Pulsar Surveys

(Bhaswati Bhattacharyya, Jayanta Roy, Yashwant Gupta, Bhal Chandra Joshi)

Pulsars have tremendous untapped potential to probe the behaviour of matter, energy, space and time under extraordinarily diverse conditions. Even though pulsars are frequently getting discovered with ongoing surveys at major telescopes over the world, the presently known pulsar population is only 1\% of that predicted by stellar synthesis models. Modeling by Faucher-Giguere et al. (2006) indicates that the Galaxy contains about 100,000 pulsars, while only a few thousand are known today! This implies that the vast majority of pulsars is waiting to be discovered.
Studies of pulsars yield a better understanding of a variety of physics problems, from acceleration of particles in ultra strong magnetic fields (primarily via the study of the emission properties of normal pulsars, having spin period > 30 ms) to probes of ultra-dense matter (mostly via studying the timing properties of millisecond pulsars, having spin period <30 ms, that are very stable rotators). The fractional rotational stability of millisecond pulsars (MSPs), one part in a million billion, is comparable to that of atomic clocks (Lorimer et al. 2004). Such rotational stability, compactness second only to black holes, and their presence in binary systems, make MSPs ideal laboratories to test the physics of gravity and as detectors for long-wavelength gravitational waves. Moreover, MSP evolutionary processes can be tracked through individual interesting discoveries of MSPs in special evolutionary phases (e.g. Roberts et al. 2011, Archibald et al. 2009; Roy et al. 2015). In addition to the regular radio emission from pulsars, millisecond transient bursts, called Fast Radio Bursts (FRBs; Lorimer et al. 2007, Thornton et al. 2013) are observed at the locations of dynamic events, making them useful probes for extreme matter states.
The known pulsar population is increasing with the surveys running at major single dish and array telescopes around the world: e.g. the High Time Resolution Universe (HTRU) survey at Parkes (Keith et al. 2010), the HTRU-North survey at Effelsberg (Barr et al. 2011), the SUrvey for Pulsars and Extragalactic Radio Bursts (SUPERB) survey at Parkes, the Pulsar survey Arecibo L-band Feed Array (PALFA, Cordes et al. 2006), the Green Bank Telescope drift scan survey (Boyles et al. 2013), the Green Bank Northern Celestial Cap (GBNCC) survey (Stovall et al. 2014), the Arecibo all-sky 327 MHz drift pulsar survey (AO327, Deneva et al. (2013)), the LOFAR Pulsar Pilot Survey (LPPS) survey (Coenen et al. 2014), the LOFAR Tied-Array All-Sky Survey (LOTAAS), and the Fermi-targetted MSP surveys (Ray et al. 2012). Because of the generally steep spectral nature of pulsars, lower frequencies are an obvious choice for searching for fainter pulsars away from the Galactic plane, where the search sensitivity is not severely affected by sky temperature and increased scattering. Such surveys away from the Galactic plane will detect relatively older pulsars.
The high sensitivity of the GMRT at low frequencies (<~ 800 MHz) makes it an outstanding telescope for pulsar surveys; such surveys are hence an important research area at NCRA. With the aid of reduced quantised noise and high time resolution supported by the flexible GMRT Software Backend (GSB; Roy et al. 2010), the GMRT search sensitivity for MSPs improved significantly, resulting in the discovery of 8 MSPs from Fermi-directed searches (Bhattacharyya et al. 2013, Roy et al. 2015).

The GMRT High Resolution Southern Sky survey (GHRSS)
The outstanding GMRT potential for low-frequency pulsar surveys has been recently emphasized by the GMRT High Resolution Southern Sky (GHRSS) survey, a low-frequency survey for pulsars and transients away from the Milky Way's plane. The GHRSS survey covers Galactic latitudes |b|>5 degrees, scanning the southern sky, with declination -40 degrees to -54 degrees. This declination coverage is complementary to the coverage of other ongoing low-frequency sky surveys around the world. The first phase of the GHRSS survey was carried out using the narrow bandwidths of the GMRT Software Backend, at 322 MHz, and has already resulted in the discovery of 13 new pulsars (Bhattacharyya et al. 2016). The second phase, using the GMRT Wideband Backend and the 250-500 MHz receivers of the upgraded GMRT is now under way.

