Some pro's and con's for single particle acceptance calculations - W. Brooks
The following are disadvantages
I see to using single particle acceptance (SPA)
calculations:
1) It does not conform to the stated mission of our group,
which is to produce
publication quality acceptances. Neither I nor anyone
I've talked to believes this
approach will be good enough to produce publishable results,
unless we invest a great
deal of time studying all the possible effects such as
the following ones (2). If
you do all these studies, you might as well do it the
right way from the beginning.
2) Multiple particle interactions with detectors are not
considered in SPA, yet all
of the detectors are vulnurable to them. Two nearby tracks
in the drift chamber,
double hits on the TOF counters, and multiple nearby
hits on calorimeter strips are
examples, not to mention the start counter (!). It will
require a lot of study to
understand how the reconstruction code responds to such
events, and to determine
what fraction of the data is affected, which will depend
on which corner of phase
space you consider.
3) To calculate the acceptance, we have to reconstruct
the event somehow.
We do not have a reconstruction code which will reconstruct
events with an
arbitrary single particle in them. Therefore we will
have to invent a scheme to fool
the reconstruction code into accepting the events. A
significant modification to the
reconstruction code for this purpose is out of the question,
given the very limited
manpower in our group.
A possible scheme for this is somehow to
merge, e.g., an electron into the event,
then reconstruct the overall event. Alternately we could
merge a start counter signal
into the event, but this may result in a different acceptance
for some particles.
Assuming we could fool the reconstruction code into reconstructing
the event, it would
be easiest to do all the SPA calculations using only
one sector. However, if we do
this, it would preclude using the dead channel map to
include the effect of the
holes in the drift chambers. I guess I would propose
to ignore the dead channels,
estimate their effect separately, and enlarge the error
estimate for the acceptance
accordingly.
4) It is not at all clear how to do the acceptance calculation
for decaying particles.
Our reconstruction codes are too primitive to handle
most decays in flight. The
acceptance calculation will depend dramatically on the
status of the reconstruction
code, rather than being dominated by geometry. Therefore
making a SPA table is not
a very well-defined task yet for these particles.
5) Unlike a full simulation, using the SPA will not help
you in the least to model
your trigger, which you will certainly need to do for
the photon data. If we spent
our group's time doing a full simulation, you could use
the data to model your trigger
and test the model.
6) Although the idea behind SPA was to have something
useful for both photon and
electron beam data, in reality there are some differences
which have to be discussed.
For instance, will it be necessary to make bins in the
photon target for events which
occur off the beam axis (due to the transverse extent
of the beam), or will it be
sufficient to generate events on beam axis? Should the
start counter acceptance be
folded in to the SPA, or is that unimportant for the
photon data?
-----------------------------------------------------------------------------------
The advantages I see to SPA include:
1) A modest savings in overall cpu time. We will be forced
to generate over a wide
range of kinematics for each particle, extending into
regions which are unphysical
or very unlikely, unlike doing a full simulation. We
have to do this for all particles
of interest, over bins in theta, phi, momentum, target
position, target radius, etc.
The savings comes in that we may be able to use the same
tables for electron beams
and photon beams, and that we do not have to do a separate
simulation for each beam
energy. We still have to do one for each magnetic field
setting.
2) It will produce moderate quality acceptance calculations
fairly quickly for
multiple groups. While people use these, we can be accumulating
statistics in full
simulations over the next few months.
3) It facilitates trying different event generators for
rapid tests. E.g., to test
the effect of using phase space event generators compared
to model-dependent event
generators.
4) We can do some calculations now, while needed improvements
are being made to the
event generators (CELEG, aao, etc.) So as soon as we
make a plan, we can get started
right away.
------------------------------------------------------------------------------------
I propose the following plan:
The GSIM Focus Group generates events over a grid of theta,
phi, momentum, target
position, and particle type, in one sector; then we will
run these events through
GSIM, producing a BOS output file for two different magnetic
field settings.
Someone from the groups interested in SPA's can then
use this BOS file to produce the single particle
acceptance tables in a format which is useful for people
in the CLAS collaboration.
(Laurent is one possibility, since he is technically
associated with our group; or
another person might be interested, such as Thierry.)
Our group could help this
person with the task, as required.