1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[apj]{emulateapj}
3: \usepackage{rotate,amssymb}
4: \def\lsim{\ifmmode\stackrel{<}{_{\sim}}\else$\stackrel{<}{_{\sim}}$\fi}
5: \def\gsim{\ifmmode\stackrel{>}{_{\sim}}\else$\stackrel{<}{_{\sim}}$\fi}
6:
7: \begin{document}
8:
9: \journalinfo{The Astrophysical Journal, in press}
10: \submitted{The Astrophysical Journal, in press}
11:
12: \title{High-Resolution Timing Observations of Spin-Powered Pulsars
13: with the \emph{AGILE} Gamma-Ray Telescope\altaffilmark{1}}
14:
15: \shorttitle{\emph{AGILE} Observations of Spin-Powered Pulsars}
16: \shortauthors{A.~Pellizzoni et al.}
17:
18: %\slugcomment{FINAL Draft 11 - 2/7/2008}
19: %
20: \author{
21: %*** AUTHORS ORDER TBC*** \\
22: A.~Pellizzoni,\altaffilmark{2}
23: M.~Pilia,\altaffilmark{2,3}
24: A.~Possenti,\altaffilmark{3}
25: F.~Fornari,\altaffilmark{2}
26: %..
27: P.~Caraveo,\altaffilmark{2}
28: E.~Del Monte,\altaffilmark{4}
29: S.~Mereghetti,\altaffilmark{2}
30: M.~Tavani,\altaffilmark{4,5}
31: A.~Argan,\altaffilmark{4}
32: A.~Trois,\altaffilmark{4}
33: %...
34: M.~Burgay,\altaffilmark{3}
35: A.~Chen,\altaffilmark{2,6}
36: I.~Cognard,\altaffilmark{7}
37: E.~Costa,\altaffilmark{4}
38: N.~D'Amico,\altaffilmark{3,8}
39: P.~Esposito,\altaffilmark{2,9,10}
40: Y.~Evangelista,\altaffilmark{4}
41: M.~Feroci,\altaffilmark{4}
42: F.~Fuschino,\altaffilmark{11}
43: A.~Giuliani,\altaffilmark{2}
44: J.~Halpern,\altaffilmark{12}
45: G.~Hobbs,\altaffilmark{13}
46: A.~Hotan,\altaffilmark{14}
47: S.~Johnston,\altaffilmark{13}
48: M.~Kramer,\altaffilmark{15}
49: F.~Longo,\altaffilmark{16}
50: R.~N.~Manchester,\altaffilmark{13}
51: M.~Marisaldi,\altaffilmark{11}
52: J.~Palfreyman,\altaffilmark{17}
53: P.~Weltevrede,\altaffilmark{13}
54: %...
55: G.~Barbiellini,\altaffilmark{16}
56: F.~Boffelli,\altaffilmark{9,10}
57: A.~Bulgarelli,\altaffilmark{11}
58: P. W. Cattaneo,\altaffilmark{9}
59: V.~Cocco,\altaffilmark{4}
60: F.~D'Ammando,\altaffilmark{4,5}
61: G.~De Paris,\altaffilmark{4}
62: G.~Di Cocco,\altaffilmark{4}
63: I.~Donnarumma,\altaffilmark{4}
64: M.~Fiorini,\altaffilmark{2}
65: T.~Froysland,\altaffilmark{5,6}
66: M.~Galli,\altaffilmark{18}
67: F.~Gianotti,\altaffilmark{11}
68: A.~Harding,\altaffilmark{19}
69: C.~Labanti,\altaffilmark{11}
70: I.~Lapshov,\altaffilmark{4}
71: F.~Lazzarotto,\altaffilmark{4}
72: P.~Lipari,\altaffilmark{20}
73: F.~Mauri,\altaffilmark{9}
74: A.~Morselli,\altaffilmark{21}
75: L.~Pacciani,\altaffilmark{4}
76: F.~Perotti,\altaffilmark{2}
77: P.~Picozza,\altaffilmark{21}
78: M.~Prest,\altaffilmark{22}
79: G.~Pucella,\altaffilmark{4}
80: M.~Rapisarda,\altaffilmark{23}
81: A. Rappoldi,\altaffilmark{9}
82: P.~Soffitta,\altaffilmark{4}
83: M.~Trifoglio,\altaffilmark{11}
84: E.~Vallazza,\altaffilmark{16}
85: S.~Vercellone,\altaffilmark{2}
86: V.~Vittorini,\altaffilmark{5}
87: A.~Zambra,\altaffilmark{2}
88: D.~Zanello,\altaffilmark{20}
89: %...
90: C.~Pittori,\altaffilmark{24}
91: F.~Verrecchia,\altaffilmark{24}
92: B.~Preger,\altaffilmark{24}
93: P.~Santolamazza,\altaffilmark{24}
94: %...
95: P.~Giommi,\altaffilmark{24}
96: and
97: L.~Salotti\altaffilmark{25}
98: }
99:
100: \altaffiltext{1}{Based on
101: observations obtained with \emph{AGILE}, an ASI (Italian Space
102: Agency) science mission with instruments and contributions directly
103: funded by ASI.}
104:
105: \altaffiltext{2}{INAF/IASF--Milano, via E.~Bassini 15, I-20133 Milano,
106: Italy}
107: \altaffiltext{3}{INAF--Osservatorio Astronomico di Cagliari,
108: localit\'a Poggio dei Pini, strada 54, I-09012 Capoterra, Italy}
109: \altaffiltext{4}{INAF/IASF--Roma,
110: via del Fosso del Cavaliere 100, I-00133 Roma, Italy}
111: \altaffiltext{5}{Dipartimento di Fisica, Universit\`a ``Tor Vergata'', via della Ricerca
112: Scientifica 1, I-00133 Roma, Italy}
113: \altaffiltext{6}{CIFS--Torino, viale Settimio Severo 3,
114: I-10133, Torino, Italy}
115: \altaffiltext{7}{Laboratoire de Physique et Chimie de l'Environnement,
116: CNRS, F-45071 Orleans, France}
117: \altaffiltext{8}{Dipartimento di Fisica, Universit\`a di Cagliari,
118: Cittadella Universitaria, I-09042 Monserrato, Italy}
119: \altaffiltext{9}{INFN--Pavia, via Bassi 6, I-27100 Pavia, Italy}
120: \altaffiltext{10}{Dipartimento di Fisica Nucleare e Teorica,
121: Universit\`a di Pavia, via A.~Bassi 6, Pavia, I-27100, Italy}
122: \altaffiltext{11}{INAF/IASF--Bologna, via
123: Gobetti 101, I-40129 Bologna, Italy}
124: \altaffiltext{12}{Columbia Astrophysics Laboratory, Columbia University,
125: New York, NY 10027 }
126: \altaffiltext{13}{Australia Telescope National Facility, CSIRO,
127: P.O.~Box~76, Epping NSW~1710, Australia}
128: \altaffiltext{14}{Curtin University of Technology, 78 Murray Street,
129: Perth, WA 6000, Australia}
130: \altaffiltext{15}{University of Manchester, Jodrell Bank Observatory,
131: Macclesfield, Cheshire SK11 9DL, UK}
132: \altaffiltext{16}{Dipartimento di Fisica, Universit\`a di Trieste and INFN--Trieste, via Valerio 2, I-34127 Trieste,
133: Italy}
134: \altaffiltext{17}{School of Mathematics and Physics, University of
135: Tasmania, Hobart, TAS 7001, Australia}
136: \altaffiltext{18}{ENEA--Bologna, via Biancafarina 2521,
137: I-40059 Medicina (BO), Italy}
138: \altaffiltext{19}{Astrophysics Science Division, NASA/Goddard Space
139: Flight Center, Greenbelt, MD 20771}
140: \altaffiltext{20}{INFN--Roma ``La Sapienza'', piazzale A. Moro 2,
141: I-00185 Roma, Italy}
142: \altaffiltext{21}{INFN--Roma ``Tor Vergata'', via della Ricerca
143: Scientifica 1, I-00133 Roma, Italy}
144: \altaffiltext{22}{Dipartimento di Fisica, Universit\`a dell'Insubria, via Valleggio 11,
145: I-22100 Como, Italy}
146: \altaffiltext{23}{ENEA--Roma, via E. Fermi 45, I-00044 Frascati (Roma),
147: Italy}
148: \altaffiltext{24}{ASI--ASDC, via G. Galilei, I-00044 Frascati (Roma),
149: Italy}
150: \altaffiltext{25}{ASI, viale Liegi 26, I-00198 Roma, Italy}
151:
152:
153: \email{alberto@iasf-milano.inaf.it}
154:
155: \begin{abstract}
156:
157: \emph{AGILE} is a small gamma-ray astronomy satellite mission of the
158: Italian Space Agency dedicated to high-energy astrophysics launched in
159: 2007 April. Its $\sim$1 $\mu$s absolute time tagging capability
160: coupled with a good sensitivity in the 30 MeV--30 GeV range, with
161: simultaneous X-ray monitoring in the 18--60 keV band, makes it
162: perfectly suited for the study of gamma-ray pulsars following up on
163: the \emph{CGRO}/EGRET heritage. In this paper we present the first
164: \emph{AGILE} timing results on the known gamma-ray pulsars Vela, Crab,
165: Geminga and B\,1706--44. The data were collected from 2007 July to 2008
166: April, exploiting the mission Science Verification Phase, the
167: Instrument Timing Calibration and the early Observing Pointing
168: Program. Thanks to its large field of view, \emph{AGILE} collected a
169: large number of gamma-ray photons from these pulsars ($\sim$10,000
170: pulsed counts for Vela) in only few months of observations. The
171: coupling of \emph{AGILE} timing capabilities, simultaneous radio/X-ray
172: monitoring and new tools aimed at precise photon phasing, exploiting
173: also timing noise correction, unveiled new interesting features at
174: sub-millisecond level in the
175: pulsars' high-energy light-curves.