Recent Results
The GMRT High Resolution Southern Sky Survey for Pulsars and Transients. IV. Discovery of Four New Pulsars with an FFA Search
Fast Folding Algorithm (FFA)-based searches are known to be more efficient in searches for isolated long-period and low duty-cycle pulsars. The reprocessing of the GMRT High Resolution Southern Sky (GHRSS) survey data with a newly implemented FFA search pipeline resulted in the discovery of six new pulsars. Singh et al. (2023) reported that three of these pulsars have very low duty cycles, within the bottom 1% of the pulsar duty cycle distribution. The figure shows the six new pulsars along with 1181 pulsars from the known population on a plot of duty cycle versus pulsar rotation period. This finding highlights the efficiency of FFA-based searches to discover low duty cycle pulsars. The new discoveries also include an extreme nulling pulsar with a nulling fraction of 90%. There are only a few other pulsars in the currently known population that show such extreme nulling. Considering the discoveries of many long-period pulsars in recent years, mostly in single-pulse searches, the authors anticipate that a fainter population of such long-period pulsars is still waiting to be discovered. A periodicity search will be needed to recover this putative faint long-period pulsar population. FFA-based searches with their superior sensitivity are best suited for searching for such pulsars with long periods and low duty cycles. Singh et al. (2023) also recommend a significant increase in the integration time per pointing in major pulsar surveys to recover the fainter population of long-period pulsars that are not detectable in single-pulse searches.
A novel greedy approach to harmonic summing using GPUs
A convenient and computationally efficient way of detecting pulsars in a time-domain search is to use the technique of Fourier transform. The Fourier transform distributes the power contained in the pulsar’s signal into the fundamental frequency bin and multiple higher harmonic bins. The incoherent harmonic sum aims to increase the pulsar’s signal-to-noise ratio (SNR) by accounting for power (either fully or partially) from an increasing number of harmonics. However, in such harmonic sum algorithms, there are limitations due to unfavourable memory access patterns (which significantly reduces data utilisation per cache-line) and the number of possible partial sums explored per a single fundamental bin. For example, the Lyne-Ashworth algorithm (described in Lorimer and Kramer 2004) used in the SIGPROC software package, sums only powers-of-two harmonics. The harmonic sum algorithm used in the PRESTO software package (Ransom 2002) also uses a subset of the harmonics. This paper reports a new harmonic sum algorithm based on a greedy approach and implementation of this on NVIDIA GPUs using the CUDA programming language. This algorithm determines which time samples to sum according to the short-term gains rather than finding the optimal sum of harmonics. The Greedy harmonic sum considering all harmonics, not only powers-of-two, achieves higher sensitivity with a performance similar to or higher than the standard harmonic sum algorithms. The figure presents a comparison between the Greedy harmonic sum algorithm and the PRESTO harmonic sum algorithm (which sums only even harmonics), its two modified versions, one that sums elements of all higher harmonics (PRESTOall) and another that performs the harmonic sum for all harmonic orders in addition to summing all higher harmonics (PRESTO+), and a simple harmonic sum which ignores the drift and sums only integer multiples of the fundamental bin.  The Greedy algorithm encounters minimum sensitivity loss as a function of pulse frequency similar to PRESTO+. The GMRT High Resolution Southern Sky (GHRSS) survey data analysis with Greedy Harmonic sum reports 10-30% more recovered SNRs than with PRESTO. The compute performance of the new algorithm in terms of the number of fundamental frequency bins processed per second is 10% faster than PRESTO and >50% faster than the updated version of PRESTO (PRESTO+). Thus, the new Greedy harmonic sum algorithm has lower signal loss and better-recovered SNR than the standard algorithm used in PRESTO while achieving similar or better performance in the number of processed fundamental frequency bins per second.
