Detector Characterization Report & Issues
K. Riles (Michigan)
(LIGO Scientific Collaboration Telecon - February
20, 2002)
Focus here on E7 and S1 issues
Some E7 info: (click here
for official E7 web site - your computer's i.p. address must be recognized
by Hanford)
-
Inteferometer livetimes: (numbers from G. Gonzalez,
D. Chin, P. Shawhan)
-
H1(72 %), H2(53%), L1(71%), H1*H2*L1(35%) - Raw
-
H1(48 %), H2(28%), L1(53%), H1*H2*L1(11%) - "Clean", longer than 15 minutes
-
"Playground" data selected for initial, exploratory upper limits analysis
(G. Gonzalez, P. Brady)
-
Single-IFO (anti-coincidence): H1 (14 hours), H2 (30 hours), L1 (28
hours)
-
Triple-IFO (coincidence): H1*H2*L1 (5 hours = 10% of total triple coincidence)
-
Preliminary sensitivities (M. Landry, S. Marka, R.
Adhikari, L. Matone)
-
H1: ~10-18 / Hz-1/2 @ 200 Hz
-
H2: ~2 x 10-20 / Hz-1/2 @ 900 Hz
-
L1: ~7 x 10-20 / Hz-1/2 @ 900 Hz
Click here to see preliminary sensitivity
curves for all three IFO's.
-
Astrophysical test signals injected simultaneously at both sites (P.
Shawhan, S. Bose)
-
Bursts, Inspirals, Stochastic
(Much more detailed info on the above topics can be found in
the minutes (and via links) of the February
detector characterization teleconference. See Gonzalez,
Landry, and Shawhan reports.)
-
Thirteen E6-E7 investigation teams - four preliminary reports at February
detchar telecon (N. Christensen, S. Klimenko,
M. Landry, P. Shawhan). More expected at March 8 telecon and March
LSC meeting.
-
Scientific monitoring shifts - 140 12-hour shifts (70 expert, 70 trainee)
taken by 51 scientists (36 travellers)
-
Initial feedback from shift takers, including unhappiness with 12-hour
shifts, summarized in minutes of January
detchar telecon
-
Detailed statistics and ongoing discussion of utility and allocation of
scimon shifts can be found in the ligo-dc
e-mail archive
-
More than a dozen DMT background monitors running continuously with varying
output options:
-
Summary html/ascii pages
-
DMT viewer plots (time series, spectra, histograms)
-
Database triggers
-
Trend files
-
Detailed log files
Some lessons learned from (or reinforced by) E7: (what needs to
be fixed)
-
Need more effective feedback from background data monitors to operators
and scientists on shift
-
Many monitors lacking summary html page (several authors working on this
now)
-
Need more guidance on what is "normal" for each monitor
-
Need global DMT status and alarm handler web page (S.
Marka now working on prototype)
-
Need online figure of merit for astrophysical sensitivity (Penn. State
student working with S. Finn on this now)
-
Need integration of DMT-produced trends into Data Viewer playback
-
Need more automation of mundane scimon tasks and more interactive tools:
-
Keeping track of "clean" locked segments (streamlined scheme to be
tried in E8 using "physics mode" button)
-
Producing graphical output sometimes cumbersome
-
Periodic storage & easy retrieval of reference spectra similar to autoburt
(looking for volunteer!)
-
GUI-driven real-time time/frequency display (S. Penn
now working on prototype)
-
GUI-driven real-time band-limited RMS display (E.
Daw adapting his background blrms monitor)
-
Need feedback from on-site downstream data analysis
-
Need easy confirmation that LDAS pipeline (data & trigger flow) functioning
properly (D. Brown now working on a GUI monitor/control
interface)I
-
Need low-threshold trigger rates from burst / inspiral analyses to spot
problems early
Questions for upper limits groups to consider:
-
How is detector noise level most usefully summarized for your analysis?
-
Power levels in selected bands?
-
Global astrophysical figure of merit, e.g., standard candle distance sensitivity
-
Do you want dedicated database "clean" segments using pre-defined criteria
(binary real-time decision)
-
Or do you want "analog" trends to base your own decision upon?
-
Calibration:
-
Does present preliminary calibration information work well for you online?
-
Do you want fitted pole/zero info or raw measurement data?
-
How do you wish to keep track of sensitivity variations
(e.g., parametrized time-dependent scale factor, raw injected line
strength)?
Some longer-term issues:
-
Scientific monitoring shifts
-
Need (I believe) to man 24/7 shifts at each site for forseeable future
-
Present voluntary scheme unlikely to work well when data runs extend to
months
-
Ongoing
discussion mentioned already
-
Barry & Rai working on a proposal to discuss at March LSC meeting -
Let's try to resolve this well before S1 begins!
-
Reduced data sets
-
An initial E7 RDS being generated by I. Leonor
that may well satisfy needs of E7 analysis with factor of three reduction
coming from channel elimination (no decimation). UL groups reluctant in
first round to decimate, in part because un-MDC-validated, stand-alone
program being used.
-
Hope for feedback from UL groups on channel selection / bandwidth from
first RDS iteration.
-
Next iteration, using LDAS infrastructure, may contain more channels and
demand decimation accordingly, given limited disk space at sites.
-
In any case, we must eventually confront the decimation issue in S1 and
should get some experience with it sooner rather than later