Minutes of Detector Characterization Teleconference
(July 13, 2001)
Present:
Caltech: Mours, Shawhan,
Yamamoto
Louisiana Tech: Greenwood
LHO: Mavalvala, Raab,
Rahkola, Savage, Whitcomb
Michigan: Riles
MIT: Sylvestre
Oregon: Schofield
Introduction - KR
-
(Because of the Amaldi meeting and summer vacations, more persons than
usual are reporting by e-mail this week.)
-
DMT monitor developers are reminded of the upcoming E5 run (August 3-6).
New monitors and improved old monitors are strongly encouraged. In particular,
adding an interface to the DMT viewer to existing monitors should be high
priority. Providing an html summary page is also welcome, along with more
documentation on how monitors work and what their outputs mean.
-
There have been a handful of requests to give talks in the detchar sessions
at the upcoming LSC meeting (August 13-16). In addition, all E3/E4 investigation
team leaders will be put on the agenda to give reports. Final E3/E4 written
reports will be due at the August meeting. These should be separate from
any E5 reports to be written up in the future, but time will be made for
any preliminary E5 reports that are ready in time.
-
KR plans to show at the meeting a status table of DMT monitors like that
shown at the March meeting, with columns for the various features that
the monitors support. As a carrot to developers, all changes in the table
between March and August will be highlighted in bold color.
-
Similarly, tables will be shown that list the status of E2 and E3/E4 investigation
reports with columns for preliminary telecon reports, oral reports at LSC
meetings, and final written reports. (Click here
to see the current E2 final reports web page.)
-
Another teleconference will be held in two weeks (Friday July 27), mainly
to discuss E5 preparations, but preliminary E3/E4 reports are welcome too,
as this will be the last detchar telecon before the August LSC meeting.
Preliminary E3/E4 Reports:
-
Data integrity - P. Shawhan
Peter has been looking at data in the HPSS archive to find global problems
in data recording/transfer. Examining raw data along with the second and
minute trends from engineering runs E2-E4, he has found five different
classes of problems that recur:
-
1) Second-trend files missing from archive. The omissions were due to the
manual process now used for ingesting into the archive and have now been
fixed by Stuart Anderson.
-
2) Gaps in the raw data because of DAQ reboots. These typically last about
a minute. Because second-trend files are written once/minute, there is
usually a 1-2 file gap for those, with no gap in minute-trend files, which
are written once/hour.
-
3) However, the DAQ system pads the trend files with leading zeroes after
a reboot, meaning that up to a minute of zeroes can appear in a second-trend
file and up to an hour of zeroes in a minute-trend file.
-
4) There were sporadic truncations of minute-trend files, making them unreadable.
Several hours from E2-E4 were lost as a result. The problem (due to a timing
mismatch) has been identified and fixed.
-
5) There are occasional short segments of second-trend data where all channels
examined are zero. This is not yet understood, but affects a very small
fraction of the data.
Peter remarked that his data checking to date has been quite tedious because
it's done manually. We need automated checking as part of data ingestion.
A natural place for such checking is in the LDAS data ingestion API, which
hasn't yet received much programming attention.
Peter also mentioned that all of the data is available via the getframes
utility and soon will be available via Lidax (see below). Lidax will provide
lists of known contiguous data segments to handle the sort of gaps found
above.
E5 Preparations:
-
Interferometer configurations - S. Whitcomb
-
It's reasonably likely that the Hanford 2km IFO will be in a full recycled,
recombined mode, giving us the first engineering data in that configuration.
The 2K is nearly back to its pre-earthquake state.
-
It's unlikely that either of the 4K IFO's will have any arms locked in
time, given recent shutdowns. Their mode cleaners may be operational, however.
-
Common-mode, feed-forward tidal actuation (acts on reference cavity temperature)
has been tested on the 2K and is tentatively calibrated at about the 20%
level. Differential-mode tidal actuaction will require some more hardware
and software work to be ready by E5.
-
Investigations, scientific monitoring shifts, and DMT monitors - KR
-
Investigation teams are forming up with the E3/E4 lineup as defaults. A
preliminary list will be posted next week.
-
Many scientific monitoring shift slots remain to be filled for E5, especially
at Hanford. More volunteers are requested. A preliminary shift list will
be posted next week.
-
An E5 telecon will be held Friday July 27 at 11:00 a.m. EDT at which investigation
leaders will be asked to summarize briefly their plans and to make any
final special requests for running conditions.
-
Any new or revised DMT monitors to be run centrally by the process manager
during E5 should be given to John Zweizig no later than Friday July 27.
Data access using Lidax - D. Sigg (attending Amaldi meeting,
reported by e-mail)
-
Daniel has very nearly completed the Lidax package, which allows flexible
data access from various sources, including the hpss archive, the observatories
and local files. A prototype version is now available for testing on fortress
and will be released for wider use soon. A quick-start
guide is available.