The GMRT High Resolution Southern Sky Survey for Pulsars and Transients. III. Searching for Long-period Pulsars
Standard pulsar radio emission models predict a critical value of period derivative (P_dot) corresponding to the spin period (P) of pulsars below which radio emission ceases. These critical values of period and period derivative trace a curve on the P-P_dot plane called the death-line. No radio pulsar should exist below this line. The discovery of long period pulsar J2144-3933 and its location on the P-P_dot plane has questioned all the existing radio emission models, but there are only a handful of such interesting objects. Over the last decade, the number of millisecond pulsars has increased four-fold, whereas there has been only a marginal increase in the number of long-period pulsars. Along with intrinsic and observational biases, susceptibility of conventional fast Fourier transform (FFT)-based searches to red noise can be the primary reason behind the lack of long period pulsars. Searching for periodic non-accelerated signals in the presence of ideal white noise using the fully phase-coherent fast-folding algorithm (FFA) is theoretically established as a more sensitive search method than the FFT search with incoherent harmonic summing. Some major pulsar surveys (e.g. SUPERB and PALFA) have implemented FFA search to get optimal sensitivity for long period pulsars. This paper reports a detailed comparative study of FFA and FFT search sensitivity under various noise conditions (ideal white noise, real telescope noise, and simulated red noise) and over a range of signal parameters (period, duty-cycle, and profile shape). Singh et al. find that the FFA search with appropriate de-reddening of the time series performs significantly better than the FFT search with spectral whitening for long-period pulsars under real noise conditions. They describe an implementation of an FFA-based search pipeline for the GMRT High Time Resolution Southern Sky (GHRSS) survey. With processing of 1500 square degrees of GHRSS sky, the paper reports the re-detection of 43 known pulsars and the discovery of 2 new pulsars. Panel (a) of the figure shows a comparison of FFA and FFT detection signal-to-noise (S/N) of these pulsars. All these pulsars are better detected in the FFA search and the long-period pulsars have a higher ratio of FFA to FFT detection significance. Five of these pulsars were missed by the FFT search. Panel (b) of the figure shows the time versus phase and folded profile plots for a newly-discovered pulsar J1517-31b, with a period of 1.1 s and at a DM of 61.7 pc-cm^{-3}. This pulsar, with a long period and an unusually narrow duty-cycle, was missed by the FFT search. The authors find that the FFA search can reduce the algorithmic bias against long-period pulsars. Increased observing time per pointing along with the implementation of the FFA search in major pulsar surveys will possibly recover the missing population of long-period pulsars and populate the region close to the death-line in the P-P_dot plane.
Serendipitous Discovery of Three Millisecond Pulsars with the GMRT in Fermi-directed Survey and Follow-up Radio Timing
Only a minor fraction (~15%) of the known pulsars spin with millisecond periodicity. The intrinsic faint nature of millisecond pulsars (MSPs) have hindered the discovery of these objects. This paper reports the discovery of three MSPs: PSRs J1120-3618, J1646-2142, and J1828+0625 with the Giant Metrewave Radio Telescope (GMRT) at a frequency of 322 MHz using a 32 MHz observing bandwidth. These sources were discovered serendipitously while conducting deep observations to search for millisecond radio pulsations in the directions of unidentified Fermi Large Area Telescope (LAT) gamma-ray sources. Phase coherent timing models for these newly discovered MSPs were derived using ~5 yr of observations with the GMRT. These are plotted in the figure, where the red points denote the timing residuals at 322 MHz and blue at 607 MHz. PSR J1120-3618 has a 5.5 ms spin period and is in a binary system with an orbital period of 5.6 days and a minimum companion mass of 0.18 solar masses, PSR J1646-2142 is an isolated object with a spin period of 5.8 ms, and PSR J1828+0625 has a spin period of 3.6 ms and is in a binary system with an orbital period of 77.9 days and minimum companion mass of 0.27 solar masses. The two binaries have very low orbital eccentricities, in agreement with expectations for MSP-helium white dwarf systems. Using the GMRT 607 MHz receivers having a 32 MHz bandwidth, PSR J1646-2142 and PSR J1828+0625 were detected but not PSR J1120-3618. Spectral indices for these MSPs using the GMRT observations are reported in this paper. PSR J1646-2142 has a wide profile, with significant evolution between 322 and 607 MHz, whereas PSR J1120-3618 exhibits a single peaked profile at 322 MHz and PSR J1828+0625 exhibits a single peaked profile at both the observing frequencies. These MSPs do not have gamma-ray counterparts, indicating that these are not associated with the target Fermi LAT pointing. This emphasizes the significance of deep blind searches for MSPs. The serendipity of the discovery of these millisecond pulsars indicates a population of weak MSPs waiting to be discovered with deep enough blind searches.
