AstroSat-CZTI as a hard X-ray Pulsar Monitor
K.G. Anusree, D. Bhattacharya, A.R. Rao, S. Vadawale, V. Bhalerao, A. Vibhute
aa r X i v : . [ a s t r o - ph . H E ] F e b J. Astrophys. Astr. (0000) :
AstroSat-CZTI as a hard X-ray Pulsar Monitor
Anusree, K.G. , Bhattacharya, D. , Rao, A.R. , Vadawale, S. , Bhalerao, V. & Vibhute, A. . School of Pure and Applied Physics, Mahatma Gandhi University, Kottayam 686560, India. Inter-University Centre for Astronomy and Astrophysics, Pune 411007, India. Physical Research Laboratory, Ahmedabad, Gujarat 380009, India. Indian Institute of Technology, Bombay 400076, India. Tata Institute of Fundamental Research, Mumbai 400005, India * Corresponding author. E-mail: [email protected] received 30 October 2020; accepted 18 January 2021
Abstract.
The Cadmium Zinc Telluride Imager (CZTI) is an imaging instrument onboard AstroSat. This instru-ment operates as a nearly open all-sky detector above 60 keV, making possible long integrations irrespective of thespacecraft pointing. We present a technique based on the AstroSat-CZTI data to explore the hard X-ray character-istics of the γ -ray pulsar population. We report highly significant ( ∼ σ ) detection of hard X-ray (60–380 keV)pulse profile of the Crab pulsar using ∼ ff -axis sensitivityof the instrument and establish AstroSat-CZTI as a prospective tool in investigating hard X-ray characteristics of γ -ray pulsars as faint as 10 mCrab. Keywords.
Pulsars: individual (Crab, PSR J0534 +
1. Introduction
Pulse profiles of rotation powered pulsars are shaped bythe geometry of the emission region as well as the radi-ation processes at work in pulsar magnetosphere (Wat-ters and Romani, 2011, Pierbattista et al., 2014). Theemission is usually broadband, covering radio through γ -rays. However, emission at di ff erent wavebands typ-ically arises in di ff erent regions of the magnetosphere,giving rise to pulse profiles that are wavelength depen-dent. There exist a variety of magnetospheric models(e.g. Cheng, Ho & Ruderman 1986, Muslimov & Hard-ing 2004, P´etri 2011, Cerutti & Beloborodov 2017)which di ff er in their prediction of the shape and distri-bution of acceleration and radiation zones. BroadbandSpectral Energy Distribution (SED) and the energy de-pendence of the pulse shape and arrival time provideclues to the distribution of these zones and have beenused to constrain the magnetospheric geometry and todiscriminate between theoretical models (e.g. Romani& Yadigaroglu 1995, Cheng et al., 2000, Abdo et al.,2009c, Abdo et al. 2010c, Bai & Spitkovsky 2010,YuanJie et al., 2012, Pierbattista et al., 2012, 2014).The number of known γ -ray pulsars stood at justseven until the launch of NASA’s Fermi Large AreaTelescope (LAT) in 2008. Since then, the continu- ous accumulation of data, together with highly e ffi -cient searching algorithm, has resulted in the numberof pulsars detected by LAT in the γ -ray band to mountto 253 (Ray et al., 2020), of which 71 have no radiocounterpart. The modelling of γ -ray and radio emis-sion together can provide important constraints on theglobal magnetospheric properties (e.g. see P´etri & Mi-tra 2020). For radio-quiet γ -ray pulsars, the magne-tospheric X-ray emission provides the only additionalclue to the emission process. The radiation energy out-put of a rotation powered pulsar typically peaks in theX-ray / γ -ray region. However, only 18 pulsars havebeen detected in the X-ray band till date (Caraveo 2014,Kuiper and Hermsen 2015) and the SED of many ofthem are sparsely sampled. Using ephemerides deter-mined from LAT data, four of the LAT radio-quiet pul-sars have been observed at photon energies <
