did
require opening up cuts on "bad" tracks and projections. The problem
with 2 gaps and no anodes seems to be in the fitting and projecting, especially
of station 2-3 stubs. This may be partly an artifact of the fitting method
but clearly fewer measurements in a station will provide you with a
poorer measurement of the track vector in any case.
I tried to see how the performance changes with and without anodes in
higher occupancy events. To do this I "mixed" _two_ HIJING events with
the J/psi signal events and ran with the baseline configuration and anodes
and with the baseline configuration and no anodes. There were some problems
running in this configuration just because some array bounds/table bounds
were reached in some events, but nevertheless the code did run and
came up with a handful of tracks that should have been found. The result
was that when anodes were present 80% of the tracks were found and when
they were not present 40% were found. I also looked at the track finding with 2 gaps
and anodes and found I could get ~80% efficiency _if_ I increased the
track finding cuts. So, I reran the "no anodes" option
with the increased track finding cuts and the track finding efficiency
went up to 60%. What is the difference between the baseline with no
anodes and the baseline with 2 gaps?: Since the track finding with smaller
numbers of hits (i.e. gaps removed or anode wire readout removed)
seems to run in to the most trouble in the station 2-3 fit, I think the
reason why no anodes gives the poorer performance than anodes and 2 gaps
is because station 3 with no anodes has only 4 measurement points (2 cathodes
in 2 gaps since 2 gaps at station 3 is the baseline) but the 2 gaps with
anodes has 6 measurement points at station 3. This means that 1) the projection
to station 1 to pick up hits is worse and 2) the discrimination power between
good and bad station 2-3 track stubs is not as good. Again, this is just my guess
as to the difference.
Last update 10/16/98 by
Melynda Brooks