Data acquisition
With ALFA, we need about 47 pointings to cover one square degree,
compared to about 330 pointings needed to cover one square degree with
similar density with a single-pixel feed. Up until now, we have
been using the Wideband Arecibo Pulsar Processors (
WAPPs) to detect the signal from
ALFA's seven beams. These cover 100 MHz of band (with dual
polarization capability), initially centered at 1420 MHz and now at
1440 MHz. For search purposes, we produce 256-channel spectra every 64
microseconds.
From August 1 to October 8 2004, we conducted a preliminary survey that
covered the two regions closest to the Galactic plane (|b| < 1
degrees) visible from Arecibo: the "Inner Galaxy" (40 < l < 75
degrees) and the "Anti-center" (170 < l < 210 degrees). Each
pointing
was 134 seconds for the Inner Galaxy and 67 seconds for the
Anti-center. This was done in sparse mode, where we do only 1/3 of the
pointings needed to cover the whole region. This preliminary survey
found a total of 11 new pulsars, and detected 30 previously known
pulsars. For a detailed description of this survey, and the strategy
of the present survey, see
Cordes
et
al. (2006).
Since March 16 2005, we are conducting the main ALFA pulsar survey
with the WAPPs. This will cover the Galactic plane (|b| < 5
degrees) visible with the Arecibo 305-m radio telescope (35 < l
< 75 degrees), it is unclear what the latitude coverage will be in
the Anti-center. Each pointing lasts about 268 seconds in the Inner
Galaxy and 134 seconds in the Anti-center.
In 2009, the survey transitioned to new and improved back-ends, the Mock
polyphase filterbank spectrometers, which are capable of covering 300
MHz (from 1225 MHz to 1525 MHz, the bandwidth covered by ALFA) for
each of the seven beams (see detailed technical
specifications
here).
This will lead to greatly increased search sensitivity, provided we
can effectively deal with all the radio frequency interference.
PALFA pulsar survey coverage and PRESTO data
processing status as of December 2010, in Galactic coordinates (center
of the Galaxy is at (0,0)). The stars indicate positions of known
pulsars, grey indicates observed pointings with the WAPPs, while red
indicates pointings processed by the PRESTO pipeline so far.
Data processing and storage
Many of the detections to date have been made with a quick reduction
package that allows us to find pulsars almost in real time. This is
made possible by reducing the spectral and time resolution by a factor
of 16, and using a computer cluster,
the
Arecibo
Signal Processor to search for pulsars in the data. This is a
nice and quick way of detecting slow pulsars, but the sensitivity to
fast pulsars is severely degraded. Re-processing these data with full
resolution is, computationally, a very challenging task but is
essential for detecting many fast (both young and recycled) pulsars so
far hidden by Galactic plasma.
It is expected that, over the next several years, this survey will
generate over
1000 Terabytes of data. The data is stored at
the
Cornell University Center
for Advanced Computing. The full-resolution raw data is processed
independently by three software pipelines.
The Cornell University pipeline does a standard periodicity search and
single-pulse search without doing an acceleration search. It has been
run on all WAPP data archived at the Cornell University Center for
Advanced Computing and has provided 2.5 million signal
candidates. Winnowing of this vast set of candidates is currently under
way.
The second pipeline is based
on
PRESTO, a
large suite of pulsar search and analysis software developed by Scott
Ransom. It employs a Fourier-Domain acceleration search technique,
which compensates for the loss of detection sensitivity in a
traditional periodicity search due to a rapidly changing frequency of
the periodic pulsar signal. Such frequency modulation can occur, for
instance, due to a pulsar's orbital motion in a compact binary. This
approach thus significantly boosts sensitivity to binary pulsars. The
PRESTO pipeline is run on dedicated clusters at several institutions
that participate in the ALFA survey, producing ~3 million signal
candidates so far.
Since March 2009, part of the
Einstein@Home computing power is
used to analyze WAPP data. The Einstein@Home algorithm is particularly
sensitive to radio pulsars in tight binary systems, with a phase-space
coverage that is complementary to that of the PRESTO pipeline. To
date, it has re-detected 123 known radio pulsars as well as two
previously unknown pulsars.
The data processed thus far has revealed that the radio frequency
interference (RFI) environment at Arecibo significantly affects the
detection threshold of the survey, creating unforseen challenges in
identifying the many weak pulsars that are likely lurking in the
data. To address this, the PALFA consortium is actively developing
novel techniques for identification, mitigation, and excision of RFI.
We are also implementing a variety of heuristics as well as machine
learning algorithms for identifying real pulsars among the millions of
signal candidates, most of which appear to be due to RFI. The
inevitable growth in the incidence and variety of man-made RFI
suggests that this problem will likely be important for all future
radio pulsar surveys.
Outreach Efforts
The
Arecibo
Remote Command Center (ARCC) at the University of Texas at
Brownsville and the University of Wisconsin at Miwaukee is currently
engaged in searching for radio pulsars in ALFA data. ARCC is an
integrated research/education facility that allows students at the
high school and undergraduate level to be directly involved with the
research at the Arecibo telescope. Web based tools have been
developed so that students could rank the pulsar candidates created
by the PRESTO analysis.