20 keV(Lin et al., 2010; Caraveo et al., 2010; Marelli et al.,2014; Lin et al., 2013 and Lin et al., 2014). These fallin the category of “Geminga-like” pulsars, for whichthe profile and spectra are known at soft X-rays (i.e., < γ -rays alone (PSR B0633 +
17, Halpernand Holt 1992; Bertsch et al., 1992). Unfortunately,at energies above 20 keV, the major X-ray observa-tories have had relatively low sensitivity. Additionaldeep observations in the hard X-ray band would there- © Indian Academy of Sciences 1
J. Astrophys. Astr. (0000) : fore augment the existing information, providing a bet-ter estimate of spectral shape and bolometric luminos-ity of these and other pulsars hitherto undetected inX-rays. Three major emission mechanisms operate inpulsar magnetospheres, namely synchrotron radiation,curvature radiation and inverse-Compton scattering. Awell-measured SED can distinguish the relative contri-butions of these components, leading to a model of theparticle energy distribution in the emission zones.To increase the sample of pulsar detections in thehard X-ray / Soft γ -ray bands, and to investigate howtheir properties fit in the general picture emerging fromthe theoretical studies of the Fermi’s young gamma-ray pulsars, we need “open all-sky X-ray detectors”.The pulsar spectra are steep at high energies. In gen-eral, photon flux of the young / middle-aged LAT γ -ray pulsars can be represented by a power-law witha simple exponential cuto ff , i.e. F γ = k · ( E γ / E ) Γ · exp( − ( E γ / E c ) β ) where β ≈ E γ is the photon en-ergy, E a normalisation energy, E c the cuto ff energyand k is the normalization. The photon index Γ hasbeen found to lie in the range − . − . γ -ray pulsar. Such long integrationsare not a ff ordable by any of the missions at present.Open all-sky detectors, on the other hand, can collectphotons during other observations, making it possibleto search for these pulsars.The first Indian multi-wavelength Satellite AstroSatwas launched in 2015 with five instruments onboard(Singh et al., 2014). One of them, the Cadmium ZincTelluride Imager (CZTI; Bhalerao et al., 2017), can de-tect photons in 20–380 keV energy range. Its hous-ing and collimators are made of Aluminium alloy andthin Tantalum shields that define its low-energy fieldof view but allow su ffi cient uncollimated penetrationabove ∼
60 keV to make CZTI an excellent wide-angle monitor at higher energies, covering roughly one-third of the sky at all times. This monitoring capabilityhas been leveraged to detect many transients, includ-ing over 300 Gamma-Ray Bursts (Sharma et al., thisvolume). A more detailed description of CZTI can befound in Bhalerao et al., 2017. The aim of this paperis to assess the suitability of CZTI for the detection andstudy of pulsars in the hard X-ray band, using its o ff -axis detection capability.During CZTI pointing observations, photons fromcandidate hard X-ray sources that shine in through thewalls will come to piggyback on any ongoing observa-tion, with varying sensitivity depending on the point-ing. Arrival times of the photons from a pulsar carrythe signature of its spin period, enabling us to search the data for hard X-ray pulsars with known ephemeris.This presents the detection of Crab Pulsar in the energyrange 60-380 keV from o ff -axis CZTI observations, us-ing a custom algorithm developed by us. The Crabpulsar (PSRJ0534 + ff -axis sensitivity of the instrument.This paper is structured as follows. Section 2 de-scribes the instruments used in this work and the analy-sis of data, followed by a discussion of results in section3 which includes the comparison of hard X-ray pulseprofiles of the Crab pulsar obtained from CZTI with γ -ray profiles from LAT data. Finally, we assess thepotential of AstroSat-CZTI in the investigation of hardX-ray counterparts of γ -ray pulsars.
