EMC Notes for LEVEL 3 trigger studies

There are a few basic questions which come up in both p-AU for
gluon structure in Nuclei, and in pp for normalizing physics and
spin physics. 

These mainly have to do with how TPC data can be handled, both in
the 24 sector crates,and in higher level 3.

Most  of these questions are related to the fact that we can't simply
write out all the TPC data to tape at over 100 Hz.

1) The EMC data must be corrected for the energy deposited by charged
tracks.  This can be done crudely over large areas with multiplicity counts
from CTB. In order to do triggering utilizing  localized areas of energy
deposit shch as high order multipole analysis or jets, it should be
done on the basis of track information.  In AuAu this can probably be
done offline, but for p-Au and pp the rate reduction needed in level
3 trigger requires that this be done in level 3 to get good results
such as sharp trigger thresholds..
  
2) We need enough momentum resolution on all tracks from one collision
event to reconstruct jets to adaquate resolution for triggering in level 3
and better than that for offline reconstruction.
The momentum resolution required is something like 20% rms independent of
energy up to 50 GeV/c in level 3, and maybe 15% offline.
As far as I know, this can be done offline for Au-Au, but not for
p-Au and pp.
The level 3 requirements are both softer in p-Au than in pp and 
easier to deal with because of fewer events stacked up in the TPC at one time.
(see appendicies).
The offline requirements are the same in p-Au as in  pp.

3) In level 3 trigger we need to deal with tracks that cross TPC sector 
boundaries because of the magnetic field.  We also have to get the
data from these tracks onto the tape for offline.  
Again, this is to get the jet rate down low enough for taping.
(see appendicies for reasons for jets, comparison of p-Au and pp, etc)
Question: How well do we need to deal with these tracks?
My quick calculations indicated 30% of all 300 MeV tracks were lost
and there was a very significant loss as high as 10 GeV/c tracks.
if nothing is done.  (I lost my notebook with the calculation
but this should be Monte-Carloed anyway.)

4) Level 3 trigger should select a relevant subset of tracking information
from each pp event at high Luminosity such that 100 HZ of events with 
EMC, TPC tracks, etc  go to tape.
The relevant tracks can probably be tagged as hits in EMC towers,
and the tags used as selection criteria in level 3, but the real 
problem is maintaining the momentum resolution needed in level 3 and offline.

Appendix on assumptions:
A)   We assume that in Au-Au the rate is low enough that all the EMC data
and all the TPC track information can be read out to tape and is of
adaquate quality. 
(This in turn assumes that multiplicities are more or less as predicted,
and that the momentum resolution plots I have seen of 12% at 40 GeV
are right.)
B) We assume that all the EMC data goes to each of the 24 sector crates
such that tagging of tracks across sector boundaries is possible.
C) We assume that if p-Au is run without the pp luminosity upgrade in place,
then the nucleon-nucleon luminosity is about the same for Au-Au,
p-Au and pp. This gives regular luminosity of 10^26 for Au-Au,
10^28 for p-Au, and 10^30 for pp.    
D) We assume that we need  resolution in the reconstruction of jet Pt
of roughly 15%  or 20% rms independent of energy.
Our simulations which combine EMC and Tracking  do roughly  this.
The resoution is dominated by tracking at the relevant energies, 
since momentum resolution  gets worse at higher Pt while EMC gets better 
at higher pt.
E) We assume that in p-Au we do gamma plus jet to look at the 
gluon structure function in Nuclei, and we do jet-jet to check
our normalization for quarks and to learn about  shadowing vs 
jet quenching based on p-Au vs Au-Au.
The level 0 trigger rates are a factor of at least 20 and maybe 100
higher than the physics rates of interest.  The analysis in level 3 
is the means of getting the rate down to a sensible level to tape.
This involves isolation cuts around gamma candidates, based on tracking,
as well as jet reconstruction with EMC and Tracking.
F) We assume from Monte-Carlo that to trigger on jets of a given Pt,
we must have an EMC threshold significantly lower. This is because the
probability of pi-zero of a given fraction of the jet momentum
goes as roughly exp(-6z) where z is the fraction.
 This in turn gives a level 0 jet trigger rate about a factor of
20 higher than the full jet rate at the Pt of interest.