176: \end{abstract}
177: \keywords{stars: neutron -- pulsars: general --
178: pulsars: individual (Vela, Crab, Geminga, PSR~B\,1706--44)
179: -- gamma rays: observations }
180:
181:
182: \section{Introduction}
183:
184: Among the $\sim$1,800 known rotation-powered pulsars, observed mainly
185: in the radio band, seven objects have been identified as gamma-ray
186: emitters, namely Vela (B\,0833--45), Crab (B\,0531+21), Geminga
187: (J\,0633+1746), B\,1706--44, B\,1509--58, B\,1055--52 and B\,1951+32
188: \citep[][]{thompson04}. In addition, B\,1046--58 \citep{klm00},
189: B\,0656+14 \citep{ramanamurthy96} and J\,0218+4232
190: \citep{kuiper00,kuiper02} were reported with lower confidence
191: (probability of periodic signal occurring by chance in gamma-rays of
192: $\sim$10$^{-4}$). In spite of the paucity of pulsar
193: identifications, gamma-ray observations are a valuable tool for
194: studying particle acceleration sites and emission mechanisms in the
195: magnetospheres of spin-powered pulsars.
196:
197: So far, spin-powered pulsars were the only class of Galactic sources
198: firmly identified by \emph{CGRO}/EGRET and presumably some of the
199: unidentified gamma-ray sources will turn out to be
200: associated to young and energetic radio pulsars discovered in recent
201: radio surveys \citep{manchester01,kramer03}. In fact,
202: several unidentified gamma-ray sources
203: have characteristics similar to those of the known gamma-ray pulsars
204: (hard spectrum with high-energy cutoff, no variability, possible X-ray
205: counterparts with thermal/non-thermal component, no prominent optical
206: counterpart), but they lack a radio counterpart as well as a supernova
207: remnant and/or pulsar wind nebula association. Radio quiet,
208: Geminga-like objects have been invoked by several authors
209: \citep{romani95,yadigaroglu95,harding07} but without evidence of
210: pulsation in gamma-rays, no identification has been confirmed.
211:
212: Apart from the Crab and B\,1509--58, whose luminosities peak in the
213: 100 keV and about 30 MeV range respectively, the energy flux of the remaining gamma-ray
214: pulsars is dominated by the emission above 10 MeV with a spectral
215: break in the GeV range.
216:
217: \emph{AGILE} (Astro-rivelatore Gamma ad Immagini LEggero) is a small
218: scientific mission of the Italian Space Agency dedicated to
219: high-energy astrophysics \citep[][Tavani et al. 2008, in preparation]{tavani06} launched on 2007 April 23. Its
220: sensitivity in the 30 MeV--30 GeV range, with simultaneous X-ray
221: imaging in the 18--60 keV band, makes it perfectly suited for the
222: study of gamma-ray pulsars. Despite its small dimensions and weight
223: ($\sim$100 kg), the new silicon detector technology employed for the
224: \emph{AGILE} instruments yields overall performances as good as, or
225: better than, that of previous bigger instruments. High-energy photons
226: are converted into e$^{+}$/e$^{-}$ pairs in the Gamma-Ray Imaging
227: Detector (GRID) a Silicon-Tungsten tracker
228: \citep{prest03,barbiellini01}, allowing for an efficient photon
229: collection, with an effective area of $\sim$500 cm$^{2}$, and for an
230: accurate arrival direction reconstruction ($\sim$0.5$^{\circ}$ at 1
231: GeV) over a very large field of view, covering about 1/5 of the sky in
232: a single pointing. The Cesium-Iodide mini-calorimeter
233: \citep{labanti06} is used in conjunction with the tracker for photon
234: energy reconstruction while supporting the anti-coincidence shield in
235: the particle background rejection task \citep{perotti06}.
236: The \emph{AGILE}/GRID
237: is characterized by the smallest dead-time ever obtained for gamma-ray
238: detection (typically 200 $\mu$s) and time tagging with uncertainty
239: near $\sim$1 $\mu$s.
240: The SuperAGILE hard X-ray monitor is positioned on top of
241: the GRID. SuperAGILE
242: is a coded aperture instrument operating in the 18--60 keV energy band
243: with about 15 mCrab sensitivity in one
244: day integration, 6 arcmin angular resolution
245: and $\sim$1 sr field of view \citep{feroci07,costa01}.
246:
247: In this work we analyze all available \emph{AGILE}/GRID data suitable
248: for timing analysis collected up to 2008 April 10 for the four known
249: gamma-ray pulsars included in the \emph{AGILE} Team source
250: list:\footnote{See http://agile.asdc.asi.it for details about
251: \emph{AGILE} Data Policy and Target List.} Vela, Crab, Geminga and
252: B\,1706--44. The other two EGRET pulsars, B\,1055--52 and B\,1951+32,
253: are part of the \emph{AGILE} Guest Observer program. As expected, only the Crab
254: pulsar has been detected by SuperAGILE and the X-ray data have been
255: used to cross-check and test \emph{AGILE} timing performances.
256:
257: The \emph{AGILE} observations are presented in \S~\ref{datareduction},
258: as well as the criteria for photons selections. The observations and the
259: timing analysis from the parallel radio and X-ray observations of the
260: four targets are described in \S~\ref{sec:radio}. The procedures
261: for the timing analysis of gamma-ray data are introduced in
262: \S~\ref{timinganalysis} and the results of their application are
263: reported in \S~\ref{section-tctr}, where timing calibration tests
264: are also dealt with. Discussion of the scientific results and
265: conclusions are the subjects of \S\S~\ref{discuss} and \ref{conclusions},
266: respectively.
267:
268: \section{\emph{AGILE} Observations and Data Reduction}
269: \label{datareduction}
270:
271: The \emph{AGILE} spacecraft was placed in a Low Earth Orbit (LEO) at
272: $\sim$535 km mean altitude with inclination $\sim$2.5\degr. Therefore,
273: Earth occultation strongly affects exposure along the orbital plane,
274: as well as high particle background rate during South Atlantic Anomaly
275: transits. However, the exposure efficiency is $>$50\% for most
276: \emph{AGILE} revolutions. \emph{AGILE} pointings consist of long
277: exposures (typically lasting 10--30 days) slightly drifting
278: ($\lesssim$1\degr/day) with respect to the starting pointing direction in order
279: to match solar-panels illumination constraints. The relatively uniform
280: values of the effective area and point spread function within
281: $\sim$40\degr\ from the center of the field of view of the GRID, allow for
282: one-month pointings without significant vignetting in the exposure of
283: the target region.
284:
285: Pulsar data were collected during the mission Science Verification
286: Phase (SVP, 2007 July--November) and early pointings\footnote{1. Cygnus
287: Field 1 ($l=89\degr$, $b=9.9\degr$), 2. Virgo Field ($l=264\degr$,
288: $b=56.5\degr$), 3. Vela Field ($l=283\degr$, $b=-6.8\degr$), 4. South
289: Gal. Pole ($l=240\degr$, $b=-50\degr$), 5. Musca Field ($l=303\degr$,
290: $b=-9\degr$), 6. Gal. Center ($l=332\degr$, $b=0\degr$),
291: 7. Anti-Center ($l=193\degr$, $b=8.1\degr$).} (2007 December--2008 April)
292: of the AO\,1 Observing Program. It is worth noting that a single
293: \emph{AGILE} pointing on the Galactic Plane embraces about one-third
294: of it, allowing for simultaneous multiple source targeting
295: (e.g. the Vela and Anti-Center regions in the same field of view with
296: Crab, Geminga and Vela being observed at once; Figure \ref{pulsarmap}).
