Minutes of Detector Characterization Teleconference
(September 5, 2003)
Present:
Caltech: Stan Whitcomb
Hobart: Steve Penn
LHO: Rick Savage, Daniel
Sigg
Louisiana Tech: Natalia
Zotov
Loyola: Jennifer Barre,
John Whelan
Michigan: Vladimir Dergachev,
Keith Riles
MIT: Shourov Chatterji,
Peter Fritschel
Oregon: Ray Frey, Masahiro
Ito, Isabel Leonor, Rauha Rahkola, Brian Stubbs
Salish-Kootenai: Tim
Olson
Syracuse: Alessandra
DiCredico, Peter Saulson
M3 Planning:
-
Schedule reminder: M3 is a six-hour run September 13 18:00-24:00 PDT at
LHO
-
Daniel reported that H1 is doing well at high powers with record inspiral
ranges (>1.5 Mpc) and is expected to be ready for M3. H2, however, is not
in as good a shape, and there is further acoustic mitigation work planned
next week on its dark port optical table. We cannot count on its running
during M3, much less on it reliability.
-
KR reported that roughly half of the DMT authors plan to attend M3 at LHO
and half to attend M4 at LLO one week later.
DMT Software Status Review:
-
Status of DMT infrastructure
-
John Zweizig: Version updates, etc. (not present)
KR reminded everyone of John's September 10 deadline for DMT code changes
to be run under the process manager during M3.
-
Status of DMT monitors
-
Dave Chin / Tim Bodiya: LockLoss and ServoMon/SpecMon
KR reported that the LockLoss config file has been updated with the
new arm power transmission channel names for LHO. The new UP bit in the
IFO state vector will be checked (instead of the CM bit) for defining "common
mode feedback" as soon as Dave Barker confirms the bits are defined, a
change that will affect several other DMT monitors. Minute trends will
be provided for the LockLoss version of the state vector (previously visible
only in the DMT viewer). No code changes are planned for ServoMon,
but the config file will be updated for new IFO conditions. KR will try
out SpecMon before and during M3 and give feedback to its undergraduate
author (Tim Bodiya). It is unlikely that SpecMon will be ready for background
monitor production use in S3, but it may prove useful as an exploratory
foreground tool during the run.
-
Masahiro Ito: glitchMon
Shourov's linear predictor filter code has been implemented and is
being tested, with a planned CVS update in a few days. Masahiro is working
to set thresholds for new channels added to glitchmon at the request of
Rana Adhikari.
-
Sergey Klimenko: LineMonitor and WaveMon
There have been minor changes to LineMonitor, going into CVS soon.
Following some discussion about monitoring of calibration lines, Sergey
offered to add a heterodyne option to LineMonitor to allow comparison between
fluctuations with the present tracking algorithm and the heterodyne method.
He didn't want to promise readiness of the new option by M3, but believes
it will be ready soon afterward. KR reported that Mike Ashley (see below)
plans to provide a heterodyne measure of calibration line amplitudes/phases
offline (non-DMT) for alpha coefficient determination during S3.
More channels have been added to the WaveMon list, and a larger band
is now examined (up to 4 kHz). The time stride has been lengthened from
30 seconds to 1 minute. Sergey is concerned that the CPU resources required
will increase enough to become a problem, particularly given present limits
on the DMT machine memory buffer size. He may need to split WaveMon into
more than one program (as he has done already for LineMonitor).
-
Szabi Marka: IRIG-B, TimeMon, and ShapeMon (report
by e-mail)
All three monitors will be ready and running for M3. Szabi is also
working on a new feature to permit temporary silencing of DMT alarms. It
may be ready for testing in M3.
-
Tim Olson: SpectrumArchiver
The new version of SpectrumArchiver will be up and running for M3 with
an expanded channel list and coarse/fine resolution storage of fast channels.
The new SAM-QFS file system should allow 24/7 storage of spectra, with
old data rolling off onto tape, but still accessible on demand as it were
still on tape. The DTT should be able to pull up the stored xml files,
whether they are physically on disk or tape. Since the monitor will generate
more than a TB / year at LHO, pure disk storage is unattractive. Daniel
reminded everyone that in the long term, the resolutions used for different
channels should be tailored to the channels, based on experience. He also
suggested looking at schemes to save space when IFO's are out of lock,
perhaps not generating the files at all, or generating a truncated xml
file. PeterF suggested using the xml header to flag out-of-lock spectra.
KR suggested including the IFO state vector in the header. Daniel was still
concerned, though, about knowing which files contain good data without
having to open the files.
-
Steve Penn / Vijay Chickarmane: BicoMon
Steve expects his new background version of BicoMon to be ready next
week with trending and dmt viewer display of the bilinearity index (global
measure of bicoherence/nonGaussianity in AS_Q). This version will serve
as a template for an expanded version with particular bi-frequency regions
picked out, to be ready for S3. Steve will also provide an html summary
page. There was some discussion of revising the name. In a later e-mail,
Steve proposed BicoMon as the background monitor's name, with the foreground
version (previously called BicoMon) renamed to BicoViewer.
-
Brian Stubbs: SuspensionMon
Brian has been looking at an H2 event in S2 where side motion was kicked
up on April 11, to set his channel thresholds. He expects SuspensionMon
to be ready for testing in M3.
-
Patrick Sutton / Mike Ashley: SenseMonitor / LineAmp and RayleighMonitor
(separate phone conversation with KR)
The baton pass of SenseMonitor responsibility from Patrick to MikeA
did
not occur, as had been planned, with Patrick believed now to
be on the road and unreachable until after M3. KR is very concerned about
the situation, both for M3 and for later in the fall, and will be exploring
fallback arrangements with Mike Landry for M3.
On a brighter note, MikeA reported that his matlab-based LineAmp program
can determine revised alpha/beta coefficients for the entire S2 data set
in about five days, and he expects to be able to speed the code up by a
factor of 10 relatively easily. LineAmp used heterodyning to determine
calibration line strength.
-
Natalia Zotov: PTMon (report by e-mail)
Still trying to track down a bug associated with multiple triggers
/ frame. Will work on the problem next week.
-
John Zweizig: BitTest, DataQual, HistCompr, PSLMon,
and SegGener (no report)
A.O.B.
-
PeterF wondered whether the next version of SenseMonitor would account
for the dynamically changing input matrix planned for S3. KR didn't know
and offered to look into it. (Followup: this should be straightforward
to implement, using same EPICS channel scheme planned to monitor changes
in DARM gain.)