Personal tools
You are here: Home Virgo Data Analysis VDAS vdas_20111215
Log in

Forgot your username or password?
Document Actions


by Marie Anne Bizouard last modified 2011-12-16 06:29


1. Announcements:

2. Current issues:
    • Data transfer/data storage @ Cascina
    • Pre-staging tools @ CNAF: Demand from CNAF to be answered
    • NINJA data set transfer - update about the transfer with gridftp
    • ROOT @ Lyon

3. Software Engineering Run 1 preparation: projects + organisation Software Engineering Run 1 (Nicolas)

4. Special topic:
- NMAPI & NOEMI (Elena)


MAB(minutes), Alberto, Loic, Elena, Benoit, Nicolas, Franco, Livio,


Current issues:
  • Data transfer/data storage @ Cascina: nothing to report
  • Pre-staging tools @ CNAF: agreed to have pre-staging tools. MAB will answer to CNAF.
  • NINJA data set transfer - update about the transfer with gridftp: wait for new NINJA files
  • ROOT @ Lyon: MAB asked if any version is prefered. Loic mentioned the request came from Michel Yvert. Not sure if he has special request. MAB hacked the v5r26p0 version of the Virgo root package to point to version 5.30.00 of root at Lyon as a temporary workaround.

ER1: Nicolas summarized the discussion of last week LV meeting. See email sent 1 week ago:
attendance : Patrick, Carsten, Ryan, Chad, Nicolas, Benoit, Chris, Didier

* Virgo data will be gaussian, based on latest broadband virgo noise
curve which will be released very soon.  Not very different from

* LIGO data will be gaussian, using a reference aLIGO noise curve that
gives the detectable rate ~40 per year for neutron stars. Probably the
curve used in the rates paper.

* Each instrument will have ~80% duty cycle with science segment
durations distributed uniformly between 1/2 hour and 1 day.

* Each instrument will have a state vector channel to indicate this
information. For LIGO, will be the same as S6 online h(t) statevector.
  For Virgo, will be the same as VSR3 online h(t) state vector.

* Virgo will also provide a simulated DQ stream with ~3% deadtime (flat between .
LIGO will not have a low-latency DQ system for ER1.

* Basic segment information will be inserted into the two segment databases.

* Astrophysical signal populations will be present in the low-latency
h(t) data streams. These will be blind during the run. Noise only data
will be released weeks after the run at which point the astrophysical
signal populations will necessarily be unblinded.

* Hardware injections, i.e. known signals added to the data to mimic
the presence of hardware injections, will be included in pre-run data.
A high rate of burst hardware injections may adversely effect the
detectability of astrophysical compact binary signals. The proposal is
to have a couple of pre-run days without hardware injections, a couple
of pre-run days with hardware injections as in S6/VSR3, and then ramp
down hardware injections as the run begins.

* The mechanics of data generation is: 1) noise generated by
independent LIGO and Virgo teams (in advance for LIGO and onfly for Virgo;
2) astrophysical and hardware injections generated in advance;
3) for Virgo, injection only frames will be transferred to Cascina
where they will be added to the simulated noise;
4) for LIGO, combined noise + signals frame files will be made
available to John Z who will ingest them into the low-latency transfer network.

* Deadlines:
  - 21 Dec 2011:  1 day of test data to be available
  - 6 Jan 2012: pre-run data to be available
  - 13 Jan 2012: full run data available

* Data preservation:
  - low-latency h(t) frames to be preserved
  - signal only injection frames to be preserved
  - noise only frames to be preserved
  - clean up of all intermediate data products 6 months after run

Disagreement about injections between burst and cbc groups. Still under discussions.


NOEMI on LIGO data: several scenarios. Transfer of data at Cascina or run on LIGO cluster. All 3 scenarios have technical difficulties.
NMAPI provides a common infrastructure for all noise monitors that are plugins. No clear agreement of other developers to use this interface. Instead of starting from scratch, MAB proposed to check with each existing package developers what are the plans for the future. The priority is not to re-develop existing tools but to consolidate what exists, check the requirements and identify the missing tools/activities. Non linear noise study is a clear missing topic for AdV.
Elena said she already started to prepare a list of package. MAB will work on that.

Discussions for NOMEI: concurrent data access is still an issue. Can read many channels in parallel from the same core. Solution proposed by Stefano: extract the channels and store them on local disk which are 37 GB big.

Action item: check if this could solve the problem at cascina with Virgo data