297: \begin{figure*} % figure 1
298: \centering
299: \resizebox{.8\hsize}{!}{\includegraphics[angle=00]{f1.eps}}
300: \caption{\label{pulsarmap} Gaussian-smoothed \emph{AGILE} intensity
301: map ($\sim$120\degr\ $\times$ 60\degr, units: ph cm$^{-2}$ s$^{-1}$ sr$^{-1}$, E$>$100 MeV) in Galactic coordinates
302: integrated over the whole observing period (2007 July 13--2008 April
303: 10) and centerd at $l=223\degr$, $b=0\degr$. The \emph{AGILE} field of
304: view (radius $\sim$60\degr) can embrace in a single pointing Vela ($l=263.6\degr$, $b=-2.8\degr$),
305: Geminga ($l=195.1\degr$, $b=4.3\degr$) and Crab ($l=184.6\degr$, $b=-5.8\degr$) as well as diffuse emission from the Galactic Disk.}
306: %(PSR~B1706$-$44 [$l=343.1\degr$, $b=-2.7\degr$] lies outside this map).}
307: \end{figure*}
308:
309: The \emph{AGILE} Commissioning and Science Verification Phases
310: lasted about seven months from 2007 April 23 to November 30 including
311: also Instrument Time Calibration. On 2007 December 1 baseline nominal
312: observations and a pointing plan started together with the Guest
313: Observer program AO\,1. Timing observations suitable for pulsed signal
314: analysis of the Vela pulsar started in mid 2007 July (at orbit 1,146)
315: after engineering tests on the payload.
316:
317: The Vela region was observed (with optimal exposure efficiency) for $\sim$40
318: days during the SVP and again for $\sim$30 days in AO\,1 pointing number 3
319: (2008 January 8--February 1). PSR B\,1706--44 was within the Vela and Galactic
320: Center pointings for $\sim$30 days during the SVP and for $\sim$45 days
321: during AO\,1 pointings numbers 5 and 6 (2008 February 14--March 30). The
322: Anti-center region (including Crab and Geminga) was observed for $\sim$40
323: days, mostly in 2007 September, and in 2008 April (AO\,1 pointing
324: number 7) with the addition of other sparse short Crab pointings for
325: SuperAGILE calibration purposes during the SVP (see Table \ref{counts}
326: and Figure \ref{fig:residual} for details about targets coverage).
327: \begin{deluxetable*}{clccccc} % table 1
328: % \tabletypesize{\scriptsize}
329: \tablecolumns{1}
330: \tablewidth{0pc}
331: \tablecaption{\label{counts} Observation parameters of the relevant
332: data subsets (grouped in uniterrupted target observations) for each
333: analyzed pulsar considering for comparison all G+L class events (see
334: text) with $E>100$ MeV observed within 60\degr\ from the center of the
335: field of view and extracted within 5\degr\ from the pulsar position.}
336: \tablehead{
337: \colhead{PSR} & \colhead{ObsID\tablenotemark{a}} & \colhead{T$_{FIRST}$} & \colhead{T$_{LAST}$} & \colhead{$\langle\theta\rangle$\tablenotemark{b}} & \colhead{Total Counts\tablenotemark{c}} & \colhead{Exposure\tablenotemark{d} } \\
338: \colhead{ } & \colhead{ } & \colhead{[MJD]} & \colhead{[MJD]} & \colhead{[deg]} & \colhead{ } & \colhead{[10$^8$ cm$^2$ s]} }
339:
340: \startdata
341:
342: Vela & SVP 1 & 54,294.552 & 54,305.510 & 26.2 & 6440 & 1.87 \\
343: Vela & SVP 2-3-4 & 54,311.512 & 54,344.569 & 40.9 & 12651 & 3.26 \\
344: %Vela & 3 & 54,322.896 & 54,337.063 & 42.5 & 6055 & ... \\
345: %Vela & 4 & 54,337.067 & 54,344.569 & 49.8 & 1377 & ... \\
346: Vela & SVP 5 & 54,358.525 & 54,359.521 & 41.4 & 391 & 0.11 \\
347: Vela & SVP 6 & 54,367.525 & 54,377.521 & 48.9 & 1355 & 0.26 \\
348: Vela & SVP 8 & 54,395.554 & 54,406.501 & 55.3 & 113 & 0.03 \\
349: Vela & AO\,1 2 & 54,450.540 & 54,457.157 & 56.2 & 564 & 0.01 \\
350: Vela & AO\,1 2-3 & 54,464.784 & 54,480.966 & 22.1 & 4726 & 3.13 \\
351: %Vela & 14 & 54,478.778 & 54,480.966 & 23.8 & 1403 & ... \\
352: Vela & AO\,1 3-4 & 54,492.948 & 54,505.377 & 43.2 & 6293 & 1.16 \\
353: Vela & AO\,1 4-5-6 & 54,508.528 & 54,528.537 & 43.6 & 9747 & 2.22 \\
354: %Vela & 17 & 54,521.295 & 54,528.537 & 45.8 & 2844 & ... \\
355: Vela & AO\,1 6 & 54,546.095 & 54,561.427 & 41.7 & 703 & 0.17 \\
356: %Vela & 19 & 54,549.629 & 54,561.427 & 43.3 & 228 & ... \\
357: Vela & total\tablenotemark{e} & 54,294.552 & 54,561.427 & 36.1 & 42983 & 12.22 \\
358: Vela & gamma\tablenotemark{f} & 54,294.552 & 54,561.427 & 36.1 & 6140 & 5.28 \\
359:
360: \hline
361:
362: Geminga & SVP 2 & 54,308.871 & 54,314.507 & 55.0 & 187 & 0.06 \\
363: Geminga & SVP 3 & 54,324.514 & 54,335.508 & 41.3 & 781 & 0.24 \\
364: Geminga & SVP 4-5-6-7 & 54,344.518 & 54,351.235 & 22.2 & 17823 & 6.19 \\
365: %Geminga & 5 & 54,351.240 & 54,365.407 & 29.7 & 5654 & ... \\
366: %Geminga & 6 & 54,365.411 & 54,379.576 & 13.5 & 6400 & ... \\
367: %Geminga & 7 & 54,379.582 & 54,386.502 & 11.2 & 3652 & ... \\
368: Geminga & SVP 8 & 54,395.520 & 54,406.504 & 37.6 & 1926 & 0.48 \\
369: %%%Geminga & AO\,1 3-4 & 54,494.514 & 54,505.390 & 57.7 & 23 & ... \\
370: %%%Geminga & AO\,1 4 & 54,508.511 & 54,510.473 & 55.6 & 192 & ... \\
371: Geminga & AO\,1 6 & 54,528.531 & 54,529.270 & 50.7 & 38 & 0.01 \\
372: Geminga & AO\,1 6 & 54,546.093 & 54,549.428 & 39.0 & 132 & 0.03 \\
373: Geminga & AO\,1 6-7 & 54,555.514 & 54,566.5 & 11.7 & 4860 & 2.03 \\
374: Geminga & total\tablenotemark{e} & 54,308.871 & 54,566.5 & 20.2 & 25962 & 9.04 \\
375: Geminga & gamma\tablenotemark{f} & 54,308.871 & 54,566.5 & 20.2 & 3874 & 4.54 \\
376:
377: \hline
378:
379: Crab & SVP 1-2 & 54,305.509 & 54,314.50 & 50.2 & 2404 & 0.58 \\
380: %Crab & 2 & 54,308.731 & 54,314.507 & 48.8 & 1423 & ... \\
381: Crab & SVP 3 & 54,324.512 & 54,335.508 & 28.2 & 761 &.0.32 \\
382: Crab & SVP 4-5-6-7 & 54,344.518 & 54,386.502 & 23.9 & 20499 & 6.41 \\
383: %Crab & 5 & 54,351.239 & 54,365.407 & 25.8 & 6660 &... \\
384: %Crab & 6 & 54,365.411 & 54,379.579 & 19.3 & 6707 &... \\
385: %Crab & 7 & 54,379.582 & 54,386.502 & 22.8 & 3853 & ... \\
386: Crab & SVP 8 & 54,395.518 & 54,406.506 & 45.4 & 1537 & 0.32 \\
387: Crab & AO\,1 3-4 & 54,494.447 & 54,505.390 & 55.8 & 350 & 0.05 \\
388: Crab & AO\,1 4 & 54,508.506 & 54,510.474 & 54.7 & 359 & 0.06 \\
389: Crab & AO\,1 6 & 54,528.531 & 54,529.140 & 46.9 & 100 & 0.02 \\
390: Crab & AO\,1 6 & 54,546.091 & 54,549.423 & 41.5 & 176 & 0.03 \\
391: Crab & AO\,1 6-7 & 54,555.516 & 54,566.5 & 21.8 & 4973 & 1.89 \\
392: Crab & total\tablenotemark{e} & 54,305.509 & 54,566.5 & 26.1 & 31159 & 9.68 \\
393: Crab & gamma\tablenotemark{f} & 54,305.509 & 54,566.5 & 26.1 & 4062 & 4.11 \\
394:
395: \hline
396:
397: B1706--44 & SVP 1 & 54,294.552 & 54,305.503 & 50.9 & 4641 & 0.91 \\
398: B1706--44 & SVP 2-3-4 & 54,311.531 & 54,322.891 & 34.5 & 15657 & 4.39 \\
399: %J1709--4429 & 3 & 54322.896 & 54337.063 & 35.3 & 6296 & ... \\
400: %J1709--4429 & 4 & 54337.068 & 54344.509 & 17.6 & 4501 & ... \\
401: B1706--44 & SVP 7-8 & 54,386.524 & 54,393.752 & 41.5 & 5521 & 1.28 \\
402: %J1709--4429 & 8 & 54393.753 & 54397.336 & 44.0 & 1497 & ... \\
403: %J1709--4429 & 14? & & & & 0 & ... \\
404: B1706--44 & AO\,1 2-3 & 54,470.292 & 54,480.965 & 53.7 & 2169 & 1.46 \\
405: %J1709--4429 & 14 & 54478.781 & 54480.965 & 52.6 & 767 & ... \\
