LSC-Virgo Burst Analysis Working Group
Review Committee Meeting Monday 13 February 2006 09:00 PST / 12:00 EST
Minutes: Monday 13 February 2006 09:00 PST / 12:00 EST
Agenda and Contact Info
- Continue the review of the GRB GWB search
AccuConference teleconferencing service:
Phone: 1-800-704-9896, participant code: 038621#
International callers ++1-404-920-6472 with same code
- Code Review
- Peter Kalmus will be conducting the code review. He has received
code from Isabel and Soumya and will be reviewing it. He is preparing
a list of claims of what the code does from the technical documentation.
- Objectives of code review are:
- Establish that there are no conceptual errors in
- Establish the code implementation and use is in complete
agreement with the documentation.
- Establish that this code implementation is free from any
- Establish the provenance and correct accounting of the data
- Make suggestions of what conceptual implementation changes that
authors should be persuing in improving astrophysical reach.
- Fourth item is new. The goal here is to make sure that
all of the data that should have been analyzed was analyzed successfully,
produced valid data products, and was correctly accounted for in the
post analyses. It includes things such as making sure that all the data
was analyzed with the same CVS-tagged version of the software, that due
to some bug some analyses were accidently omitted, etc.
- Items 1 and 5: these items refer more to the work of the review
committee as a whole rather than the code reviewer. The code reviewer
is not really responsible for performing these tasks, but may have unique
insights because of their intimate knowledge of the working of the code.
- Code review should feel free to comment on possible improvements to
code efficiency and style. But these are of lesser importance to those
points listed above.
- Suggestion: maintain code review documentation on a burst enotebook
- Target is to review substantial parts of the code by the LSC meeting.
- What needs to be done by Sunday, Feb 19 deadline?
- Review committee would like specifics about what is to be
in APS presentations. For example, descriptions of the figures,
sensitivity statements, and results statements that are to be given
(that need review).
- Preliminary presentations are not required.
- We need to know what we have to review in advance so that we can plan
to have it reviewed by the LSC meeting. Goal is for us to be able to
say "we have reviewed this material" after the presentation at the LSC
meeting so that the APS presentation can be approved then.
- It will not be possible to have complete reviews of the various
analyses so we need to isolate those aspects that are to be presented.
We may need to do triage if we cannot review everything that people want
- What about S5 results? These will not be available by Feb 19.
Depending on what exactly will be presented we do not perhaps need to
have full details by Feb 19. For example, if the statement will be
"we find no evedence for a GWB associated with the GRBXXXXXX with our
preliminary analysis" then the review should be fairly straighforward.
If the statement is "we bound the strength of GWBs associated with
GRBXXXXXX to h less than YY" then substantial review may be required.
- Also: be aware that there is traditionally only one "preliminary"
result. So if preliminary S5 results were shown, we would not want to
have another set of preliminary S5 results with new calibration, etc.
The next set of results would have to be "final". This should be
considered when deciding whether a result is sufficiently mature
to be shown at APS.
- Unlikely that we will have results from the population study of S5
triggers for the APS presentation.
- Isabel has prepared new plots for the data conditioning sanity checks
[ HTML ].
- There is another set of plots comparing Tukey vs. Hann windows.
- There are corrected plots for the L1 time-series segmentation.
- Jolien is still worried about spectral leakage causing the data
not to be properly "whitened" at low frequencies. (See, e.g., Fig 4
of technical documentation, frequencies between 40 and 65 Hz.) This
is probably OK since these frequency should be suppressed.
Still, need to make sure that there is not any spectral leakage and
this probably needs to be done for each IFO and each data run (and
perhaps several epochs within the runs) since the nature of the ASQ
- Brian is worried about up-conversion and whether the spectrum
accurately captures this effect. Specifically in the 60 - 100 Hz
region where ther might be some difference between the 4s Hann
spectrum and the 1s Tukey spectrum. Again: would need to see average
spectra to clearly distinguish between two estimates.
- Jolien is still worried about discontinuities in the whitened
timeseries. ACTION: produce one whitened time series with one
set of transition times, then shift the data so that the transition
times are different and produce a second whitened time series. Overlay
these two series to see if the discontinuities are actually there in
the data or if they are associated with the data conditioning
- If discontinuities are due to data conditioning, need to establish
whether they significantly affect the CC statistic. Perhaps by comparing
distribution of CC stats that span a transition vs. those that do not.
$Id: minutes-2006-02-13.html,v 1.3 2006/02/13 22:21:08 jolien Exp $