The first four paragraphs summarize this note. The rest is for
gluttons.
We are converging on a long-term configuration for the WFD and monopole
triggering systems, but before finalizing anything we want a consensus the
three major remaining issues: the final monopole search configuration, the
remaining triggers to be added to the system, and, of course, how to treat all
the new data.
We have suggestions for solutions to all these problems, but include
two caveats: (1) only the collaboration can decide on critical issues regarding
detector acceptance, which is the reason that the spokesmen and the most
affected groups are being notified of what we plan; and (2) we welcome
criticism, but unless we are supplied with viable alternatives the hardware
will soon be configured in the "default" state outlined below.
Essentially, we plan to (1) open the FMT time window to overlap with
the ERP, using the ERP/FMT/SMT to provide full beta coverage in the WFD;
(2) read out a 1 ms time window for all triggers, archive one "full" copy, and
strip all but a narrow muon window from the "default" data, and (3) run in
final configuration with SMT, multiple-box ERP, HIPT, FMT, and LIP WFD readout
(only). We still expect to wrap these things up by the nominal 1 August start
date.
Those of you who agree with this plan and who are not concerned with
technical details can stop reading now; as for the rest, remember that the WFD
represents the technical heart of the scintillator rare particle search (it's
not "ACRO"!) and we will need some time to explain how everything works. We
have worked hard to focus on issues, but there's a fair amount of pretty dense
information here.
I. FINAL MONOPOLE SEARCH CONFIGURATION
There are now a total of four "basic" monopole triggers that call WFD
readout (ERP, HIPT, FMT, and SMT), as well as the LIP:
Trigger # Faces Time Window Physics
ERP 1 0- 6 us Mu/Fast Monpole
FMT 2 1- 10 us Fast Monpopole
HIPT 2 0- 10 us (at 10x FMT thresh) Highly Ionizing
LIP 3 0-250 ns Lightly Ionizing
(CSPAM) (2) (0- 1 us) Mu (no WFD now)
The CSPAM trigger identifies scintillator panels rather than individual
tanks, and so would in principal force twice as much WFD readout as the ERP.
For this reason, and becaue the ERP provides a great deal of overlap, we have
provisionally decided to leave the CSPAM out of the default WFD readout set.
The HIPT and the ERP/FMT also overlap, but because of the limited
amount of data added by the HIPT (it has a wuite low trigger rate) and because
of the great interest in the physics involved, we did include it in the default
set. We expect to circulate hard numbers on how much the HIPT adds to the
data stream in the next few days, but preliminary estimates are in the range of
10%, +/- a factor of two.
II. NEW TRIGGERS
Although we are very near the final default configuration of the WFD
system, there are a few points to face before completion: the attico slow
monopole trigger (SMT), the streamer tube slow monopole trigger, and changes in
the LIP trigger.
(1) The addition of the attico SMT could increase the total data flux
by as much as a factor of 50-60%. This appears to be unavoidable, however, and
we will add the attico SMT as soon as it becomes available.
(2) We expect to implement WFD readout for the streamer tube slow
monopole trigger as soon as the readout software is available. It is not
expected that this will add significantly to the data stream as long as (a)
the number of tanks to be read out on each trigger is reasonably small; and
(b) either the total trigger rate is not large compared to the muon rate, or
the trigger identifies primarily only tanks which would be read out on muon
events anyway. One caution is in order, however: as the WFD team will begin to
leave the LNGS on 14 July, we are on a very tight schedule for getting slow
monopole WFD readout ready for the nominal 1 August start date, and of course
we can do nothing until the code is finalized.
(3) Additional LIP triggers: the week the inter-SM LIP trigger will be
installed, the present extremely conservative streamer tube track requirement
will be somewhat loosened, and the scintillator coincidence time window will be
slightly increased. Because the LIP requires a three-plane scintillator
coincidence, however, none of these changes are expected to add significantly
to the total data rate.