406: B1706--44 & AO\,1 3 & 54,492.948 & 54,497.509 & 44.9 & 1507 & 0.20 \\
407: B1706--44 & AO\,1 5-6 & 54,510.482 & 54,555.509 & 22.4 & 24219 & 7.36 \\
408: %J1709--4429 & 17 & 54521.289 & 54535.453 & 17.8 & 7546 & ... \\
409: %J1709--4429 & 18 & 54535.459 & 54549.627 & 14.6 & 7351 & ... \\
410: %J1709--4429 & 19 & 54549.631 & 54555.509 & 23.8 & 3157 & ... \\
411: B1706--44 & total\tablenotemark{e} & 54,294.552 & 54,555.509 & 28.7 & 53714 & 15.6 \\
412: B1706--44 & gamma\tablenotemark{f} & 54,294.552 & 54,555.509 & 28.7 & 8463 & \phantom{6.56}6.56\enddata
413: \tablenotetext{a}{Observation ID: SVP=Science Verification Phase
414: (grouped in subset of 200 orbit each starting from 1,146),
415: AO\,1=Scientific Observations Program pointings (see \emph{AGILE}
416: Mission Announcement of Opportunity Cycle-1:
417: http://agile.asdc.asi.it/).}
418: \tablenotetext{b}{Mean off-axis angle.}
419: % from center of the field of view.}
420: \tablenotetext{c}{Source photons + diffuse emission photons + particle
421: background.}
422: \tablenotetext{d}{Good observing time after dead-time and occultation
423: corrections.}
424: \tablenotetext{e}{Total G+L class events
425: %(including ``gamma'' G events)
426: with $E>100$ MeV, 5\degr\ max from pulsar position, 60\degr\ max
427: from field of view center.}
428: \tablenotetext{f}{High-confidence photon events (G class only) with $E>100$
429: MeV, 5\degr\ max from pulsar position, 60\degr\ max from field of view
430: center.}
431: \end{deluxetable*}
432:
433: \begin{figure} % figure 2
434: \resizebox{\hsize}{!}{\includegraphics[angle=00]{f2.eps}}
435: \caption{\label{fig:residual} Post-fit timing residuals (in
436: milli-turns) as a function of the Modified Julian Day
437: resulting from the observation at 1.4~GHz of the three radio pulsars
438: which are discussed in this paper: from the top Vela, Crab and
439: PSR~B1706$-$44. The panels in the left (right) column report the
440: residuals of the best available timing solutions obtained over the
441: data span not including (including) the correction of the timing noise
442: via the use of the $\Delta R$ term (see \S~\ref{timinganalysis}).
443: Note that for the Crab pulsar the scale on the vertical axis of the
444: panel in the right column is amplified by a factor 10 with respect to
445: that in the left panel. The time intervals corresponding to the
446: useful \emph{AGILE} pointings for each target are also given as the
447: black sections of the bar at the bottom of each panel.}
448: \end{figure}
449:
450: GRID data of the relevant observing periods were grouped in 20 subsets
451: of 200 orbits each (corresponding to $\sim$15 days of observation)
452: starting from orbit 1,146 (54,294 MJD). Data screening, particle
453: background filtering and event direction and energy reconstruction
454: were performed by the \emph{AGILE} Standard Analysis Pipeline
455: (BUILD-15) for each subset with an exposure $>$$10^6$ cm$^2$ s at
456: $E>100$ MeV. Observations affected by coarse pointing, non-nominal
457: settings or intense particle background (e.g. orbital passages in
458: South Atlantic Anomaly) and albedo events from the Earth's limb were
459: excluded from the processing.
460:
461: A specific optimization on the events extraction parameters is
462: performed for each target in order to maximize the signal to noise
463: ratio for a pulsed signal. The optimal event extraction radius around
464: pulsar positions varies as a function of photon energy (and
465: then it is related to pulsar spectra) according to the Point Spread
466: Function. However, for $E>100$ MeV broad band analysis, a fixed
467: extraction radius of $\sim$5\degr\ (a value slightly higher than the
468: Point Spread Function 68\% containment radius) produces comparable
469: results with respect to energy-dependent extraction.
470:
471: Quality flags define different GRID event classes. The G event class
472: includes events identified with good confidence as photons. Such
473: selection criteria correspond to an effective area of $\sim$250 cm$^2$
474: above 100 MeV (for sources within 30 deg from the center of the
475: field-of-view). The L event class includes events typically affected by
476: an order of magnitude higher particle contamination than G, but
477: yielding an effective area of $\sim$500 cm$^2$ at $E>100$ MeV, if
478: grouped with the G class. We performed our timing analysis looking for
479: pulsed signals using both G class events and the combination of G+L
480: events. In general, the signal-to-noise ratio of the pulsed signal is
481: maximized using photons collected within $\sim$40\degr\ from the
482: center of GRID field of view and selecting the event class G. For
483: very strong sources, such as Vela, or sources located in low
484: background regions, it is possible to include also photons in the
485: 40--60\degr\ off-axis range and belonging to the G+L event class
486: typically %improving up to a factor of 3 the count statistics.
487: improving the count statistics by up to a factor of three (obviously
488: implying also a much higher background),
489: without affecting or even improving the detection significance.
490: For each pulsar, Table \ref{counts} summarizes the exposure parameters
491: of the relevant data subsets grouped in uninterrupted target
492: observations.
493: For simplicity and comparison, we reported in Table \ref{counts} and Table \ref{counts2}
494: observation and detection parameters obtained with the same extraction
495: criteria and energy range.
496: For all observations with the targets within 60\degr\
497: from the pointing direction, we list the average angular distance
498: $\langle\theta\rangle$ of the source position from the pointing direction, the
499: sum of all G+L events with $E>100$ MeV whose direction is within
500: 5\degr\ from the pulsar position and the G only events (``gamma
501: photons''), selected with the same criteria. The choice to include all
502: G+L events yields a ``dirty'' data set: for the case of Vela,
503: $\sim$10\% of the total counts are ascribable to the Galactic Plane,
504: and $>$50\% are particle background ($\sim$10,000 source counts).
505: The contamination is reduced in
506: the ``gamma'' entry, characterized by $\sim$20\% diffuse emission and
507: $\sim$5\% particle background ($\sim$5,000 source counts).
508: % Thus, looking at the Vela case,
509: %the dirty data set yields $\sim$10,000 source counts while the clean
510: %one
511: The corresponding exposure for each
512: data subset was calculated with the GRID scientific analysis task
513: AG\_ExpmapGen according with the above parameters. Summing
514: up the photon numbers, we see that the overall photon statistics
515: accumulated (accounting for background contamination) is comparable to
516: that collected by EGRET for the same four pulsars.