Status reports on DMT infrastructure:
-
General - J. Zweizig (attending Amaldi meeting, reported by e-mail)
John has finished a new trigger-directed data analysis environment
class (TrigEnv). It's similar to the standard DatEnv class, but it reads
data offline from epochs instead of from an infinite stream. The epochs
are taken from an xsil table (e.g., produced by guild). A preliminary template
monitor is available on Caltech computers.
A new major DMT release is planned soon that will include these changes
and DTT-related changes (see below). It will also include Masahiro Ito's
histogram class, Dave Chin's revised operational state conditions, and
Ed Daw's IIR filter class (see below).
-
Merging with DTT - D. Sigg, J. Zweizig (attending Amaldi meeting,
reported by e-mail)
Daniel and John are putting together a combined version of the GDS
code with the DMT and DTT better integrated via a merged source tree. The
DMT/DTT integration will also include the present DTT calibration infrastructure.
In the farther future, plotting tools used now by the DTT and DMT viewer
will be made available on the root command line.
-
Histogramming - M. Ito (reported by e-mail)
Masahiro has completed the code that defines 1-dimensional histograms
in the DMT environment and allows passing of histogram structures to the
DMT viewer via the MonServer, MonClient and MonDataMsg classes. He is now
updating Daniel's ROOT extensions within the DMT viewer itself to provide
a histogram display gui.
-
Operational state conditions - D. Chin, K. Riles (KR reporting)
Prompted by requests from John, Dave has reorganized some of the code
to make it more efficient and to speed up compilation, but there is more
work to be done on that. As noted above, the functional changes reported
at the last detchar telecon will be available in the next DMT release.
Status reports on DMT monitors:
-
Line noise (multi-taper method) & inter-channel correlation monitors
- B. Allen, A. Ottewill
(Adrian reported by telephone afterward)
Nelson Christensen continues to use the correlation monitor for E3/E4
studies. A dramatically revised version of the multi-taper line monitor
has been sent to Bernard Whiting for performance testing. The new version
is better integrated into the DMT background monitor format and includes
both a DMT viewer interface and database trigger generation. Adrian expects
to send the new code to John next week for use in E5.
-
Line noise (quasi-monochromatic method) monitor - E. Daw, S. Klimenko
(not present, attending Amaldi meeting)
-
Band-limited / seismic noise monitors - E. Daw (not present)
-
Bilinear cross-couplings & broad-band non-Gaussianity monitor - S.
Penn (reported by e-mail)
The stand-alone bispectrum program now includes bicoherence which is
expected to be more useful than the unnormalized bispectrum of the old
version. Steve plans to be at Livingston before and during E5 and will
give tutorials on using the program to anyone interested.
-
Lock transition and servo instability monitors - D. Chin, K. Riles
(KR reporting)
Both the lock transition and servo instability monitors now have dmt
viewer interfaces that show histories of lock transitions and servo alarm
triggers. Dave has given the revised monitors to John for running in E5.
-
Rayleigh behavior monitor (non-Gaussianity vs frequency) - S. Finn,
G. Gonzalez, P. Sutton (not present, attending Amaldi meeting)
-
Power spectral transient monitor - S. Mohanty (not present, attending
Amaldi meeting)
-
Transient identification & event catalog monitors - J. Sylvestre,
R. Weiss (Julien reporting)
Julien plans to release (today or Monday) new versions of the tid program
suite. The suite consists of a DMT background monitor (server) that computes
time-frequency plots, an interactive, foreground gui program (client) that
can run anywhere and communicates with the background monitor to display
the plots in real-time scroll mode, and another DMT background monitor
(event catalog) that identifies patterns in the time-frequency plane. The
default Nyquist frequency binning can be changed with an option that turns
on a decimation filter.
The time-frequency plotter can keep up in real-time with plotting several
2kHz-rate channels. It's meant for exploratory finding of recurring patterns.
With trial and error one can determine pattern templates (rectangular only,
for now) and thresholds that can then be inserted into an ascii configuration
file that the event catalog reads. At the moment, a config file exists
to identify a known transient due to a compressor near the LHO mid-X station,
and Julien is working to develop others. For the moment, when a pattern
is recognized the event catalog produces ascii and html files. In the future
it will also generate a database trigger.
Documentation, including a detailed example can be found here.
-
Glitch monitor - M. Ito (reported by e-mail)
Some minor improvements are planned for E5: suspending the calculation
of ambient standard deviations during detected events and adding more information
to the database output. Some sample
events found by the glitch monitor during the engineering runs have
been posted.
-
Magnetic field and seismic transient monitors - R. Frey, R. Rahkola
(Rauha reporting)
An earthquake monitor that sets visible and audio alarms in the control
room is now being tested. The monitor uses its own EPICS interface, different
from the default DMT interface for triggers, allowing a dedicated EPICS
channel for the monitor. In its first test the monitor picked up seismic
triggers during a recent windstorm.
A.O.B.
-
Fred Raab reported that after some installation work Hugh Radkins is now
working again on tidal modelling. E3 data has been looked at, and E4 will
be soon. A preliminary report will be given in two weeks.
-
Next teleconference (E5 preparations & E3/E4 reports): Friday
July 27 at 11:00 a.m. EDT (8:00 PDT).