Minutes of Detector Characterization Teleconference
(November 3, 2000)
Present:
AEI-Potsdam: Bose, Chassande-Mottin, Mohanty,
Sintes
Caltech: Drever, Kells,
Lazzarini, Marka, Mours, Shawhan, Whitcomb, Yamamoto, Zweizig
Carleton: Christensen
Dublin: Ottewill
LHO: Gustafson, Landry,
Ottaway, Raab, Savage, Sigg, Weiss
LLO: Daw
Loyola: McHugh
Michigan: Riles
MIT: Adhikari, Sylvestre
Oregon: Rakhola, Schofield,
Strom
Penn State: Finn
Syracuse: Penn
Issues related to next week's E2 run (Special requests? Questions?)
- All
-
Items posted on agenda page:
-
The baseline reduced data set gives a volume rate of about 1.2 MB/second
which will make it hard to store all collected E2 rds samples on fortress
disks. Investigating teams were asked to send any remaining channel requests
to KR over the weekend. The standard RDS channel list and any special lists
are to be given to Daniel Sigg on Monday. So far only the lock loss investigation
has requested channels not readily incorporated into the standard list.
-
Note added on Monday: Three channels lists have been sent to Daniel: Standard
RDS, RDS for lock loss studies with
1 arm locked, and RDS for lock loss
studies with 2 arms locked and/or the power recycled Michelson cavity locked.The
lock loss lists are for extra channels to be added to the standard
RDS during time periods of order(1 shift). Data volume rates (Bytes/second)for
these lists are
Baseline RDS(standard) = 1,411,392
Extra RDS(lockloss1)
= 328,704
Extra RDS(lockloss2)
= 1,254,400
-
An earlier e-mail from Stan Whitcomb concerning likely interferometer configurations
was briefly summarized. The run will focus on two stable configurations:
single-arm and recombination with both arms locked (no recycling mirror).
An effort will be made to take data with full recycling, but for only a
shift or two at the end of the run.
-
Investigation leaders were urged to look over Daniel's proposed schedule
of shift activities and give him feedback. The schedule starts the run
with several days of recombination running, then a day of parameter studies,
followed a couple of days of one-arm running.
-
All E2 participants were urged to send any travel schedule corrections
to KR and to give feedback to John Zweizig on the proposed end-of-shift.
Data Monitor Tool Status (John Zweizig):
-
John has released a preliminary version 1.2.3 with recently contributed
code from Daniel Sigg (power line monitor), Szabi Marka (GPS time ramp
monitor), Benoit Mours (data summary set generator), Sergey Klimenko (wavelet
tools), and Dave Chin/KR (servo instability monitor). Tweaking of the code
is ongoing.
-
John urged authors of new monitors that are to run in background to make
sure he has useful configuration files, environmental variable settings,
and any command line sequences needed to start the monitors up. Brief documentation
of the monitors should be provided to Szabi Marka.
-
Investigation team leaders were urged to decide what monitors they will
want during the run and to make sure John knows about them, including how
to run them.
-
He has tried running 13 DMT monitors in parallel on the sand machine to
verify that cpu and memory load is indeed approximately linear with increasing
monitor processes. No surprises were seen, but more rigorous testing with
recently contributed monitors will start over the weekend to help shake
down the system.
Status reports on performance characterization priority 1 tasks:
-
Line noise:
-
B. Allen, A. Ottewill (Adrian reporting)
Nothing new to report on installed code.
-
S. Klimenko
Sergey will be present during E2 and will be using his and Daniel's
line monitoring code as part of the line noise investigation. His own DMT
monitor will be run in foreground using root macros. He requested that
voltage monitor channels be available during the run.
-
A. Sintes
Alicia will be visiting the U. Florida in December prior to the GWDAW
meeting to meet with Bernard Whiting to sort out unresolved discrepancies
they find in the performance of her line removal algorithm.
-
Characterizing seismic noise:
-
E. Daw
Despite much effort and consultation with John, Ed hasn't solved the
problem reported last meeting of spontaneous monitor crashes after a few
hours of running. He now suspects a decimation routine in the DMT infrastructure
to be at fault. He will be present at LHO for the last several days of
E2 and will be looking at his seismic monitor's output from Louisiana in
the meantime.
