Atacama Large Millimeter/submillimeter Array Expanded Very Large Array Robert C. Byrd Green Bank Telescope Very Long Baseline Array New VLBA capabilities with DiFX Wide-field imaging, multi-field imaging and more Adam Deller NRAO / UC Berkeley

VLBA Outline The DiFX software correlator and its usage with the VLBA New capabilities offered by DiFX compared to the VLBA hardware correlator: – Broad compatibility – Spectral/temporal resolution – Pulsar analysis – Commensal science – Wide-field / multi-field capabilities

VLBA The DiFX software correlator A C++ program running on commodity computer hardware (rack-mounted, multi-core servers) Development commenced in 2005, adopted by Australian Long Baseline Array in 2006, NRAO testing from 2008 and complete switch by December 2009 Supported by numerous libraries and applications for job configuration, FITS file building etc; ~10 active developers (NRAO, MPIfR, ATNF/Curtin, Haystack)

VLBA The DiFX software correlator

VLBA The DiFX software correlator Performance is good; hardware capable of supporting 10 stations x 512 Mbps would cost ~$12,000 in 2011 Low barriers to getting started has encouraged many adopters – Many contributors to code – This combined with ease of coding in C++ c.f. FPGAs has contributed to the rapid development of new features like the ones focused on today

VLBA Unique DiFX capabilities Compatibility, expandability – Initial reason for adoption - needed something capable of expansion to 4 Gbps system – incremental nature is extremely useful (hardware purchased in 4 stages, minimizing overall cost through Moore’s Law) – Handles all input/output VLBI formats Flexibility in parameter setting – Time, frequency resolution in particular

VLBA Unique DiFX capabilities Much more flexible pulsar processing (dynamic allocation of resources); allows pulse-phase dependant studies (binning) and “matched filtering” for recovery of optimal S/N from complex profiles

VLBA Unique DiFX capabilities Ease of adding new features has allowed low- overhead commensal functionality One such feature produces ms time resolution spectrometer and spectral kurtosis data The V-FASTR project has been approved to search for fast transient events during all DiFX correlations of VLBA data Real-time pipeline captures, re-orders and flags data and searches for dispersed pulses

VLBA Unique DiFX capabilities frequency time raw filterbank data bandpass, tcal corrected data

VLBA Unique DiFX capabilities V-FASTR has detected both normal and giant pulses from multiple (targeted) pulsars Running near full-time now Exploring an unknown area of parameter using a new technique at near-zero cost Highly visible pathfinder for SKA transient searches Also produces valuable RFI information for routine VLBA operations

VLBA Wide-field imaging DiFX is the most capable VLBI correlator in the world for wide-field imaging, due to the attainable time and frequency resolution primary beam: 30’ Smearing-limited field of view 15” phase centre Calculations for 1.6 GHz, total smearing = 10% Time resolution: 2000 ms Freq. resolution: 500 kHz 12hr VLBA dataset: 2.4 GB primary beam: 30’ Smearing-limited field of view 2’ phase centre Time resolution: 200 ms Freq. resolution: 50 kHz 12hr VLBA dataset: 240 GB

VLBA Wide-field imaging This ability has been widely used since the introduction of DiFX However, full-beam VLBA imaging is still a logistical impracticality Calculations for 1.6 GHz, total smearing = 10% primary beam: 30’ Smearing-limited field of view 30’ phase centre Time resolution: 20 ms Freq. resolution: 4 kHz 12hr VLBA dataset: 30,000 GB

VLBA Wide-field imaging Generally, however, the sky is almost entirely empty at VLBI resolution Thus, usually do not want “full beam” imaging; rather, many targeted small “fields” This can be achieved by uv shifting after correlation, but spectral/temporal resolution requirements are identical to imaging DiFX has moved the uv shift inside the correlator, allowing “multi-field” correlation and avoiding the logistical problem

VLBA Multi-field imaging primary beam Smearing-limited field of view Correlate at high resolution for ~10ms phase centre Apply uv shift primary beam Smearing-limited field of view phase centre phase shift Average in frequency primary beam Smearing- limited field of view phase centre Repeat for many phase centres primary beam THEN: Repeat for next ~10ms (average in time)

VLBA Multi-field imaging primary beam Image: Random cutout, NRAO FIRST survey VLBI fields still not to scale! Satisfactory “finder” catalogs already exist for most applications of this technique

VLBA Multi-field imaging 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 VLBA is unparalleled for multi-field VLBI applications due to homogeneous, relatively small dishes (large antennas or phased arrays reduce useful field of view)

VLBA Multi-field imaging For mJy-sensitivity secondary calibrator searches (me, later) with ~20 targets/pointing, net factor of 7 increase For sub-mJy sensitivity deep field AGN searches (e.g. Middelberg) with ~300 targets/pointing, net factor of ~100!

VLBA Multi-field imaging Efficient VLBI surveys of mJy and sub-mJy objects are feasible for the first time Middelberg et al. (2011) already published VLBA results on Chandra Deep Field South, more on the way covering variety of area and sensitivity ranges From Middelberg et al., 2011

VLBA Conclusions In addition to facilitating the ongoing sensitivity upgrade, DiFX has opened a number of new areas of parameter space for the VLBA – Advanced pulsar processing – Commensal transient observations – Wide-field and multi-field observations Of these, multi-field observations have the potential for opening up the most new applications - VLBI surveying is now practical