Minutes of Detector Characterization Teleconference
(October 1, 2004)
Present:
Caltech: Peter Shawhan, Patrick Sutton,
John Zweizig
Florida: Sergey Klimenko
LLO: Natalia Zotov
Loyola: Marc Cenac, John Whelan
LSU: Gaby Gonzalez, Chad Hanna
Michigan: Keith Riles
MIT: Lindy Blackburn, Erik Katsavounidis
Oregon: Ray Frey, Rauha Rahkola
PSU: Shantanu Desai
S2 / S3 Investigation Report:
-
Data quality investigations: (S2,
S3 )
& segment repository (S2, S3)
- (John Zweizig, Keith Riles)
John reported he is working on indentifying airplanes at LLO during S2
(S2 LHO and S3 LHO/LLO already done). Gaby mentioned that after the final
S3 calibration release is ready (version 3), there will be a new set of
associated data quality flags indicating problematic calibration intervals.
These new flags should supersede the old flags. KR briefly summarized the
recent release of S3 data quality segments.
DMT Software:
-
Critical DMT issues for M5 preparation:
-
John Zweizig: Infrastructure status and deadlines for new monitors
Because the new EasyCalibrate class makes frequency domain analysis
more useful, in general, John has been cleaning up the FSeries infrastructure,
including adopting the FFTW library. One issue is that the old DMT DFT
normalization has differed from the official LSC spec. John will conform
to the spec and send out a note to DMT authors warning of the change.
DEADLINE for new code to be run in M5: Friday October 15
John Whelan mentioned very recent problems running online on fortress.
JohnZ suspects a recent reboot and will look into the problem. He is concerned
about frame losses approaching 20% at LLO in recent days. Preliminary investigations
haven't revealed the source. He has contacted Chethan Parameswariah for
help.
Sergey requested that JohnZ create workspaces or provide sample makefiles
so that monitor authors can more easily work on only their monitors directly
in the CVS tree without having to rebuild all other monitors after
a CVS update. JohnW and Keith agreed that it would be nice to provide
a sample makefile for stand-alone monitor building. JohnZ agreed to look
at breaking out a subset of the current global makefile.
-
Patrick Sutton: Calibration infrastructure
The new EasyCalibrate class is now available for general use. SenseMonitor
itself has been updated to use the new class and should serve as a useful
template for other authors wanting to use EasyCalibrate. Documentation
is available in html. JohnZ mentioned that in addition to being available
as source code in the CVS tree, the new code is in compiled form on
the Caltech DMT machines. The new class works only for H1, using the
old S3 calibration reference functions. The calibration team will need
to run a root script to generate new EasyCalibrate xml files, presumably
before E11.
-
John Whelan / Mark Cenac / Brian O'Reilly: StochMon
StochMon compiles and runs sucessfully on frame file and will be put
into CVS by the end of the weekend. For the M5 mini-run, it will work
for H1-H2 and H2-H1 (true ifo against reference ifo). They are hoping
to incorporate an overlap reduction function into the monitor before M5.
KR suggested focussing instead on incorporating EasyCalibrate as the higher
priority.
-
Ken Ganezer: seis_blrms
Ken has a new version of seis_blrms compiled and running which he believes
will correctly handle data dropouts. KR suggested running it at LLO where
data dropouts are now rampant (see above).
-
Sergey Klimenko: BurstMon
Sergey has started incorporating EasyCalibrate into BurstMon and hopes
to have a running version in a couple of days.
-
Giovanni Santostasi: PulsarMon
Giovanni reported by e-mail that he has implemented but not yet tested
dmt viewer plots. He will try incorporating EasyCalibrate now that it is
available.
-
Rauha Rahkola: SuspensionMon
Rauha will be putting the new SuspensionMon version (bug in trending
fixed) into CVS this weekend.
-
Other DMT reports:
Lindy reported on KleineWelle which is a DMT program developed primarily
for offline burst search analysis, but which he plans to adapt to online
control room use. The single-IFO glitch finding is based on pixel clustering
in the time-frequency plane where the frequency basis functions are dyadic
Haar wavelets (similar to BurstMon). Linear predictive filtering (order
1024) is used for conditioning. Lindy showed efficiency vs hrss plots for
several Gaussian waveorms. He also showed glitch rates from S3 playground
for H1, H2, and L1, along with times of triple-coincidence triggers. Rate
variations of more than an order of magnitude are observed. KleineWelle
has also been used to study glitches in auxiliary channels as veto candidates.
Half of the 30 triple-coincidence AS_Q triggers can be vetoed from other
channels, but deadtimes have not yet been determined. Lindy will be adding
KleineWelle to the GDS CVS treee soon and will continue development
of an online gltich rate monitor. JohnZ wondered if Sergey's wavelet library
could be used by the monitor instead of its own library, so that the DMT
supports only one set of generic wavelet routines. Lindy will take a look
at Sergey's routines.
A.O.B.
-
Next detchar telecon: Friday October 15 at 1:30 p.m. EDT (M5
and DMT software)