Some recent EMC Requirements additions that may be of interest:

Requirement:
     TPC tracks are required for  almost all EMC events to be used in   
physics analysis.  The efficiency and momentum resolution of 
tracking must be adaquate to
a) reconstruct jets soon enough to use in triggering before
tape writing with resolution adaquate to get theoretical trigger
rates within a factor of 2 and without losing rare events,
(approximately 20% resolution on leading particle up to 50 GeV,
assuming that enough TPC information goes onto tape to get
about 15% resolution up to 50 GeV offline.0
(losses of low mom. tracks less than 20% of total jet energy)
b) reconstruct jets offline with resolution not significantly worse
than they would be reconstructed with ideal TPC + EMC.
c) reconstruct electrons of 40 GeV/c offline with better than 12%
momentum resolution, 
d) reconstruct electrons in the trigger with sufficient resolution
to make E(emc) vs P(tpc) cuts to get the trigger rates down to
acceptable levels.
(in pp , we need data  down to  12 GeV Pt ).
Justification:
   Tracks are needed with EMC data to trigger and measure jets.
Tracks are needed with EMC data to make energy-based isolation cuts
on direct photon and electron candidates.  

Requirement:
     EMC requires knowlwdge of the beam centroid position at the time 
of each event to within 0.3 mm.
Justification:
       A momentum resolution for TPC tracks of better than 12% 
at 40 GeV (rms) is needed in pp running. This is for electrons from W and Z.
There is a similar  requirement for leading particles in jets
in pp and pA running.  Also, full acceptance form 80% of the interaction
diamond is needed at eta from -1 to +2.  The SVT may not be useful for
every event in this case.


Requirement:
EMC requires calculation of the correlation between EMC phototubes
and Shower Max information within the level1-level2 time frame.
These correlations cross EMC phi boundaries and TPC sector boundaries
by one EMC nearest neighbor.
Justification:
     The correleation between EMC and Shower max should be done as 
soon as possible because the rate of events into Level 3 will
be too large to handle if it is not done.
Implementation:
      We specifically reject
a) direct gamma candidate events which have an obvious pi-zero
in the tower,
b) events  which do not pass a chi-square cut as being direct
gamma or electron,
c) events in which the direct gamma candidate is too close to
boundaries (outside fiducial cuts)
d) events which have a "splash" of hits in multiple areas
of a single shower max box and are not reconstructable offline.

Requirement:
     EMC shall put the pedestal-subtracted phototube data for
events accepted by Level 1  into a dual-port memory for use by Level 2.
Justification:
     Level 2 needs the EMC data for calculation, but does not have the
bandwidth into DAQ to pass all the data to DAQ.

Requirement:
     EMC shall supply a few bits to Level 2 indicating the quality
of the event asfound in EMC Level 1.  
Justification:
     EMC events may be needed  by triggers that are not initiated
by EMC.

Requirement,
     Upon a Level2 accept, EMC shall supply the EMC data to 
24 TPC Sector Crates for correleation with TPC tracks and/or
passing to DAQ .
Justification:  
     The data path for EMC to DAQ is through the 24 sector crates.
The amount of processing to do the correlation with tracks is
large and can best be handled by multiple processors 
operating on independent parts of the TPC data.

Requirement:
     EMC shall make correlations with CTB/TOF at two stages of the
trigger. a) in Level 0 using hit counts from CTB slats and 
EMC isospin patches (approximately 1 unit in eta by 1 unit in phi)
b) in late Level 1 or in level 2 using analog multiplicity from
CTB with EMC PMT data  in programmable patch sizes.
Justification:
    The best physics calculations  possible at each trigger stage are made 
in order to reduce the rate into the next stage of the trigger.

Requirement:
     It shall be possible to process built events to the
following extent before writing to tape:
a) Corrections to EMC to get neutral energy when there are
 charged tracks hitting the calorimeter must be made.
This involves both tracks within a sector and crossing sectors. 
b) Energy based isolation cuts around direct gamma candidates
must be made using both neutral energy and charged tracks.
This involves both tracks within a sector and crossing sectors. 
c) Jet energies must be calculated using neutral energy  plus TPC tracks.
This involves both tracks within a sector and crossing sectors. 
Justification:
    The bandwidth to tape may be exceeded if we are not sufficiently
selective in choosing quality events.  We cannot do the physics
if a high percentage of good events are lost due to the bandwith
to tape being saturated by non-useful events.
This is most critical in high Luminosity pp running, but may apply
in other cases as well.