Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.naic.edu/alfa/ealfa/meeting1/minutes/palfa.html
Дата изменения: Mon May 8 23:01:36 2006
Дата индексирования: Sun Dec 23 01:45:21 2007
Кодировка:

Pulsar ALFA Surveys

Next: HIPASS Up: Minutes of the 1st Previous: NAIC Intro

Minutes of the talk on:
Pulsar ALFA Surveys

Jim Cordes (Cornell U.)

A more detailed version of this talk by the speaker is available in PDF or PPT format.

The pulsar consortium met on 1-2 November 2002 in Arecibo. A report is available on the web as is a summary of the talks presented at the meeting. At the meeting, we appointed a coordination committee with 8 members which I chair. The Point of Contact (POC) at AO is Paulo Freire.

Preliminary parameters for the pulsar surveys will produce on the order of 1 Pbyte of raw data over the course of 3-5 years. Long term archiving is an immediate challenge.

Targeted classes of pulsars

Young canonical pulsars (galactic plane)
Recycled pulsars/MSPs (out of plane)
High velocity pulsars
NS-NS and NS-BH binaries
Pulsars beyond the "death line" (radio magnetars)
Precessing pulsars
Globular cluster millisecond pulsars
X-ray and Gamma Ray pulsars

Why more pulsars?

Our objectives are to try to double the number of known pulsars. We hope to find 1000 more in the galactic plane alone. More pulsars will include extreme pulsars, say those with

P 1ms P 5sec (long ones interesting too)
P $_{\rm orb} <$ hours.
V 1000 km/s
B $> 10^{14}$ G.

Study of such objects are of interest for numerous applications:

Population and stellar evolution.
As ISM probes.
High energy connections (GLAST).
Physics payoff (liSA, liGO, etc) esp GR.

Pulsar search requirements

100 million time samples
1000 channels
200 times real time to process data (Beowulf cluster; actually cluster of clusters) but we think we can keep up with real-time rate.
Have to compensate for dispersion (unknown); dedisperse.
- Pulse broadening from multipath; strongly frequency dependent; good for L-band but impacts/limits integration time.

Dmax ve Flux density threshhold suggests a low integration time which thus leads you to be luminosity limited. But as you increase t $_{\rm int}$ , you become limited by first dispersion and then by scattering. Like everyone, we will need to trade off to develop optimal survey strategy.

Possible Surveys

Deep survey of galactic plane of available (to Arecibo) longitudes covering 3 to 5 degrees in latitude
Less deep intermediate latitude
Deep surveys towards specific objects
- high energy targets (Multibeam for RFI)
- extended targets
Extragalactic targets, for example M33
Piggyback (with other projects)

Requirements for Galactic plane:

300 sec per position; takes about 2000 hours
400 TB of data in 2000 hours,

This translates to about 3 years at 50% of galactic plane R.A. Of course, such a large allocation has to be discussed in light of other programs using Arecibo but note that NRAO has a large survey projects benchmark number of 50% of time in any RA range.

Comparison of AO, GBT and Parkes

The Parkes Multibeam Pulsar Survey adopted an integration time of 2100 sec. Surveys with ALFA need only 20 sec to match this sensitivity, but we want 300 sec to go much deeper. A simulation of a galactic plane search with ALFA predicts that we should detect 1000 new pulsars. A plot of the distribution in galactic coordinates of what ALFA surveys should detect in the galactic plane compared with other surveys shows that an ALFA survey would go much deeper into the Galaxy than other surveys.

Spectrometer

For pulsar surveys, we require the following in a spectrometer:

1000 channels with < 0.3 MHz channels.
Need fast dump capability.
FPGA Correlator or FPGA-FFT or Polyphase filter approach.
Polarization summing mode.

A decision needs to be made rapidly on how such a device can be constructed in time for ALFA.

Intermediate latitude survey

Another possible pulsar survey would be conducted at intermediate latitudes. It would require about 1500 hours of telescope time. The dwell time more like 60 sec, with fair amount of flexibility.

Issues for optimizing surveys.