2. Data and analysis
This work is based on data from AstroSat-CZTI point-ing observations that were released for public use onor before 30th April 2019. We have also used pub-licly available archival data from NASA’s Fermi-LATmission. All material informations about the instru-ments and data, along with the characteristics of anal-ysis methods, are described in this section. Moreover,we also discuss some of our checks against the vulner-abilities anticipated during long integration.2.1
AstroSat-CZTI
CZTI is the AstroSat instrument primarily designedfor simultaneous hard X-ray imaging and spectroscopyof celestial X-ray sources in the energy band 20-150keV. Its functioning employs the technique of codedmask imaging. The CZTI instrument bears a two-dimensional coded aperture mask (CAM) above itspixellated, 5-mm thick solid-state CZT detector mod-ules spread over four quadrants. Passive collimators areplaced in each quadrant of CZTI that support the codedmask. The set up defines a 4.6 deg × ∼ ∼ ff ective up to ∼
100 keV above which they become transparent. Thetransparency being a function of energy and angle of in-cidence, appreciable sensitivity for o ff -axis sources ex-tends down to ∼
60 keV (see Bhattacharya et al., 2018).The 976 cm of the detector’s total geometric area isdistributed over 16384 pixels, with 4096 pixels in eachindependent quadrant. After the launch, about 15 per-cent of the pixels were disabled for having shown ex-cessive electronic noise. Nearly 25 percent of the re-maining pixels seemed to have an inadequate spectro-scopic response. Considering that the coded mask has a . Astrophys. Astr. (0000) : Table 1
Telescope Energy range Start MJD Stop MJD Exposure(ks)AstroSat-CZTI( Crab at nominal pointing) 30-60 keV 57290 58232 519AstroSat-CZTI(Crab within 5-70 degree of nominal pointing) 60-380 keV 57290 58232 4752Fermi-LAT( within 1 degree of γ -ray position of Crab pulsar) 0.1-300 GeV 57290 58500 5962 Brief summary of the observations. Participating telescopes and their instruments, the energy range chosen for the work, range of MJDsfor which the data have been collected for this work and the total exposure achieved are listed ∼
50% open fraction, the total e ff ective area at normalincidence is ∼ cm in all active pixels at energiesbelow 100 keV. The detected events are recorded witha time resolution of 20 µ s. The absolute timestamp as-signed to CZTI events is estimated to have a jitter of ∼ µ s RMS (Bhattacharya, 2017) and a fixed o ff setwith respect to Fermi of 650 ± µ s (Basu et al., 2018).2.1.1 Analysis of CZTI data:
All available CZTIpointing observations during MJD 57366-58362, withthe Crab pulsar within 5-80 degrees from normal inci-dence, were selected for this work, resulting in a totalexposure of 5096ks. The merged Level-1 data of allthe selected CZTI observations were reduced to Level-2 using standard CZTI analysis software. Details andthe sequence of analysis modules can be found in thelatest version of the CZTI user-guide available at theASSC website . There are intervals during pointing ob-servation where data is absent due to SAA passage anddata transmission loss. There are also intervals whenthe earth occults the target source. To generate scienceproducts from such observations, identifying such in-tervals and removing the data for that duration by ade-quately accounting for the gaps is essential. This taskis performed by the module cztgtigen . It generatesGood Time Interval (GTI) files based on various thresh-old parameters. We generated custom GTI files to takeinto consideration the earth occultation of the o ff -axissource and filtered the original event file accordingly.These filtered, clean event files were used in the subse-quent steps of analysis.In this work, we have used AstroSat CZTI observa-tions with on-axis pointing of the Crab pulsar as well asothers when the pulsar is o ff -axis at