517: \begin{deluxetable*}{cccccrcc} % table 3
518: % \tabletypesize{\scriptsize}
519: %\rotate
520: \tablecolumns{1}
521: \tablewidth{0pc}
522: \tablecaption{\label{counts2} The photon harvest from the \emph{AGILE}
523: observations of gamma-ray pulsars.}
524: \tablehead{
525: \colhead{PSR} & \colhead{P} & \colhead{$\dot{E}$} & \colhead{$d$} & \colhead{Pulsed Counts\tablenotemark{a}} & \colhead{$\chi^{2}_{r}$ (d.o.f)} & \colhead{Exposure\tablenotemark{b}} & \colhead{Pulsed Flux\tablenotemark{c}} \\
526: \colhead{ } & \colhead{(ms)} &\colhead{(erg s$^{-1}$)} &\colhead{(kpc)} &\colhead{ } & \colhead{ } & \colhead{(10$^8$ cm$^2$ s)} & \colhead{10$^{-8}$ ph. cm$^{-2}$ s$^{-1}$}}
527:
528:
529: \startdata
530:
531: %Vela & 9,173$\pm$580 & 225.51 (9) & 1.24 & 927$\pm$59 \\
532: %Geminga & 1,897$\pm$480 & 10.44 (9) & 0.85 & 280$\pm$71 \\
533: %Crab & 2,001$\pm$530 & 10.71 (9) & 0.92 & 271$\pm$72 \\
534: %J1709--4429 & 2,364$\pm$720 & 9.11 (9) & 1.56 & 189$\pm$57 \\
535:
536:
537: Vela & 89.3 & $6.92\times10^{36}$ & 0.29 & $9,170\pm580$ & 225.51 (9) & 12.22 & $940\pm60$\\
538: Geminga & 237.1 & $3.25\times10^{34}$ & 0.16 & $2,200\pm480$ & 10.44 (9) & 9.04 & $300\pm70$\\
539: Crab & 33.1 & $4.61\times10^{38}$ & 2.00 & $2,120\pm530$ & 10.71 (9) & 9.68 & $270\pm70$\\
540: B1706--44& 102.5 & $3.41\times10^{36}$ & 1.82 & $2,370\pm720$& 9.11 (9) & 15.6 & \phantom{10}$190\pm60$\enddata
541: \tablecomments{See text for details on data reduction and timing analysis.}
542: \tablenotetext{a}{Pulsed counts (G+L event class) with $E>100$ MeV,
543: 5\degr\ max from pulsar position, 60\degr\ max from field of view
544: center, 10 bins.}
545: \tablenotetext{b}{Good observing time after dead-time and occultation
546: corrections.}
547: \tablenotetext{c}{Calculated with the expression $C_P f/E$, where
548: $C_p$=pulsed counts, $E$=exposure, $f$=factor accounting for source
549: counts at angular distance $>$5\degr\ from source position according
550: to the point spread function ($f\sim1.25$).}
551: \end{deluxetable*}
552:
553:
554: A maximum likelihood analysis (ALIKE task) on the \emph{AGILE}
555: data from the sky areas containing the four gamma-ray pulsars yielded
556: source positions, fluxes and spectra in good agreement with those
557: reported both in the EGRET catalogue for the corresponding 3EG sources
558: \citep{hartman99} and in the revised catalogue by \citet{casandjian08}.
559: As an example, Figure \ref{pulsarmap} reports an \emph{AGILE} intensity
560: map displaying Vela, Geminga and Crab. Details on gamma-ray imaging and
561: spectra of the four sources will be reported in future papers
562: when count statistics significantly higher than that currently available
563: are collected and in-flight Calibration files are finalized.
564:
565: In this paper we focus on timing analysis. However, it is worth
566: noticing that pulsed counts provide a gamma-ray pulsar flux
567: estimate independent from the likelihood analysis, as described in
568: \S~\ref{section-tctr}.
569:
570: \section{Radio/X-rays observations and timing}
571: \label{sec:radio}
572:
573: In order to perform \emph{AGILE} timing calibration through accurate
574: folding and phasing, as described in \S\S~\ref{timinganalysis}
575: and \ref{section-tctr}, pulsar timing solutions valid for the epoch of
576: \emph{AGILE} observations were required. Thus, a dedicated pulsar
577: monitoring campaign (that will continue during the whole \emph{AGILE}
578: mission) was undertaken, using two telescopes (namely
579: Jodrell Bank and Nan\c cay) of the European Pulsar Timing Array
580: (EPTA), as well as %the Parkes and the Mt. Pleasant radio telescopes
581: %of The Australia Telescope National Facility (ATNF).
582: the Parkes radio telescope of the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Australia Telescope
583: National Facility (ATNF) and the 26m Mt Pleasant radio telescope operated by the
584: University of Tasmania.
585:
586: In particular, the observations of the Vela Pulsar have been secured
587: by the Mt.~Pleasant Radio Observatory. Data
588: were collected at a central frequency of 1.4~GHz. Given the pulsar
589: brightness, it is possible to extract a pulse Time of Arrival (ToA)
590: about every 10 seconds, so that a total of 4,098 ToAs have been
591: obtained for the time interval between 2007 February 26 (MJD 54,157) and
592: 2008 March 23 (MJD 54,548), encompassing the whole time span of the
593: \emph{AGILE} observations. During this time interval the pulsar
594: experienced a small glitch (fractional frequency increment
595: $\Delta\nu_g / \nu \simeq 1.3 \times 10^{-9}$), which presumably
596: happened around 2007 August 1 (with an uncertainty of $\pm$3 days
597: due to the lack of observations around this period) (see \S~\ref{velaglitch}). Ephemeris for the pre- and post-glitch time
598: intervals were then separately calculated and applied to the gamma-ray
599: folding procedure.
600:
601: The observations of the Crab Pulsar have been provided by the
602: telescopes of Jodrell Bank and Nan\c cay. At Jodrell Bank the Crab
603: Pulsar has been observed daily from 2006 December 09 (MJD 54,078) to
604: 2007 October 06 (MJD 54,379) and again from 2008 February 06
605: (MJD 54,502) to 2008 April 10 (MJD 54,566) which result in 334 ToAs.
606: The observations were mainly performed at 1.4~GHz with the 76 m Lovell
607: telescope, with some data also taken with
608: %the 25m Mark II
609: the 12-m telescope at a central frequency of 600~MHz. The Nan\c cay
610: Radio Telescope (NRT) is a transit telescope with the equivalent area of a 93 m dish and observed at a
611: central frequency of 1.4~GHz over the time interval between 2006 December 05
612: (MJD 54,074) and 2007 September 07 (MJD 54,350), producing 64
613: ToAs. A timing solution was produced by joining the ToAs from the two
614: telescopes and accounting for the phase shift that naturally ensues from
615: data sets coming from different telescopes.
616:
617: The observations of pulsar B\,1706$-$44 have been performed at the
618: 64 m telescope at Parkes, in Australia, at the mean observing frequency
619: of 1.4~GHz. They produced a total of 20 ToAs over the interval from 2007 April
620: 30 (MJD 54,220) to 2008 Apr 6 (MJD 54,562), thus covering the
621: whole \emph{AGILE} observing time span for this target.
622:
623: The timing of all pulsars is performed using the TEMPO2 software
624: \citep{hobbs06,edwards06}. It first converts the topocentric
625: TOAs to solar-system barycentric TOAs at
626: infinite frequency\footnote{Dispersion measure is obtained as part of timing
627: solution for Crab (DM=56.76(1)), Vela (DM=68.15(2)) and from \citet{johnston95}
628: for B\,1706$-$44 (DM=75.69(5)).}
629: (using the Jet Propulsion Laboratory [JPL] DE405 solar-system
630: ephemeris\footnote{See ftp://ssd.jpl.nasa.gov/pub/eph/export/DE405/de405.iom/.})
631: and then performs a multi-parameter least-square fit to determine the
632: pulsar parameters. The differences between the observed barycentric
633: ToAs and those estimated by the adopted timing model are represented
634: by the so-called $residuals$. The procedure is iterative and improves
635: with longer timescales of observations. An important feature developed
636: by TEMPO2 is the possibilty of accounting for the timing noise in the
637: fitting procedure (see \S~\ref{timinganalysis}). This is
638: particularly useful in timing the young pulsars: in fact most of them
639: suffer of quasi-random fluctuations (typically characterized by a very
640: red noise spectrum) in the rotational parameters, whose origin is
641: still debated. TEMPO2 corrects the effects of the timing noise on the
642: residuals by modeling its behaviour as a sum of harmonically
643: correlated sinusoidal waves (see Equation [\ref{phasing2}] in
644: \S~\ref{timinganalysis}) that is subtracted from the residuals.
645:
646: The left panels (labeled as {\it pre-whitening}) in
647: Figure~\ref{fig:residual} reports the timing residuals obtained over the
648: data span of the radio observations without the correction for the
649: timing noise, whereas the right panels (labeled as {\it whitened})
650: show the residuals after the application of the correction. The
651: comparison between the panels in the two columns shows that this
652: procedure has been very effective in removing the timing noise in
653: Vela, Crab and PSR~B1706$-$44. The impact of this whitening procedure
654: on the gamma-ray data analysis is discussed in \S~\ref{section-tctr}.
655:
656: Geminga is a radio-quiet pulsar whose ephemeris can be obtained from
657: X-ray data. Following the demise of \emph{CGRO}, Geminga was
658: regularly observed with \emph{XMM-Newton} in order to maintain its
659: ephemeris for use in analyzing observations at other wavelengths.
660: We analyzed all eight observations of Geminga (1E\,0630+178,
661: PSR\,J\,0633+1746) taken with the X-ray
662: (0.1--15 keV) cameras of \emph{XMM-Newton} \citep{jansen01} between
663: MJD 52,368 and MJD 54,534 (2002 April - 2008 March), with exposures
664: in the range 20-100 ks.
