(Cosmological Acceleration and Radio Pulsar Experiment—Latin for 'seize')
CARPE is a compact radio telescope designed to observe gravity waves through precision pulsar timing, discover new pulsar systems through sensitive all-sky pulsar surveys, and measure cosmology and dark energy from redshifts of 1–4. CARPE leverages recent developments in correlator technology and wideband antennas to observe from 350–1500 MHz with ~2.5 thousand small steerable antennas, instantaneously observing ~400 square degrees of the sky with 400 MHz of bandwidth in ~10 thousand frequency channels. The very compact arrangement of ~200 meters provides very high surface brightness sensitivity, and the MOFF correlator produces fully calibrated full Stokes images and forms calibrated pulsar beams across the field of view.

CARPE DE (dark energy)

CARPE is designed to observe cosmological acceleration and dark energy from a redshift 1<z<4 using the 21 cm intensity mapping technique. In the next few years a large optical program such as JDEM is likely to tackle the problem of dark energy via spectroscopic galaxy surveys, supernova surveys, and/or weak lensing measurements.  If dark energy behaves like a cosmological constant, then its effect on the Hubble expansion is dominant only at z<1 and becomes negligible at z > 2. In this case, studies of the expansion history at low redshifts by JDEM would provide the most powerful measurement.

However, the origin of the dark energy is not understood, and so it is not known a priori which redshift range should be studied in order to provide optimal constraints on possible theories for it. CARPE is a concept for a new radio array that would greatly extend the redshift range over which the expansion history of the Universe can be studied with small and independent systematic biases.

After the Epoch of Reionization the hydrogen in the intergalactic medium is highly ionized, and the majority of the HI resides in dense pockets of self-shielded hydrogen, such as damped Lyman-alpha absorbers and Lyman-limit systems. The Lyman-alpha opacity is determined by the volume averaged HI fraction, and is observed to be below ~104 after reionization. However, the 21 cm emission is determined by the HI mass fraction. Due to the association with dense systems the HI mass fraction remains at a few percent from reionization to the current epoch, and the techniques developed for the Epoch of Reionization power spectrum observations can be applied to diffuse HI structure observations at intermediate redshifts.

CARPE would target statistical measurements of the matter power spectrum and large scale structure between 1<z<4.  The observations will probe the expansion history through both the baryon acoustic oscillations (BAOs) and the Alcock-Paczynski (AP) effect. 21 cm observations with CARPE have four strengths:

  • Probe a large redshift range, 1<z<4.  No other probes for precision cosmology are currently being applied to this cosmic epoch (Corasaniti et al., 2007) and redshifted 21 cm observations might be the first working method to effectively probe the dark energy at z>3.
  • Sensitive to both transverse and line-of-sight BAO scales, and the AP effect.  In contrast, spectroscopic galaxy redshift surveys are more sensitive to the transverse scale, while photometric redshift surveys will not be sensitive to the radial BAO at all (Glazebrook & Blake, 2005).
  • Low bias and independent systematics.

CARPE Pulsars

In terms of sensitivity CARPE has comparable area to the GBT and its sensitivity is significantly higher than the Parkes Multibeam Survey. The frequency range of 300–1400 MHz places it well to find both local sources at low luminosities at lower frequencies (where scattering is worse but pulsars are stronger) and more distant, brighter objects at higher frequencies (Smits et al., 2008).  The extremely large FOV acts as a multiplier, making the system work like many thousands of individual telescopes (compare to 13 for the Parkes multibeam), rapidly surveying the whole sky and aiding in RFI rejection. The number of CARPE pulsar beams is limited by the processing capability of the pulsar backend (see MOFF correlator). We will be able to repeatedly survey the sky, aiding in detection of sporadic sources like the Rotating Radio Transients (RRATs;  McLaughlin et al. 2006) and intermittent sources (Kramer et al., 2006), and also rapidly precessing systems like the double pulsar.  We will also be able to do repeated, deep searches of the Magellanic clouds in single pointings (assuming a Southern Hemisphere site).  Given the good bandwidth (especially at lower frequencies), wide frequency range, and enormous FOV, pulsar searches with CARPE will directly complement recent and ongoing surveys: it will repeat the Parkes Multibeam Survey with better time and frequency resolution, and deeper pointings, and can similarly probe deeply at low frequencies for a larger part of the Galactic plane than the GBT350 survey.  The exact survey details — frequencies, coverage strategy, processing parameters — remain to be determined with detailed simulations, but we are confident that CARPE will prove very powerful.  Concurrent advances in pulsar processing (using dedicated digital electronics like the GUPPI system; GPUs; Ransom et al. 2003) will hopefully keep pace with the enormous increases in data-volume that CARPE can provide.

CARPE GW (gravity waves)

Once the pulsars are found, many detailed studies require dedicated pulsar timing observations to probe emission physics, establish orbital parameters, and test gravity.  Here, the large area and FOV will allow multiple pulsars to be timed simultaneously.  This will help refine pulsar ephemerides and remove systematic effects, with likely dramatic improvements to our sensitivity and ability to detect gravity waves. The sensitivity of a pulsar timing array to gravitational wave detection increases with the number of pulsars in the array and decreases with increased timing precision. CARPE will provide improvements on both of these fronts by providing more pulsars and by increasing precision through longer pulsar observations across a large bandwidth, essential for correcting for propagation effects.

Enabling technologies

Two of the key CARPE technologies are the broadband feed and the MOFF correlator. antenna

It is difficult to construct an antenna with a very broad feed, a stable focus point, and good impedance match (choose 2 of 3). Rich Bradley is leading the design of the CARPE antenna, and is looking at broadband fractal feeds that look directly at the sky. The lack of a reflector removes the constraint needing a stable focal location, and reduces concern of beam spill over the dish edge. The dual polarization sinuous cone (figure at right) has a very good impedance match and stable beam pattern over the 350–1500 MHz frequency range of CARPE. To increase the collecting area at higher frequencies and avoid grating lobes, a second set of sinuous cones sensitive to only the higher 750–1500 MHz frequency range are inserted at the interstices in a Sierpinski fractal pattern (below right).

The MOFF correlator (Morales 2009) combines the calibration and gridding typically done post-correlation as a part of the correlation operation. The result is an imaging correlator that scales as  NlogN with the number of antennas (as opposed to N2) for compact interferometers and produces a fully calibrated optimal map output. This greatly reduces the correlation and post processing requirements, and enables instruments like CARPE with many thousands of antennas. In addition, a fully calibrated electric field image is formed as an intermediate product within the correlator. The electric field pixels can be read out as calibrated full Stokes pulsar beams, and the number of pulsar beams is limited only by the output bandwidth (how much data you can process). In addition we exploring ideas for performing de-disperison within the correlator and modularizing the correlator so CARPE can be built in subunits, with the correlator naturally growing as additional ‘tiles’ of the CARPE instrument are completed.


A technical talk by Miguel Morales at the University of Washington (keynote, pdf, both files are quite large).