-
Inter-channel correlations:
-
B. Allen, A. Ottewill (Adrian reporting)
New dmt code has been given to John to help with the correlations investigation
during E2. The new code can be run as a background monitor. Nelson Christensen
will provide John with an updated configuration file for the monitor on
Monday.
-
Bilinear cross-couplings
-
S. Penn
Steve has run into new problems installing the DMT on his powerpc linux
box, despite having successfully installed an earlier DMT version. He has
consulted several experts on linux DMT (JZ, Ed Daw, Ed Maros) and believes
the problem arises from a powerpc assembler error. He will not have code
ready in time for E2, but the non-Gaussian investigation will be based
on post-run analysis, anyway.
-
Operational state:
-
D. Chin, R. Gustafson, K. Riles (KR reporting)
Graduate student Dave Chin has taken over the osc code and has delivered
a revised version to John, including a background monitor for detecting
servo instabilities. The parameters for the monitor's config file still
need considerable tuning, which will be attempted during E2. The monitor
does not yet talk to the monitor display manager, more development
to be done during E2 by Dave who will be on site for two weeks. He is working
now, however, (at Daniel's request) on a modification of the osc code to
allow triggered dumps of data with run-time-definable duration and deadtime
parameters. Dave hopes to have the code ready by the start of E2. The triggered
data dumps should be immediately useful to the lock loss investigation
and perhaps to other transient investigations.
Status reports on performance characterization priority 2 tasks:
-
Bandlimited RMS
-
Time / Frequency plots
-
S. Mohanty (had to leave telecon early - reported later by e-mail)
No recent progress to report, getting back into the loop at AEI after
move from PSU.
-
J. Sylvestre
The TID program has been augmented to handle DMT triggers and
to allow fixed thresholds instead of adaptable thresholds determined from
the data. The adaptable threshold algorithm is cpu-intensive, while the
fixed threshold version is much less so. The price, however, for that reduction
is a one-time highly cpu-intensive calibration procedure to determine the
fixed thresholds. Nominally that calibration requires two of the four cpus
on, for example, sand or stone, along with 600-700 MB of memory for several
hours. Julien will plan to carry out the calibration of PEM channels before
E2 begins and will work with John to make sure that any calibrations of
IFO channels needed during E2 do not create havoc with other DMT monitors.
Rai Weiss asked whether a test had been carried out to verify that
all of the available DMT monitors can be run simultaneously. John said
he would be doing that before E2 starts (see above). Rai also worried that
the deluge of visitors to LHO during E2 would saturate available computing
resources, online and offline. John said he would look into setting up
an extra data server on the general computing system at LHO to reduce the
load on the online machines. Visitors were urged to bring their own laptops
to help reduce demands on computing. Rai also wanted to know who would
be responsible to oversee run activities. The answer is Daniel and John
who were urged by Stan and Rai to make ruthless decisions when necessary
about sharing of resources and in curtailing activities (like gabbing and
milling about in the control room) that interfere with efficient E2 running.
-
Non-Gaussian noise
-
S. Finn, G. Gonzalez (not present)
Status reports on transient analysis priority 1 tasks:
-
Power spectral transients:
-
Servo instability:
D. Chin, R. Gustafson, K. Riles (see above)
-
Event catalog:
-
J. Sylvestre, R. Weiss (see above)
Status reports on transient analysis priority 2 tasks:
-
Flickering optical modes:
-
Transient detection using adaptive denoising methods
-
E. Chassande-Mottin (had to leave telecon early - no report)
-
Impulse recognition:
-
M. Ito (not present)
Rauha Rahkola believes Masahiro's code is working and will be running
during E2.
-
Magnetic field transients
-
R. Frey, R. Rahkola (Rauha reporting)
This code should also be running (in foreground) during E2 for measuring
both magnetic and seismic transients.
A.O.B.
-
KR asked about the DAQ status at LHO since he had been unable to connect
to the network data server in the morning. Daniel reported some flakiness
with a frame builder disk drive. Stan said that Rolf Bork would be making
a special visit to LHO starting on the weekend to try to resolve a number
of recent DAQ problems before the start of E2.
-
Next detector characterization teleconference: Friday December 1 at 11:00
a.m. EDT (8:00 PDT).