angles between 5to 70 deg (see below for the choice of this angle range).We have also used Fermi-LAT observations of the Crabpulsar spanning a similar time range for comparison.A brief description of the data sets used is provided inTable 1.The arrival times of the CZTI events at thespacecraft were converted to those at the Solar- http: // astrosat-ssc.iucaa.in / system Barycenter using the JPL DE200 Solar-systemephemeris. This was done using the tool as1bary , aversion of NASA / HEASOFT AXBARY package, cus-tomised for AstroSat, using the well-known astrometricposition of the Crab pulsar.We developed a custom code to fold the barycen-tered event data spanning many months but includingfrequent gaps. This code assigns an absolute phase toeach recorded event using the polynomial model tim-ing solutions known as SSB polycos generated usingtempo2 (Hobbs et al.,2006). Polycos predict a pulsar’sparameters at a particular epoch. A polyco file containspulsar ephemerides over a short period, typically hours,in simple polynomial expansion. All the polycos weregenerated at an epoch in the centre of each 6-hour inter-val in this work. These 6-hour ephemerides were thenused for folding the LAT data as well as the CZTI data.As an initial test of the custom code, Fermi-LAT γ -rayevents for a few pulsars were folded and compared withtheir published timing models. The SSB polycos weregenerated from LAT timing models available publiclyat Fermi’s website. . Finally, the code was testedon AstroSat-CZTI data. For this, publicly availableAstroSat-CZTI data from six pointing observations ofCrab pulsar (Table 1, first row) were selected and re-duced using the default CZTI data analysis pipeline.The 30-60 keV pulse profile, thus obtained, is shown inFigure 2(a). The Crab pulsar ephemeris derived fromFermi observations and used in this work are given inTable 2 along with the reference epochs.To compare the pulse profiles between hard X-raysand γ -rays, we folded all the events extracted fromAstroSat-CZTI observations and those obtained fromFermi observation within MJD 57290–58500 using the γ ray ephemeris mentioned above. For all CZTI dataanalysis, we have directly combined the data of all fourquadrants, as they run on a synchronised time refer-ence (see Figure 1). We initially folded the o ff -axisdata separately for every 10-degree angle interval andfound that when the source is located beyond 70 de-grees from the pointing axis, the signal-to-noise ratio in https: // confluence.slac.stanford.edu / display / GLAMCOG / LAT + Gamma-ray + Pulsar + Timing + Models
J. Astrophys. Astr. (0000) : . × × c oun t s / b i n pulse phase Q0 Q1 Q2 Q3 Figure 1 : Phase histograms of PSR J0534 + ff -angle data(this work) from the four di ff erent quadrants (Q0, Q1,Q2 and Q3) of CZTI operating as independent detec-tors. All the quadrants are found to be time alignedwith no measurable relative delaya given integration time drops significantly below thosefor smaller o ff -axis angles. We, therefore, decided torestrict the accumulation of o ff -axis data to the anglerange of 5 to 70 deg (below 5 deg the source wouldappear in the main FoV).With the phase reference as presented in Table 3,we consider the “o ff -pulse” region of the profile, phase0.5 through 0.8, as the background. The signal to noiseratio is calculated as the ratio of the peak count to thestandard deviation of the counts in the o ff -pulse region.We also divided the obtained X-ray events into severaldi ff erent energy bands to check consistency, as shownin Figure 2(b).2.2 Fermi-LAT