The GMRT High Resolution Southern Sky Survey for pulsars and transients -II. New discoveries, timing and polarization properties
Bhattacharyya et al. have been carrying out the GMRT High Resolution Southern Sky (GHRSS) Survey for pulsars and transients since 2014. In this paper, they report the discovery of three pulsars, PSRs J1239−48, J1516−43 and J1726−52. They also present long-term timing solutions for three pulsars previously discovered with the GHRSS survey: (1) PSR J2144−5237, a millisecond pulsar with a period of 5 milli-seconds in a 10-day orbit around a companion of mass 0.18 solar masses, (2) PSR J1516−43, a mildly recycled 36 milli-second pulsar in a 228 day orbit with a companion of mass ∼0.4 solar masses, and (3) the 320 millisecond pulsar PSR J0514−4408. For PSR J0514−4408, Bhattacharyya et al. discover pulsed gamma-ray emission. In addition, they report radio polarimetric observations with the Parkes telescope for three of the GHRSS discoveries, PSRs J0418−4154, J0514−4408 and J2144−5237. The top panel of the figure shows a 322 MHz radio profile (red) of PSR J0514−4408 (discovered in the GHRSS survey) plotted with the Fermi Large Area Telescope (LAT) gamma−ray profile (using ∼9.2 years of LAT data above 100 MeV). The bottom panel of the figure shows the gamma-ray pulsation from PSR J0514−4408.
Discovery of PSR J1227-4853: Transitioning from Low-mass X-Ray Binary to Redback Millisecond Pulsar
Low-mass X-ray binaries (LMXB's) and radio millisecond pulsars (MSP's) are linked through stellar and binary evolution, where MSP's are the end products of an episode of accretion of matter and angular momentum from the binary companion during the LMXB state. Over the last decade, the discovery of three transitional millisecond pulsars (tMSP's) has allowed a detailed study of the recycling process. Recent studies of PSR J1824−2452I and PSR J1023+0038 have observationally demonstrated the LMXB – MSP evolutionary link. These systems show direct evidence of back-and-forth state switching between radio MSP and accreting X-ray millisecond pulsar regimes and opened a new avenue of research in pulsar astrophysics. The third such tMSP system, J1227-4853, was discovered by us using the GMRT. PSR J1227-4853 is a 1.69 millisecond pulsar at a dispersion measure of 43.4 pc/cm^3. It transited into the active radio-MSP phase associated with a sudden drop of its X-ray and optical luminosity in 2012 December. Extreme orbital perturbations as well as the signature of proper motion are revealed from our GMRT timing campaign. This pulsar, an ``eclipsing redback'', is the only transitioning system currently in an active rotation-powered state. Simultaneous imaging and timing observations with the GMRT were used to directly show that eclipses are caused by absorption rather than dispersion smearing or scattering. A long-term timing study of PSR J1227-4853 is currently under way, which will help to determine whether these transitional systems will eventually be canonical radio MSP's or whether they form a new sub-class of MSP's that continue to transition between the two states. Also, such studies will result in better understanding of the spin evolution of the systems and the dynamics of accretion during the accretion-powered, propeller stage and the rotation-powered stage. The figure shows the pulsar search output for PSR J1227-4853 showing rapid evolution of period and period-derivative in a compact binary system.
