Subject: Minutes of meeting at SNII workshop
From: Kate Scholberg (schol@mit.edu)
Date: Mon Feb 19 2001 - 12:37:51 EST
Dear SNEWSers,
Thanks to help from the SNII meeting organizers, we were able to have
a SNEWS working group meeting over the two lunchtimes of the conference.
Here are the minutes of the meeting.
Kate.
SNEWS Working Group Meeting at SNII in Marina del Rey
=====================================================
February 15-16, 2001
Present (on at least one day): Pietro Antonioli, Barry Barish, Dick
Boyd, Walter Fulgione, Szabi Marka, Ronnie Misra, Kate Scholberg, Reda
Tafirout, Nicolai Tolich, Mark Vagins, Clarence Virtue.
News from Experiments:
----------------------
-- Super-K's supernova watch "expressline" (which bypasses certain
stages of the data pipeline to do a prompt burst search and provide
an alert within <~ 5 minutes), has been running for some time. A bug
was recently fixed and it's now been in continuous operation for more
than a month.
Super-K will be off for ~3 months starting at the beginning of July for
its long-planned upgrade (the shutdown will be preceded by a
"calibration fest" in June.) During draining and filling, the supernova
monitor will be on as much as possible (e.g. at night), with
phototubes watching whatever water is in the tank.
-- SNO is coming to the end of its pure D20 phase. They plan to add
salt sometime around the end of April, after a planned pt calibration.
Reda gave a talk at the workshop detailing the supernova burst search
analysis; the final alarm rate is easily within the <~ 1/week range. Alarm
response is still manual.
-- With the completion the third tower, LVD has reached 1 kton of
active mass. They have also installed a new DAQ, which the detector
to run in a more modular way, such that it's not necessary to shut
down entirely for calibrations, problems, etc. Alarms to SNEWS
were turned off in December for the new DAQ's shakedown period,
but they expect to come back online for SNEWS within a couple of
weeks. At that time, they will also upgrade to version 2.0 of
coinccode.
-- MACRO was shut down entirely in December.
-- Nikolai Tolich from Stanford joined us as a new KamLAND rep. He's
working on KamLAND's trigger. KamLAND's PMTs are now completely
installed and oil filling should start in the fall.
-- Dick Boyd reported on OMNIS: they're now setting up a test module
with 58 tons of lead. See also the talk at this workshop.
-- LIGO: the first interferometer is in operaton at Hanford and the
second one in Louisiana is now turning on. They expect the first
coincidence data between the two in September, then first real data in
mid-2002. Szabi Marka is now working on the fast alarm channe. Adam
Burrows is also working on the expected signal, about which very
little is known. Right now the trigger is based simply on excess
power in the right frequency range, but refinements based on expected
time evolution of the signal should be possible. Relative timing
between gravitational and neutrino signals also needs to be understood
(gravitational waves will probably come first, but it's not known for
sure).
SNEWS news:
----------
-- As reported in previous messages, version 2.0 of the code was
released: this includes multiple server capability (each client simply
loops over servers), plus work by Corrinne Mills last summer on:
* secure sockets
* alarm retraction utility
* multiple platform builds
I'll upgrade the server once the Super-K and LVD clients
are ready.
-- In addition, there will be some reorganization: new servers gsboom
and mitboom will be added to the network, and the repository will get
moved to mitboom.
-- The next upgrade will be done by Ronnie Misra. The main added
functionality will be two-way communication (to allow
"anticoincidence" and consistency information exchange), inter-server
communication (to suppress duplicate alarms), plus improvements in
security and robustness. He gave a talk detailing how he will
implement this.
The dominant feature of his new design is the use of multithreading
and IPC to implement the two-way communication: the idea is to have a
separate process running in the background that deals with the SNEWS
messaging, while the regular data collection/burst searcher program
keeps running independently. This setup avoids any interruption of
data collection during conversation between client and server. He's
also working on a multi-server protocol, and improving security.
For instance, digital signing will authenticate the data itself,
rather than just the connection.
Ronnie will post his M. Eng. thesis proposal with the details of the
redesign. Please send him any comments or suggestions.
Automation
----------
-- The plan for getting to automation is the following:
The subgroup will get together to perform a high rate test, and then
prepare a report for the advisory board members of the experiments
involved in the test. These advisory members will then make the decision
to go ahead or not, based on the report.
SNO, LVD and Super-K are now ready to go ahead with this test. Presumably,
the procedure can be repeated whenever a new experiment is to be added
to the automated coincidence.
Here's the proposed specific format for the test:
* Each experiment supplies a plot of rate vs P, where P is some
software-twiddlable parameter(s) (multiplicity threshold, etc.), and
where the rate ranges between ~1 per minute and 1 per day.
* The subgroup then calculates expected coincidence rates
for different values of the individual experiment parameters.
* Then, during some specified period of time (~ 2 weeks),
we set the thresholds and count coincidences, to see
if the behavior is as we expect.
This checks that the coincidence software works, too, and
should help to shake out any problems.
We should set rates such that we get a total of ~100 or so
coincidences during the test (we can widen the time window if necessary).
I showed some sample curves sent by Fukuda-san at Super-K. The
Super-K "dials" are multiplicity in 20 seconds and Rmean (mean distance
between event vertices). It's possible to dial in alarm rates between
1/hour and 1/day, and it may be possible to go to higher rates by
twiddling the "pre-multiplicity" thresholds.
SNO and LVD can also easily produce similar plots: SNO needs a few days,
LVD a couple of weeks.
We reserved the last 2 weeks of March as the time to exchange data and
arrange the test times and parameters. April is probably good time
for the test. (It will definitely have to be before the SK shutdown).
Szabi suggested that the high rate test be arranged to coincide with a
LIGO "engineering run"-- he'll look up the schedule.
There was also a suggestion that the subgroup
look at the time series of SK and LVD alarms already sent to kaboom
over the last 1.5 years or so, along with the list of SNO alarms that would
have been sent, to include in the report to the advisory board.
We also set the goal of having an automated alert to astronomers
following the Super-K turn-on in the fall.
Astronomer Alert News
---------------------
Leif Robinson of Sky & Telescope has retired as editor in chief, but
will still remain involved in SNEWS. Rick Feinberg, the new S&T
editor in chief, is interested in keeping up the SNEWS connection. A
new persion, Roger Sinnott, has been designated as contact person and
Astro-Alert mailing list maintainer.
They're also going to revive work on the amateur astronomer test.
This archive was generated by hypermail 2a24 : Mon Feb 19 2001 - 12:38:46 EST