The LAT instrument is described by Atwood et al.,(2008). We have used the already publicly availabledata from LAT. A unique value of the LAT data is thata pulsar’s discovery in γ -rays often enables the immedi-ate measurement of the pulsar parameters over the ten-year span in which the LAT has been operating. LATdata have been used to find precise timing solutions formany pulsars, including radio-quiet and radio-faint pul-sars (Ray et al., 2011; Kerr et al., 2015; Clark et al.,2017). Table 2 : Fermi-LAT γ -ray timing solution of PSRJ0534 + σ errors in the parameters . ParametersRight ascension, α δ + f ( s − f (10 − s − ) -3.71184342371(4.6591493867e-17)Second derivative , ¨ f (10 − s − ) 3.3226958153(1.1931513864e-24)Third derivative , ... f (10 − s − ) 1.2860657170(9.2842295751e-32)PEPOCH 55555POSEPOCH 50739DMEPOCH 55107.807158553TZRMJD 56730.15526586924Solar system ephemeris model DE405Time system TDB Analysis of Fermi-LAT data
In order to get asignificant γ -ray profile to compare with the CZTIresults, we took all available photon data for theLAT source PSRJ0534 + ff ective exposure time of 5962-kilo sec-onds, as the observatory scans the entire sky once ev-ery three hours. We used the HEADAS-FTOOLS onHEAsoft-ver 6.27. (Blackburn, 1995) to perform thedata reduction. We obtained the Fermi-LAT data in theenergy range of 0.1-300 GeV within a circular region ofinterest (ROI) with a 1-degree radius from the decided γ -ray position of PSR J0534 + γ -rays. We used gtbary tool in fermiscience tools to apply barycentric corrections to pho-ton arrival times in LAT event files using correspondingFermi orbit files. After barycentering the events usingfermi science tool gtbary , the absolute phase of eachevent in 0.1-300 GeV was determined by the customcode developed by us, with 6 hour SSB polycos gener-ated using the timing parameters in Table 2 http: // heasarc.gsfc.nasa.gov / ftools https: // fermi.gsfc.nasa.gov / ssc / data / analysistools / overview.html . Astrophys. Astr. (0000) :
3. Discussion of Results
Figure 2 shows the folded pulse profile of the Crabpulsar (Period: ∼
33 ms) using the LAT and the CZTIinstruments. The CZTI profiles (purple in colour) ac-cumulate data from multiple observations spread overseveral months, with Crab position within 5 to 70 de-grees away from the nominal pointing direction. Thetotal integration time in these CZTI profiles is about5000 ks, giving a signal to noise ratio of ∼
30 in theenergy integrated profile and over ∼
15 in the energy-resolved ones. The long-known energy dependence ofthe Crab pulse profile can be seen when compared withthe softer bands (e.g. 30-60 keV in CZTI) in Figure2(a). The left peak is taller at lower energies while theright peak dominates at high energies. The bridge emis-sion connecting the two peaks is also seen relativelystronger at higher X-ray energies. The normalised lightcurves in three separate hard X-ray bands are shownin Figure 2(b). The right peak(P2) grows, but the leftpeak (P1) again starts to dominate at very high energies,beyond ( ∼
10 MeV) in the γ -regime, as seen in Figure2(a).For a more quantitative comment on the morphol-ogy change observed, we determined the intensity ra-tios P2 / P1 and Bridge / P1 in three separate hard X-ray bands shown in Figure 2(b), adopting the phaseinterval definitions of (Abdo et al., 2010) shown inTable 3. The values obtained for P2 / P1 ratios are1.301 ± ± ± / P1 are 0.536 ± ± ± γ -ray / radio timing models.Integrating the well established broken power-lawmodel spectrum for Crab (Ulmer et al., 1995), we cal-culated the hard X-ray flux in 60-380 keV band to be0.011 ph cm − s − . The observed CZTI detection countrates of 0 . ± .
01 cps in 60-380 keV band translatesto an average CZTI o ff -axis e ff ective area of ∼
80 cm ,averaged over 5-70 degree o ff -axis. This is about 20percent of the on-axis e ff ective area at lower energies.Assuming a Crab-like spectrum and based on theobserved count rates mentioned above, we estimate thepossible integration time required for a 5 σ detection ofa 10 mCrab o ff -axis pulsed source with CZTI to be ∼ γ -ray pulsars from thesecond Fermi / LAT pulsar catalog in the CZTI hard X- ray (60-380 keV) band, as shown in Figure 3.