Coherently dedispersed gated imaging of millisecond pulsars
The discovery of millisecond pulsars (MSP's) and their precise localisation using existing methods is hindered by their intrinsic fainter nature. This leads to significant delays between the discovery of MSP's and their further identification using conventional imaging methods. Motivated by the need for rapid localization of newly-discovered faint MSP's, we have developed a coherently dedispersed gating correlator for the GMRT. This gating correlator accounts for the orbital motions of MSP's in binary systems, while folding the visibilities with a best-fit topocentric rotational model derived from a periodicity search using simultaneously generated beamformer output. With this technique, the signal-to-noise ratio of the detection of an MSP in the image domain can be dramatically improved (by a factor of as much as 5). We have also incorporated a superior approach of dispersion correction, called coherent dedispersion, in our imaging technique to reconstruct the intrinsic pulse shape of such MSP's. We could unambiguously localize newly discovered Fermi MSP's in the on–off gated image plane with an accuracy of ±1”. Immediate knowledge of such a precise position enables the use of sensitive coherent beams of array telescopes for follow-up timing observations, which substantially reduces the use of telescope time (by a factor of 20 for the GMRT!). In addition, a precise a priori astrometric position reduces the effect of large covariance in timing fit, which in turn accelerates the convergence to an initial timing model. Moreover, such accurate positions allow for rapid identification of pulsar counterparts in optical and X-ray wavelengths. Figure caption: On–off gated images for newly discovered Fermi MSP's. All the MSP's are marked in the respective 10’ × 10’ facet images.
GMRT Discovery of PSR J1544+4937: An Eclipsing Black-widow Pulsar Identified with a Fermi-LAT Source
Bhattacharyya et al. used the GMRT to perform deep observations to search for radio pulsations in the directions of unidentified Fermi Large Area Telescope (LAT) gamma-ray sources, resulting in the discovery of a new milli-second pulsar (MSP), PSR J1544+4947, an eclipsing MSP in a special evolutionary state. PSR J1544+4937 is a 2.16 ms pulsar in a 2.9-hour compact circular orbit with a very low-mass companion star (mass > 0.017 solar masses). At 322 MHz, the pulsar is found to be eclipsing for 13% of its orbit, whereas at 607 MHz the pulsar is detected throughout the low-frequency eclipse phase. Variations in the eclipse ingress phase are observed, indicating a clumpy and variable eclipsing medium. Moreover, additional short-duration absorption events are observed around the eclipse boundaries. The authors used the radio timing solutions to detect gamma-ray pulsation from the pulsar, confirming it as the source powering the gamma-ray emission. The figure shows the frequency-dependent eclipsing detected with the GMRT in PSR J1544+4937. The pulsar radiation is seen to be eclipsed by the companion star at 322 MHz, but not at 607 MHz. The figure plots the variation of the timing residuals and the electron column density around the eclipse phase (which is indicated by the shaded region) at 322 MHz (top) and 607 MHz (bottom).
The GMRT High Resolution Southern Sky (GHRSS) survey
Bhattacharyya et al. used the outstanding GMRT potential for low-frequency pulsar surveys in the GMRT High Resolution Southern Sky (GHRSS) survey, a low-frequency survey for pulsars and transients away from the Milky Way's plane. The GHRSS survey covers Galactic latitudes |b|>5 degrees, scanning the southern sky, with declination -40 degrees to -54 degrees. This declination coverage is complementary to the coverage of other ongoing low-frequency sky surveys around the world. The first phase of the GHRSS survey was carried out using the narrow bandwidths of the GMRT Software Backend, at 322 MHz, and has already resulted in the discovery of bunch of new pulsars with exciting properties. Bhattacharyya et al. discovered 13 pulsars in the GHRSS survey in a surveyed area of 1800 square degrees, i.e. 0.007 pulsars per square degree, which is one of the highest among pulsar surveys away from the Milky Way’s plane. GHRSS survey discoveries include a millisecond pulsar (in a ~10 hour orbit around a ~0.18 solar mass companion star), a pulsar for which gamma-ray pulsations have been discovered using the Fermi Large Area Telescope, and two mildly recycled pulsars. The second phase, using the GMRT Wideband Backend and the 250-500 MHz receivers of the upgraded GMRT is now under way. The figure shows the 21 pulsars discovered by the GMRT between 2012−2017 from targeted and blind surveys. Fermi-directed discoveries are shown as green points; the blue shaded region indicates the sky coverage in Galactic coordinates of the GHRSS survey, while the pulsars discovered in this survey are shown as red points Details: http://www.ncra.tifr.res.in/ncra/research/research-at-ncra-tifr/research-areas/pulsarSurveys/GHRSS