Subject: Meeting minutes from Nu2002, Munich
From: Kate Scholberg (schol@MIT.EDU)
Date: Mon Jun 10 2002 - 09:59:35 EDT
Note: I have added several people present at the meeting to this SNEWS
working group mailing list. If you do not want to be on the list,
please let me know. Also, if you need the password for the working
group web page and other info, let me know.
Minutes of SNEWS Meeting at Neutrino 2002
=========================================
Attendees: Glenn Horton-Smith, Marek Kowalski, Per Olof Hulth, Marco Selvi,
Clarence Virtue, Chris Waltham, Rex Tayloe, Art McDonald, Alex Murphy,
Carlo Vigorito, Pietro Antonioli, Walter Fulgione, John Beacom, Alessandra
Di Credico, Mark Vagins, Alec Habig
Thanks to the help of Georg Raffelt we were able to meet over lunch
at Neutrino 2002.
Experiment Updates:
-------------------
* Super-K: as everyone knows, SK is currently down but should be back
online by the end of the year. Mark reported that the new PMT density
will cause loss of perhaps 15% of SN events, but sensitivity of the
alert system to Galactic supernovae will be practically unaffected.
* SNO: Clarence reported that the trigger is now set to 35 events above
4 MeV in 2 seconds. Livetime is now effectively better than 90% for
supernovae. Response time should be less than 20 minutes. Background
is well understood. One remaining physical background is spallation
events; these occur about every 2 weeks and can be clearly tagged. NCDs
will be added some time next year (exact schedule still unclear).
* KamLAND: Glenn Horton-Smith reported for Nikolai Tolich who could
not be present. Nikolai has implemented a SN trigger which lowers
the threshold to 0.5 MeV for 3 minutes when a cluster of events is
found. Livetime is now about 90%.
* Borexino: Alessandra DiCredico reported. Filling should
begin in the fall of this year and they should start to run
at the beginning of next year.
* AMANDA: now running with a SN trigger. Soon they should have
24 hour connectivity from Iridium.
* Mini-BooNE: a supernova trigger should be possible at approximately
12 MeV threshold. A first pass at a supernova trigger was made; rates
are a little higher than expected but they are in the very early
stages of getting things running. They have about 15% dead time from
the muon veto.
* OMNIS: (post-meeting update from Alex Murphy, now at Edinburgh)
Current plans center on 2 kT of lead with scintillator layers, plus a
1/2 kt of lead perchlorate. Alex is also looking into deploying an
OMNIS-like detector to help with cosmic ray rejection for dark matter
experiments at Boulby.
SNEWS Code Status:
-----------------
Version 2.1 is currently running on kaboom and gsboom, and is robust
and stable. A few minor bugs were flushed out by the high rate test.
A completely new version, 3.0, has been completed by Ronnie Misra (as
an MIT M. Eng. thesis project). This has significant additional
functionality: two way communication (for "anti-coincidence"),
inter-server communication, and additional security enhancements.
It's now in a form ready to be put into use (although it will have a long
term test before it replaces the current tested version).
High Rate Test and Automation Proposal:
--------------------------------------
I presented the results of last year's "high rate test", and
summarized the current proposal for an automated SNEWS alert for
astronomers. The high rate test data were analyzed by the subgroup
only, as per the privacy agreement; the results were okayed by the
advisory board to be shown.
The summary (slides from the meeting) can be found at:
http://www-lns.mit.edu/~schol/snnet/nu2002.pdf
(use the SNEWS working group password; if you do not know this,
please call or email me).
Here is the very brief summary of this talk:
-- In addition to verifying that the software works, the high rate
test showed no correlations between experiments (no shocker!), and
coincidence rates were as expected from the individual experiment
rates.
-- The proposal for automated involves a gold/silver scheme: the
alert only goes out for "gold-plated" coincidences; others get
human-checking. For a gold alert, there must be no history of
previous high rates for any of the experiments involved.
Here I will note only some of the comments, suggestions and points of
discussion from the meeting (refer to above link for background info).
* How exactly does one upgrade a silver alarm to gold? For the case of
individual alarms not flagged as "good", the alert can be
automatically upgraded by an individual experiment retagging its
alarm. For the other silver cases (alarms not geographically
separated, history of high rates), after checking, the individual
experiments can send their own alerts for a truly gold-plated signal.
* Doug Cowan suggested having a continuous scale where criteria are
applied with different weights to decide whether the alarm should be
automated or not.
* It's not really feasible for the server to track in detail dead time
from each experiment, so the past-high-rate based alarm suppression
may sometimes fail if, for instance, an experiment dead for the
previous several intervals suddenly starts up with a high rate.
However, a variety of time windows including short ones should help
suppress such a possibility.
* Walter suggested having a "high rate mode" like the high rate test
(using carefully tagged high rate alarms from each experiment) as part
of normal running mode. This will be very useful for monitoring
purposes. This suggestion will be incorporated into the proposal.
* Art suggested running in silver-only mode for some time before
going to a gold-enabled state. There were no objections, and this
will be incorporated into the proposal.
* Another suggestion was to add regular fake silver alerts.
We agreed that such fake alerts should be carefully tagged.
This suggestion will be incorporated into the proposal.
* The security audit part of the proposal was emphasized by Art; he
stressed that we need not just one, but regular audits. Such audits
are available commercially, but are expensive. As yet it is not clear
how we can accomplish this in an inexpensive way; a possibility, also
discussed at previous meetings, is to enlist the help of experts at
national labs. Art volunteered to get in touch with Los Alamos people
he knows. LNGS, Kamioka people will also investigate.
I will incorporate a number of the suggestions into the proposal.
I will make the revised proposal available within a few days.
Other Notes:
-----------
On the morning of the meeting, the snnet open discussion email list
received a bogus email saying "SN alert". As it turns out this was
sent from the meeting! However it was not signed with the SNEWS PGP
signature (nicely illustrating why we require this). This list, which
is a simple alias, can be converted to a major-domo-style mailing list
(which will suppress spam, too).
This archive was generated by hypermail 2a24 : Mon Jun 10 2002 - 10:00:19 EDT