4. Conclusion
We have presented a successful attempt to detect theCrab pulsar by the AstroSat CZT Imager from point-ings where the pulsar was o ff -axis, at angles rangingfrom 5 to 70 degrees. This is possible because the Col-limator and the housing of the CZTI become graduallytransparent at energies above 60 keV, turning it into anall-sky detector at higher energies. Accumulating ano ff -axis exposure of ∼ ∼ σ pulseprofile of the Crab pulsar in the 60–380 keV energyband. Energy-resolved pulse profiles constructed atmultiple sub-bands reproduce the known energy depen-dence of the profile shape. This demonstrates the capa-bility of AstroSat-CZTI to act as a hard X-ray pulsarmonitor, in 60-380 keV band, with an average o ff -axise ff ective area of ∼
80 cm . Our results establish that theCZTI time stamps possess su ffi cient long-term stabil-ity to carry out phase connected timing spanning manyyears. We estimate that with continued accumulation ofdata, it will become possible for the CZTI to detect pul-sars with hard X-ray fluxes down to ∼
10 mCrab, thusmaking a large majority of Fermi / LAT pulsars accessi-ble for study in the 60–380 keV energy band. Such asurvey is currently ongoing with several successful de-tections already made. These results will be reportedelsewhere.
Acknowledgements
This publication makes use of data from the CZTI on-board Indian astronomy mission AstroSat, archived atthe Indian Space Science Data Centre (ISSDC). TheCZT Imager instrument was built by a TIFR-led con-sortium of institutes across India, including VSSC,ISAC, IUCAA, SAC, and PRL. The Indian Space Re-search Organisation funded, managed and facilitatedthe project. We extend our gratitude to CZTI POCteam members at IUCAA for helping with the augmen-tation of data. We thank Fermi Timing Observers PaulRay and Kerr Mattew for their timely and favourableresponse in providing LAT ephemeris for Crab pulsarand helping with queries related to SSB polyco gener-ation using tempo2. We thank the anonymous refereefor his / her valuable suggestions to improve the paper.We would like to thank Avishek Basu, Karthik Rajeevand Atul Mohan for useful discussions. We thank IU-CAA HPC facility where we carried out all the analy-sis. Anusree K. G. acknowledges support for this workfrom DST-INSPIRE Fellowship grant, IF170239, underMinistry of Science and Technology, India. J. Astrophys. Astr. (0000) :
Table 3 : Phase component definitions for the Crab pulsar (Abdo et al., 2010) adopted in this study
Component Abbreviation Phase Interbval WidthPeak 1 P1 0.87 - 1.07 0.20Peak 2 P2 0.27 - 0.47 0.20Bridge Bridge 0.098 - 0.26 0.162O ff Pulse OP 0.52 - 0.87 0.35 . Fermi−LAT0.1−300 GeVCZTI (this work)60−380 keVCZTI30−60 keV . r m a li s ed i n t en s i t y . pulse phase (a) . . r m a li s ed i n t en s i t y . pulse phase (b) Figure 2 : The pulse profile of the Crab pulsar observed with Fermi-LAT (black) and the AstroSat-CZTI (purple).The integration time in the 0.1–300 GeV LAT profile is ∼ ∼ ∼
600 ks of on-axispointing observations of the Crab pulsar with AstroSat-CZTI. In CZTI profiles, data from multiple observationsduring MJD 57290-58232 with PSR J0534 + . Astrophys. Astr. (0000) : F e r m i - L A T F l u x i n c r a b un i t s hard X-ray pulsar observability with CZTI LAT 2PCLAT discovered pulsarsObserved in X-raysCZTI