665:
666:
667:
668: %%We
669: %%%analysed all the
670: %%%%available
671: %%%data of Geminga (1E\,0630+178,
672: %%%PSR\,J\,0633+1746) taken with the X-ray (0.1--15 keV) cameras of
673: %%%\emph{XMM-Newton} \citep{jansen01};
674: %\footnote{The \emph{XMM-Newton} satellite
675: %\citep{jansen01} carries three 1500 cm$^2$ X-ray telescopes with an
676: %EPIC instrument in each focus, a reflecting grating spectrometer
677: %(RGS), and an optical monitor. Two of the EPIC imaging cameras use MOS
678: %CCDs and one uses a pn CCD. The data were processed using version
679: %7.1.0 of the \emph{XMM-Newton} Science Analysis Software (SAS) and the
680: %calibration files released in 2007 August.}
681: %%%in particular the publicly available data in the \emph{XMM-Newton} Science Archive\footnote{See http://xmm.esac.esa.int/.} include six observations (data span
682: %%%52,368--54,170 MJD) with exposures in the 20--100 ks range.
683:
684: The data were processed using version 7.1.0 of the \emph{XMM-Newton}
685: Science Analysis Software (SAS) and the calibration files released in 2007 August.
686: For the timing analysis we could only use the pn data (operated in
687: Small Window mode: time resolution 6 ms, imaging on a
688: $4\arcmin\times4\arcmin$ field), owing to the inadequate time
689: resolution of the MOS data. We selected only single and double photon
690: events (patterns 0--4) and applied standard data screening
691: criteria. Photons arrival times were converted to the Solar system
692: barycenter using the SAS task barycen. To extract the source photons
693: we selected a circle of 30$\arcsec$ radius, containing about 85\% of
694: the source counts. Using standard folding and phase-fitting
695: techniques, the source pulsations were clearly detected in all the
696: observations. We derived for Geminga the long-term spin parameters
697: valid for the epoch range 52,369-54,534 MJD.
698: %summarized in Table \ref{tab:geminga}.
699: Note that the absolute accuracy
700: of the \emph{XMM-Newton} clock is better than 600 $\mu$s
701: \citep{kirsch04timing}.
702:
703: The Crab Pulsar is embedded in the Crab Nebula and represents about
704: 10\% of its flux in the hard X-ray band. Being a bright source with a
705: relatively soft high-energy spectrum, the Crab pulsar is easily
706: detected by SuperAGILE in less than one \emph{AGILE} orbital
707: revolution. Since the on-board time reconstruction of SuperAGILE
708: \citep{feroci07} is different from that of the GRID, as a cross-check and test
709: of the SuperAGILE timing performances we processed the X-ray data of
710: an on-axis observation of $\sim$0.7 days (54,360.7--54,361.4 MJD),
711: corresponding to an effective exposure of $\sim$41 ks. In the analysed
712: data the passages of \emph{AGILE} in the South Atlantic Anomaly and
713: intervals of source occultation by the Earth were excluded. The time
714: entries in SuperAGILE event files are in the on-board time reference
715: system (Coordinated Universal
716: Time) and were converted in Terrestrial Dynamical Time before being processed and
717: analysed by the same procedures used for the GRID
718: (see \S~\ref{timinganalysis}).
719: %The Crab SuperAGILE light-curve was then
720: %produced with the same folding method reported in
721: %\S~\ref{timinganalysis}, yielding 63,700$\pm$8,700 pulsed counts
722: %($\sim$3\% of the total flux). It is shown in
723: %Figure \ref{crab} and discussed in \S~\ref{section-tctr}.
724: A complete
725: analysis of the X-ray observations of Crab with SuperAGILE is beyond the
726: scope of this work: it will be presented in a future paper, as well as
727: the search for the Crab pulsed signal in the \emph{AGILE}
728: mini-calorimeter data.
729:
730: \section{Gamma-ray timing procedures}
731: \label{timinganalysis}
732:
733: In this section we will describe the timing procedures we have adopted.
734: They have been implemented with two aims: to verify the timing performances
735: of {\em AGILE} and to maximize the quality of the detection of the
736: four targets.
737: %Results about the application of these procedures will
738: %be reported in \S~\ref{section-tctr}.
739:
740: \emph{AGILE} on-board time is synchronized to Coordinated Universal
741: Time (UTC) by Global Positioning System (GPS) time sampled at a rate
742: of 1 Hz. Arrival time entries in \emph{AGILE} event lists files are
743: then corrected to Terrestrial Dynamical Time (TDT) reference system at
744: ground segment level. In order to perform timing analysis, they have
745: also to be converted to Barycentric Dynamical Time (TDB) reference and
746: corrected for arrival delays at Solar System Barycenter (SSB). This
747: conversion is based on the precise knowledge of the spacecraft
748: position in the Solar System frame. To match instrumental microsecond
749: absolute timing resolution level, the required spacecraft positioning
750: precision is $\lesssim$0.3 km. This goal is achieved by the
751: interpolation of GPS position samples extracted from telemetry packets.
752: Earth position and velocity with respect to SSB are then
753: calculated by JPL planetary ephemeris DE405. All the above barycentric
754: corrections are handled by a dedicated program (implemented in the
755: \emph{AGILE} standard data reduction pipeline) on the event list extracted
756: according to the criteria described in \S~\ref{datareduction}.
757:
758: Pulsed signals in GRID gamma-ray data cannot be simply found by
759: Fourier analysis of the photon SSB arrival times, since the pulsar
760: rotation frequencies are 4--5 order of magnitude higher than the
761: gamma-ray pulsars typical count-rates (10--100 counts/day). The
762: determination of the pulsar rotational parameters in gamma-ray must
763: then start from an at least approximate knowledge of the pulsar spin
764: ephemeris, provided by observations at other wavelengths.
765:
766: Standard epoch folding is performed over a tri-dimensional grid
767: centered on the nomimal values of the pulsar spin frequency $\nu_0$
768: and of its first and second order time derivatives, $\dot{\nu_0}$ and
769: $\ddot{\nu_0},$ as given by the assumed (radio or X--ray) ephemeris at
770: their reference epoch $t_0= \rm PEPOCH$. The axes of the grid are explored
771: with steps equal to 1/T$_{span},$ 2/T$_{span}^2$ and 6/T$_{span}^3$
772: respectively (all of them oversampled by a factor 20), where
773: T$_{span}$ is the time span of the gamma-ray data. For any assigned
774: tern $[\nu;~\dot\nu;~\ddot\nu],$ the pulsar phase $\Phi^{*}$ associated to
775: each gamma-ray photon is determined by the expression:
776: \begin{equation}
777: \label{phasing1}
778: \Phi^{*}=\Phi_{0}+\nu\Delta t+\frac{1}{2}\dot{\nu}\Delta
779: t^2+\frac{1}{6}\ddot{\nu}\Delta t^3
780: \end{equation}
781: \noindent
782: where $\Delta t=t-t_0$ is the difference between the SSB arrival time
783: $t$ of the photon and the reference epoch $t_0$ of the ephemeris and
784: $\Phi_0$ is a reference phase (held fixed for all the set). A
785: light-curve is formed by binning the pulsar phases of all the photons
786: and plotting them in a histogram. Pearson's $\chi^2$ statistic is
787: then applied to the light-curves resulting from each set of spin
788: parameters, yielding the probabilities of sampling a uniform
789: distribution. These probabilities ${\cal{P}}(\nu;~\dot\nu;~\ddot\nu)$
790: are then weighted for the number ${\cal{N}}_{\rm st}$ of steps over
791: the grid which has been necessary to reach the set
792: $[\nu;~\dot\nu;~\ddot\nu]$ starting from
793: $[\nu_0;~\dot\nu_0;~\ddot\nu_0].$ The maximum value over the grid of
794: ${\cal{S}}=1- {\cal{N}}_{\rm st}{\cal{P}}(\nu;~\dot\nu;~\ddot\nu)$
795: finally determines which are the best gamma-ray rotational parameters
796: for the target source in the surroundings of the given ephemeris. Of
797: course, the higher the value of ${\cal{S}}$, the higher is the
798: statistical significance of a pulsating signal. We note that this
799: approach allows to avoid any arbitrariness in the choice of the range
800: of the parameters to be explored, which otherwise can affect the
801: significance of a detection.
802:
803: %Once the best set of gamma-ray spin parameters has been determined,
804: %it is also possible to compare them with the quoted uncertainties in
805: %the parameters of the original ephemeris. This comparison, as well as
806: %that of the values of ${\cal{S}},$ clearly showed the advantage of
807: %using ephemeris (derived in radio or in X-ray bands) whose range of
808: %validity brackets the gamma-ray observations.
