Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.atnf.csiro.au/vlbi/dokuwiki/lib/exe/fetch.php/difx/deller_multifield.pdf Дата изменения: Wed Dec 7 23:20:08 2011 Дата индексирования: Tue Apr 12 07:25:19 2016 Кодировка: Поисковые слова: arp 220

Single-pass, multifield correlation in DiFX
Adam Deller ASTRON
5th DiFX workshop, Haystack Observatory


Multifield motivations
Phase must stay constant within averaging window!

Adam Deller

5th DiFX workshop, Haystack Observatory


Multifield motivations
l

Standard VLBI dataset (L band, 0.5 MHz channels, 2 seconds) FOV ~ 15"
l

Just 0.007% of the VLBA primary beam!

l

l

Retaining sufficient resolution for fullbeam images impractical primary beam: 30' Unnecessary; Time resolution: the vast majority 20 ms of the sky is phase centre Freq. resolution: 4 kHz empty at VLBI Smearing-limited field of view resolution! 12hr VLBA datase 30'
30,000 GB
5th DiFX workshop, Haystack Observatory

t:

Adam Deller


Multi-field imaging
Use a lowresolution "finder" catalogue to identify where to place the VLBI fields

primary beam Image: Random cutout, NRAO FIRST survey

VLBI fields still not to scale!

Adam Deller

5th DiFX workshop, Haystack Observatory


uv-shifting: the key to multifield
l

l

At any given instant, the phase centre of correlation can be changed by rotating the visibilities by a phase value equal to the LO frequency x delay (between desired phase centre primary beam and current phase centre) This is a station-based effect, but if done after some accumulation must be done separately to each baseline
uv-shifted "pencil" fields
5th DiFX workshop, Haystack Observatory

Adam Deller


Limitations
l l

"Pointing" a correlator involves appropriate delay and phase corrections "Re-pointing" correlated visibilities requires the appropriate differential phase change (fn of baseline/freq/time)
primary beam phase shift

This is ok

Desired image Smearing-limited field of view

Adam Deller

5th DiFX workshop, Haystack Observatory


Limitations
l l

l

"Pointing" a correlator involves appropriate delay and phase corrections "Re-pointing" correlated visibilities requires the appropriate differential phase change (fn of baseline/freq/time) But if the data primary beam already averaged Ruined too heavily in image phase shift frequency and time, Desired image smearing is too Smearing-limited severe field of view
5th DiFX workshop, Haystack Observatory

Adam Deller


How multifield is implemented
l

l

l

Provide separate geometric model for each phase centre (calcif2, vex2difx) Correlation is initially directed at the pointing centre, with high spectral resolution Then on ~10ms timescales (typically once per subint, but can be more frequent) shift is applied and only then are the channels averaged

Adam Deller

5th DiFX workshop, Haystack Observatory


Inside the Core
Subint visibility Repeat for each phase centre

Amplitude Phase

Rotate phase Rotate to new field

Average Average and further integrate

After ~10 ms
Adam Deller

5th DiFX workshop, Haystack Observatory


Multi-field imaging
primary beam primary beam phase centre phase centre Smearing-limited field of view phase shift Smearing-limited field of view primary beam phase centre Smearinglimited field of view primary beam

Correlate at high resolution for ~10ms

Apply uv shift

Average in frequency

Repeat for many phase centres

THEN: Repeat for next ~10ms (average in time)
Adam Deller 5th DiFX workshop, Haystack Observatory


Multi-field imaging: cost
l

l

Some computational overhead (factor of ~2.5) due to higher upfront spectral resolution, but additional fields are almost free (factor of <1.01) Thus efficiency gain increases as number of targets per pointing increases
l

For hundreds of sources, speed up factor > 100!

Adam Deller

5th DiFX workshop, Haystack Observatory


Multi-field imaging: nuisances
l

l

large FFT sizes -> difficult to get subint/int times that are "nice round numbers" Output data rates can start to be large! 10s of MB/s can push the capabilities of disk attached to the master node

Adam Deller

5th DiFX workshop, Haystack Observatory


Multifield in action 1: calibrators
Searching for in-beam calibrators around a pulsar (blue). The background is a FIRST image, detected calibrators are shown in magenta and the VLBA primary beam is shown in red.

Adam Deller

5th DiFX workshop, Haystack Observatory


Multifield in action 2: AGN

Chandra Deep Field South, E. Middelberg et al.

Adam Deller

5th DiFX workshop, Haystack Observatory


Questions?

Adam Deller

5th DiFX workshop, Haystack Observatory