Figure 3 : Detectability of Fermi-LAT pulsars with theCZTI. The green horizontal line marks the estimatedflux of the faintest detectable hard X-ray pulsar us-ing five years of archival AstroSat-CZTI data. Thered diamonds represent the hard X-ray pulsar candi-dates discovered / observed in the soft X-ray band ( < γ -ray pulsars Vela (PSRJ0835-4510) and Geminga(PSRJ0633 + References
Abdo, A. A., Ackermann, M., Atwood, W. B., Bagagli, R. etal., 2009, The Astrophysical Journal, 696, 1084.Abdo, A. A., et al., 2010, The Astrophysical Journal Supple-ment Series, 208, 17.Abdo, A. A., Ackermann, M., Ajello, M. et al., 2010, TheAstrophysical Journal, 708, 1254.Atwood, W. B., Abdo, A. A., Ackermann, M. et al., 2009,The Astrophysical Journal, 697, 1071.Bai, X. N. & Spitkovsky, A., 2010, The Astrophysical Jour-nal, 715, 1282Basu, A., Joshi, B.C., Bhattacharya, D. et al., 2018, Astron-omy and Astrophysics, 617, A22. Bertsch, D. L., Brazier, K. T. S., Fichtel, C. E. et al., 1992,Nature, 357, 306.Bhalerao, V., Bhattacharya, D., Vibhute, A. et al., 2017, Jour-nal of Astrophysics and Astronomy, 38, 31.Bhattacharya, D. (2017). Journal of Astrophysics and As-tronomy, 38, 51.Bhattacharya, D., Dewangan, G.C., An-tia H.M. et al., 2018, AstroSat Handbook,http: // / ∼ astrosat / AstroSat handbook.pdfBlackburn, J. K. (1995). Astronomical Data Analysis Soft-ware and Systems IV, 77, 367.Caraveo, P. A. (2014). Gamma-ray pulsar revolution. AnnualReview of Astronomy and Astrophysics, 52.Caraveo, P. A., De Luca, A., Marelli, M. et al., 2010, TheAstrophysical Journal Letters, 725, L6.Cerutti, B., & Beloborodov, A. M., 2017, Space Science Re-views, 207, 111.Cheng, K.S., Ho, C. & Ruderman, M., 1986, The Astrophys-ical Journal, 300, 522.Cheng, K.S., Ruderman, M. & Zhang, L., 2000, The Astro-physical Journal, 537, 964.Clark, C. J., Wu, J., Pletsch, H. J. et al., 2017, The Astro-physical Journal, 834, 106.Du, Y., Qiao, G., & Wang, W., 2012, The Astrophysical Jour-nal, 748, 84.Halpern, J. P., & Holt, S. S., 1992, Nature, 357, 222.Harding, A., & Muslimov, A., 2004, cosp, 35, 562.Hewish, A., Bell, S. J., Pilkington, J. D. et al., 1968. Nature,217, 709.Hobbs, G. B., Edwards, R. T., & Manchester, R. N., 2006,Monthly Notices of the Royal Astronomical Society, 369,655.Kerr, M., Ray, P. S., Johnston, S., 2015, The AstrophysicalJournal, 814, 128.Kuiper, L., Hermsen, W., Cusumano, G. et al., 2001, Astron-omy & Astrophysics, 378, 918.Kuiper, L., & Hermsen, W., 2015, Monthly Notices of theRoyal Astronomical Society, 449, 3827.Lin, L. C., Huang, R. H., Takata, J. et al., 2010, The Astro-physical Journal Letters, 725, L1.Lin, L. C. C., Hui, C. Y., Hu, C. P. et al., 2013, The Astro-physical Journal Letters, 770, L9.Lin, L. C. C., Hui, C. Y., Li, K. T. et al., 2014, The Astro-physical Journal Letters, 793, L8.Marelli, M., Harding, A., Pizzocaro, D. et al., 2014, The As-trophysical Journal, 795, 168.Muslimov, A.G. & Harding, A.K., 2004, The AstrophysicalJournal, 606, 1143
J. Astrophys. Astr. (0000) : γ -ray light-curve morphology of youngpulsars: comparing Fermi observations and simulated pop-ulations. Saint-Petersburg, Russia July 28–August 1, 2014,99.Ray, P. S., Kerr, M., Parent, D. et al., 2011, The Astrophysi-cal Journal Supplement Series, 194, 17. Ray, P.S., Smith D.A. et al., 2020,https: // confluence.slac.stanford.edu / display / GLAMCOG / Public + List + of + LAT-Detected + Gamma-Ray + Pulsars //