809: Due to the brightness of
810: the sources discussed in this paper, period folding around the
811: extrapolated pulsar spin parameters obtained from publicly available
812: ephemerides (ATNF Pulsar Catalog,\footnote{See http://www.atnf.csiro.au/research/pulsar/psrcat/.}
813: \citealt{manchester05}, and Jodrell Bank Crab ephemeris
814: archive,\footnote{See http://www.jb.man.ac.uk/$\sim$pulsar/crab.html.}
815: \citealt{lyne93}) or from recent literature led us to
816: firmly detect ($>$$5\sigma$) the pulsations for all the four targets
817: with a reasonable number of trials ($\lsim$1,000). However, for three
818: of them (Crab, Vela and PSR B1706$-$44), the best gamma-ray spin
819: period fell outside the $3\sigma$ uncertainty range of the adopted
820: radio ephemeris, and the detection significance ${\cal{S}}$ resulted
821: lower than that derived using contemporary timing solutions
822: (accounting for the effects of timing noise and/or occasional glitches),
823: provided by dedicated radio observation campaigns (see \S~\ref{sec:radio}).
824: %:e.g. for a few day long observation of Crab, the period discrepancy between the publicly available
825: %radio ephemeris valid at $\sim$53,400 MJD and the best set of gamma-ray parameters was $\Delta
826: %P=8 \times 10^{-3}~\mu$s (while the nominal $3\sigma$ uncertainty on the period
827: %was $10^{-3}~\mu$s), with ${\cal{S}}=0.2$ (while ${\cal{S}}=0.9999$ was
828: %obtained using radio ephemeris covering the time span of the gamma-ray
829: %observation).
830: %This reflects the effects of timing noise and/or
831: %occasional glitches which affect the rotational stability of most
832: %young and Vela-like pulsars. As a consequence, more trials (over the
833: %rotational parameters grid) are needed for optimizing the signal when
834: %using old ephemeris, thus reducing the overall sensitivity to the
835: %pulsed signal.
836: As expected, for steady and older pulsars, such as
837: Geminga, the availability of contemporary rotational parameters
838: turned out to be less important; even using a few-year-old ephemeris, the
839: pulsed signal could be detected within only $<$100 period search trials
840: around the extrapolated X-ray timing solutions.
841:
842: An additional significant improvement (see \S~\ref{section-tctr}) in
843: the detection significance has been obtained by accounting also for the
844: pulsar timing noise in the folding procedure. This exploits a tool
845: of TEMPO2 \citep{hobbs06,edwards06}, namely the possibility of fitting
846: timing residuals with a polynomial harmonic function $\Delta R$ in
847: addition to standard positional, rotational and (when appropriate)
848: binary parameters \citep{hobbs04}:
849: \begin{equation}
850: \label{phasing2}
851: \Delta R(\Delta t)=\sum_{k=1}^{N}a_k\sin(k\omega\Delta
852: t)+b_k\cos(k\omega\Delta t)
853: \end{equation}
854: \noindent
855: where $N$ is the number of harmonics (constrained by precision
856: requirements on radio timing residuals, as well as by the span and the
857: rate of the radio observations), $a_k$ and $b_k$ are the fit
858: parameters (i.e. the $\it{WAVE}$ terms in TEMPO2 ephemeris files), and
859: $\omega=2\pi(T_{radio}(1+4/N))^{-1}$ is the main frequency
860: (i.e. $\it{WAVE_{OM}}$ in TEMPO2 ephemeris files) related to the radio
861: data time-span $T_{radio}$.
862:
863: If the spin behavior of the target is suitably sampled, the harmonic
864: function $\Delta R$ can absorb the rotational irregularities of the
865: source, in a range of timescales ranging from $\sim$$T_{radio}$ down to
866: about the typical interval between radio observations. As an example, the
867: peak-to-peak amplitude of the $\Delta R$ fluctuations for Crab related
868: to the radio monitoring epochs (54,074--54,563 MJD) covering our
869: \emph{AGILE} observations are of the order of $\sim$1 ms,
870: corresponding to a phase smearing $>$0.03, a value significantly
871: affecting the time resolution of a $>$50 bins light-curve. Under the
872: assumption that the times of arrival of the gamma-ray photons are
873: affected by timing noise like in the radio band, gamma-rays folding
874: can properly account for $\Delta R$, extending the relation
875: \ref{phasing1} to:
876:
877: \begin{eqnarray}
878: \label{phasing3}
879: %\Phi=\Phi_{0}+\nu\Delta t+\frac{1}{2}\dot{\nu}\Delta
880: %t^2+\frac{1}{6}\ddot{\nu}\Delta t^3+ (\nu+\dot{\nu}\Delta
881: %t+\frac{1}{2}\ddot{\nu}\Delta t^2)\Delta R
882: %\simeq\Phi_{0}+\nu\Delta t(1+\frac{\Delta R}{\Delta t})+\frac{1}{2}\dot{\nu}\Delta
883: %t^2+\frac{1}{6}\ddot{\nu}\Delta t^3
884: \Phi & = & \Phi^{*}+ (\nu+\dot{\nu}\Delta
885: t+\frac{1}{2}\ddot{\nu}\Delta t^2)\Delta R
886: \nonumber\\
887: & \simeq & \Phi_{0}+\nu\Delta t(1+\frac{\Delta R}{\Delta t})+\frac{1}{2}\dot{\nu}\Delta
888: t^2+\frac{1}{6}\ddot{\nu}\Delta t^3
889: \end{eqnarray}
890:
891: As reported in \S~\ref{section-tctr}, this innovative phasing
892: technique improves significantly the gamma-ray folding accuracy for
893: young and energetic pulsars, especially when using long data spans,
894: like those of the \emph{AGILE} observations. Of course, the
895: implementation of this procedure requires radio observations
896: covering the time span of the gamma-ray observations
897: making the radio monitoring described in \S~\ref{sec:radio}
898: all the more important.
899: % and thus a
900: %dedicated radio monitoring campaign (see \S~\ref{sec:radio}) was
901: %organized.
902:
903: \section{Timing calibration tests and results}
904: \label{section-tctr}
905:
906: In order to verify the performances of the timing analysis procedure
907: described in \S~\ref{timinganalysis}, a crucial parameter to check is
908: the difference between pulsar rotation parameters derived from radio,
909: X-ray and gamma-ray data. Figure \ref{redchi} shows the \emph{AGILE}
910: period search result for the Crab pulsar (corresponding to the MJD
911: 54,305--54,406 observations, significantly affected by timing noise,
912: as shown in Figure \ref{fig:residual}). The implementation of the folding
913: method described in section \S~\ref{timinganalysis} (including timing
914: noise corrections as given by Equation [\ref{phasing3}]) allowed for a
915: perfect match between the best period resulting from gamma-ray data
916: and the period predicted by the radio ephemeris with
917: discrepancies $\Delta P_{\rm Crab}\sim3\times10^{-12}$ s, comparable
918: to the period search resolution $r_{\rm Crab}\sim2\times10^{-12}$
919: s. This represents also an ultimate test for the accuracy of the
920: on-board \emph{AGILE} Processing and Data Handling Unit
921: \citep[PDHU;][]{argan04} time management (clock stability in
922: particular) and on-ground barycentric time correction
923: procedure. Standard folding without $\Delta R$ term implies
924: radio-gamma period discrepancies one order of magnitude higher
925: ($\Delta P_{\rm Crab}\sim4\times10^{-11}$ s). Moreover, it lowers also
926: the statistical significance of the detection and effective time
927: resolution of the light-curve.
928: %For the Crab, the value of the probability
929: %${\cal S}$ introduced in \S~\ref{timinganalysis} goes from $XXXX$
930: %(when folding with the $\Delta R$ term) down to $XXXXX$ (ignoring the
931: %$\Delta R$ term).
932: For the Crab, the value of the Pearson's $\chi^2$ statistic
933: introduced in \S~\ref{timinganalysis}
934: (we quote reduced $\chi^2$ values)
935: goes from $\sim$6.3 (when
936: using the $\Delta R$ term) down to $\sim$4.2 (ignoring the
937: $\Delta R$ term) when folding the data into a 50 bins light-curve (see
938: Figure~\ref{crabcompare}). Obviously,
939: % such discrepancies (and
940: %associated smearing effects on the light-curves) arising from not
941: %considering timing noise in the folding process, are expected to grow
942: %on longer timescales and
943: ignoring timing noise in the folding process would yield discrepancies
944: (and light curve smearing) which are expected to grow when considering
945: longer observing time span. Thus the contribution of timing noise
946: should be considered both in high-resolution
947: timing analysis and in searching for new gamma-ray pulsars. The same
948: analysis applied to Vela and PSR~B\,1706--44 -- much less affected
949: than Crab by timing noise in the considered data span -- led to
950: similar results for the period discrepancies ($\Delta P_{\rm
951: Vela}\sim8\times10^{-12}$ s, $\Delta P_{\rm
952: B1706}\sim3\times10^{-11}$, to be compared with the period search
953: resolution $r_{\rm Vela}\sim9\times10^{-12}$ s and $r_{\rm
954: B1706}\sim2\times10^{-11}$ s).