III. DATA DISTRIBUTION
The excellent resolution provided by the new WFD system has one major
drawback: it writes about three times as much data as all the other detector
elements put together, and it is likely to become even more generous when the
attico SMT is added. So while we all understand that WFD data are critically
important to meeting the design goal of the experiment, we also understand that
we must do everything possible to insure that we have due cause for pushing
the data distribution and processing systems as hard as we do. Here are some
ideas, of which the last has by far the biggest potential impact on the data:
(1) We can eliminate WFD readout for single-box ERP events. Nominally
single-box ERP events may have less physics content that multiple-box events,
but because they make up so little of the data stream the average gain here
would be only on the order of a few percent. Nevertheless we plan to implement
this change because it also provides some insurance against hot boxes. The
decision can be reversed, of course, given any reasonable physics objection.
(2) It may also be possible to reduce the data stream by limiting the
TOHM/LI acceptance of the SMT, or by reducing the time window for monopole
readout. Because these are physics-based rather then hardware-based decisions,
however, we belive that work on these ideas should progress within the rare
particle group. So we plan to finalize the WFD in parallel, and implement
changes from the rare particle group when and if requested to do so.
(3) Probably the most significant reduction of the data could come
from cutting the nominal 1 ms beta=10^-4 monopole window to 20 us on fast
particle triggers, and we expect to do this. There are three possible methods:
(a) While the current hardware has the ability to limit WFD readout to
an arbitrarily small window for fast particle triggers, this would greatly
reduce detector acceptance for monopole catalyzed proton decay events, in which
monopoles could be preceded by fast particle triggers (due to proton decay
events outside the detector, catalyzed by the monopole on its way in, and
stopping the WFD _before_ the monopole enters any of the scintillator tanks).
Therefore we do not recommend this approach.
(b) It is possible to run the WFD for 1 ms on all triggers (the present
state), but develop a "smart" readout that only keeps the muon section of the
data when no slow monopole trigger occurs. This could reduce the data from
its present level by a factor of almost two. Unfortunately, it is not clear
how much time such a solution would require and it is not clear when anyone with
enough experience on the WFD system would be able to come to the LNGS to
implement it. In addition, testing and debugging would require significant
down time, and detector dead time would likely be permanently increased. In
addition, even a quite clever smart readout system would fail if the catalysis
cross section were so high that proton decays occurred inside the scintillator
itself, in which case formation of the slow monopole trigger could be be
prevented by "washing out" the single photoelectron signals. Although this
plan originally seemed the most obvious, for all of these reasons we cannot
recommend it now.
(c) The remaining possibility is data reduction in the offline system.
This would leave the WFD readout window at 1 ms for all events (the most
conservative physics option). This "full data set" would be written to an
archive, from which any analysis requiring the full WFD data would proceed.
In the "normal" data distribution only the narrow muon WFD window would be
retained, even for events with SMT triggers. This approach has three distinct
advantages: it can be implemented almost immediately, the increase in the
"normal" data size would be moderate (perhaps only a factor of two above the
"pre-WFD" level, or a factor of two less than the present), and the most
conservative possible physics choices are implemented. The major disadvantage
is that the "full" data archive would be quite large, and any long-term slow
monopole analysis would have quite a lot of information to digest. As far as
we understand it, however, this the nature of slow monopole searches and it is
not clear how this can be dramatically changed without affecting sensitivity
(except by windowed readout, which will also affect sensitivity at least
somewhat anyway and also has the disadvantages listed above).
IV. CONCLUSIONS
Essentially everything we outline here can be implemented by people
at the lab before the nominal 1 August startup date. There are three obvious
exceptions: the attico SMT is still under construction, the streamer tube slow
monopole readout code is being produced "off site," and possible work on the
SMT acceptance is independent of the WFD team. Any simple adjustments called
for by these efforts can nominally be performed in a short time, but we should
caution the collaboration that significant changes will be very difficult to
implement if they are not requested essentially immediately: dedicated WFD
personnel will begin leaving the LNGS on 14 July.