We have identified a number of issues that must be explored before we settle on the survey strategy:

RFI management
- characterization
- test obs & algorithm development
- multibeam schemes
Diffractive ISS:
- multiple passes favored for low DM,
- v-t weighting for intermediate DM,
- no action for high DM.
Refractive ISS:
- muiltiple passes for low to intermed DM (Even 2 would do).
Nulling requires multiple passes.

"search" vs. "confirmation": Historically, these are two different phases. Parkes spent as much time confirming as they did in the initial survey.

What next?

Further survey simulations
Design at the telescope survey modes:
- beam interlace, hour angles, feed rotation.
RFI studies, pilot obs, simulations.
Search code development (like tempo, not aips++) - want a single pipeline.
Data management plan.
Followup observation plans.

Defined 5 working groups:

Surveys
Data Acquisition
Post processing
Data management
Follow up observations

At our meeting, the pulsar consortium developed a final formal structure organizational chart with organization, coordinating committee and tasks assigned to each subcommittee.

Chair selection: JMC was elected temporarily, 2 week period afterwards process for election but no one volunteered. Further process needs to be specified; one idea is for rotating committee chairs.

Preliminary protocols:

Membership: open policy early on, but by application later
Protection of student projects
Data access: open to all members during priority period NAIC policy 18 months, but probably will be reduced to more like 6 mos. There was some resistance to a shortened period within consortium. Access may also be granted during the proprietary period by application from nonmembers.
Uniform, baseline processing to produce a legacy product is the goal but we also hope to encourage innovative new approaches.
Authorship: rotating lead, equitable, including all consortium members but with the encouragement that inactive members opt out (honor system)
Follow up observations: similar to Authorship
Discovery of exotica: full consortium involvement

Data Management

Raw data
- Local processing (inc. quicklook), processing at consortium member institutions
- Will have short and long-term archiving (disk/tape).
- Perhaps at a central mainland location with high bw-pipe?
- Database catalog system with web based data selection.
Intermediate data products include candidate lists, RFI identification, diagnostic plots
Final productions (catalogs, pulse profiles, timing models).
- Implied linkage to the NVO as appropriate.

Pilot Database Storage with Cornell Technology Center

Database information for pilot:
- Microsoft SQL Server
- Hosted at CTC
- Stores both raw data and heavily processed data.

Discussion/Questions:

Murray L.	Could you make your survey observations during the daytime?
Jim	Probably. It largely depends on RFI.
Don C.	Could you reduce integration time/pass?
Jim	Yes. But at a cost of course.
Steve S.	Could you offset half a beam?
Jim	Yes.
Chris S.	What the draft guidelines said was that 18 months is the current proprietary period. It was stated then that the issue needed to be discussed.
Riccardo G.	The proprietary period at other national observatories is zero for large projects.
Jim	The proprietary period for GRO was one year.
Lister	Would you consider a scanning survey with telescope slowly moving.
Jim	It depends on how slow. Is that for sampling issues?
Lister	Yes.
Jim	If it were slow enough. It would add more complexity to processing but we can deal with motion.
Phil P.	1PByte costs about 1 dollar per GB, so your archiving needs cost at least 1 million dollars.
Jim	True. We don't need it all at once.
Desh	The typical pipeline delay with regular proposals is 6 months from submitting proposal to getting on telescope.
Jim	One of the issues we discussed is what happens, in the case that the raw data were available immediately, if a rogue pulsar group comes along and downloads all the data.
Phil P.	How long would it take to download that amount of data?
Jim	What if it could be stored locally.
Murray	Could you expand a bit on the thinking on protection of student projects?
Jim	Say we acquire a certain block of data and a PhD student will do follow up. We would like to insure some protection so that no other consortium member could do that project.
Murray	But not protection of data.
Jim	Right.
Desh	So protection is really for the follow-up rather than for the raw data.
Jim	In a sense yes.
Riccardo	What is mimumum acceptable integration time for the high latitude survey?
Jim	People have done drift scans (12 sec/beam), but they have not been all that successful. However, another very interesting option might be to look for transients even for a few seconds, especially if there were multiple passes of the same region.