955: \begin{figure} % figure 3
956: \resizebox{\hsize}{!}{\includegraphics[angle=00,width=15cm]{f3.eps}}
957: \caption{\label{redchi} Period search result for the Crab pulsar (period
958: trials vs. $\chi^2$ Pearson statistics). The radio period (vertical
959: line at $P_{\rm trial}-P_{\rm radio}=0$) is 33.607554009(4) ms
960: ($PEPOCH=54,362.242$ MJD). Dashed line is obtained from standard folding
961: period trials, while continuos line is from the folding technique
962: accounting for timing noise (see \S~\ref{sec:radio}). The new method
963: allow us to perfectly match the radio period in gamma rays ($P_{\rm
964: gamma\_tn}-P_{\rm radio}\sim3\times10^{-12}$ s). It is worth noting that
965: in the considered data span (54,305--54,406 MJD) the period search
966: resolution ($r_{\rm Crab}\sim2\times10^{-12}$ s) is about an order of
967: magnitude higher than the 1$\sigma$ error on the radio period.}
968: \end{figure}
969: \begin{figure} % figure 4
970: \resizebox{\hsize}{!}{\includegraphics[angle=00,width=15cm]{f4.eps}}
971: \caption{\label{crabcompare} The Crab light-curve (50 bin)
972: corresponding to the data span 54,305--54,406 MJD
973: obtained by folding including $\Delta R$ terms compared with that
974: obtained neglecting timing noise (dashed line).
975: In observing periods strongly affected by timing noise, the smearing
976: effects reduce detection
977: significance and observed pulsed counts (see text).}
978: \end{figure}
979:
980: %and the detection
981: %significance, going from $XXXX$ and $XXXX$ (when including the $\Delta R$
982: %term) down to $XXXXX$ and $XXXX$ (ignoring the $\Delta R$ term),
983: %respectively.
984:
985: For the radio-quiet Geminga pulsar we used X-ray ephemeris obtained from
986: \emph{XMM-Newton} data (see \S~\ref{sec:radio}) as a starting
987: point for the period search. Due to the stability of the spin
988: parameters of this relatively old pulsar, not significantly affected
989: by timing noise, $\it{WAVE}$ parameters are not required for the
990: folding process. The X-ray versus gamma-ray period discrepancy was
991: $\Delta P_{\rm Gem}\sim9\times10^{-12}$ s, whereas
992: $r_{\rm Gem}\sim7\times10^{-11}$ s. We here
993: note that the frequency resolution 1/T$_{\rm span}\sim10^{-7}$ Hz of
994: \emph{AGILE} is about one order of magnitude better than that
995: corresponding to a single \emph{XMM-Newton} exposure (lasting $\lsim100$
996: ks), but the few year long X-ray data span implies an overall much
997: better effective resolution 1/T$_{\rm XMM}\sim5\times10^{-9}$ Hz for the
998: \emph{XMM-Newton} data.
999:
1000: The resulting gamma-ray light-curves, covering different energy ranges
1001: for the four pulsars, are shown in Figures \ref{vela}, \ref{geminga},
1002: \ref{crab}, and \ref{J1709}. The pulsed flux was computed considering
1003: all the counts above the minimum of the light-curve, using
1004: the expression $PF=C_{\rm{tot}}-n N_{\rm{min}}$ and its associated
1005: error $\sigma_{PF}=(C_{\rm{tot}}+n^2 \sigma^2_{N_{\rm{min}}})^{1/2}
1006: \simeq n (N_{\rm{min}})^{1/2}$, where $C_{\rm{tot}}$ are the total
1007: counts, $n$ is the number of bins in the light-curve and
1008: $N_{\rm{min}}$ are the counts of bin corresponding to the minimum.
1009: This method is ``bin dependent'', but reasonable different choices of
1010: both the number of bins (i.e. $n>10$) and of the location of the bin
1011: center (10 trial values were explored for each choice of $n$) do not
1012: significantly affect the results. Several models (polar cap, slot
1013: gap) predict that gamma-ray pulsar emission is present at all phases.
1014: %(typically $\sim$xx\% of the total emission).
1015: When enough counts statistics will be available,
1016: (e.g. $\sim$10,000 counts for the Crab pulsar),
1017: it will be possible to estimate the unpulsed emission (due to
1018: the pulsar plus a possible contribution from the pulsar wind nebula)
1019: by considering the difference among the total source flux obtained by
1020: a likelihood analysis on the images and the pulsed flux estimated with the above
1021: method.
1022: \begin{figure} % figure 5
1023: \resizebox{\hsize}{!}{\includegraphics[angle=00,width=15cm]{f5.eps}}
1024: \caption{\label{vela} Vela pulsar light-curves ($P\sim89.3$ ms) for
1025: different energy bands ($E<100$ MeV, 40 bins, resolution: $\sim$2.2 ms;
1026: $E>100$ MeV, 100 bins, resolution: $\sim$0.9 ms; $E>1$ GeV, 20 bins,
1027: resolution: $\sim$4.5 ms; G+L class events) obtained by integrating all available
1028: post-glitch data (54,320--54,561 MJD). The radio ephemeris and
1029: the 8,192 bins light-curve
1030: (bottom panel) are obtained by the analysis of
1031: $\sim$4,100 ToAs observed at Mt.~Pleasant Radio Observatory in
1032: Tasmania (radio observation interval 54,157--54,548 MJD).}
1033: \end{figure}
1034: \begin{figure} % figure 6
1035: \resizebox{\hsize}{!}{\includegraphics[angle=00,width=15cm]{f6.eps}}
1036: \caption{\label{geminga} Geminga light-curves ($P\sim237.1$ ms) for
1037: different energy bands ($E<100$ MeV, 20 bins, resolution: $\sim$11.8 ms;
1038: $E>100$ MeV, 100 bins, resolution: $\sim$2.4 ms; $E>1$ GeV, 20 bins,
1039: resolution: $\sim$11.8 ms; G class events).
1040: The X-ray ephemeris and the 1--8 keV 40 bins light-curve
1041: (bottom panel) are obtained by the analysis of \emph{XMM-Newton}
1042: data (observation interval 52,369--54,534 MJD).}
1043: %corresponds to the ephemeris in Table \ref{tab:geminga}.}
1044: \end{figure}
1045: \begin{figure} % figure 7
1046: \resizebox{\hsize}{!}{\includegraphics[angle=00,width=13cm]{f7.eps}}
1047: \caption{\label{crab} Crab pulsar light-curves ($P\sim33.1$ ms) for
1048: different energy bands ($E>100$ MeV, 50 bins, resolution: $\sim$0.7 ms;
1049: $E>500$ MeV, 40 bins, resolution: $\sim$0.8 ms; G class events) obtained by integrating all available
1050: data presented in Table \ref{counts}. The X-ray (18--60 keV)
1051: SuperAGILE 50 bins light-curve is obtained from a $\sim$41 ks observation
1052: taken at $\sim$54,361 MJD. The radio ephemeris and
1053: the 2048 bins light-curve
1054: (bottom panel) are obtained by the analysis of
1055: 334 ToAs observed by Jodrell Bank and Nan\c cay radio telescopes (observation
1056: interval 54,078--54,566 MJD).}
1057: \end{figure}
1058: \begin{figure} % figure 8
1059: \resizebox{\hsize}{!}{\includegraphics[angle=00,width=15cm]{f8.eps}}
1060: \caption{\label{J1709} PSR B\,1706--44 light-curves ($P\sim102.5$ ms)
1061: for different energy bands ($E>30$ MeV, G class events, 40 bins, resolution: $\sim$2.6 ms;
1062: $0.1<E<1$ GeV, G+L class events, 20 bins, resolution: $\sim$5.1 ms; $E>1$ GeV, G+L class events, 10 bins,
1063: resolution: $\sim$10.2 ms) obtained by integrating all available
1064: data presented in Table \ref{counts}. The radio ephemeris and
1065: the 1024 bins light-curve
1066: (bottom panel) are obtained by the analysis of
1067: 20 ToAs observed by Parkes radio telescope (observation
1068: interval 54,220--54,562 MJD).}
1069: \end{figure}
1070:
1071: Pulsed counts and related Pearson statistics for the four pulsars are
1072: reported in Table \ref{counts2} (for the standard event extraction
1073: parameters as in Table \ref{counts}). The resulting fluxes (Pulsed
1074: Counts/Exposure) are consistent with those reported in the EGRET
1075: Catalogue \citep{hartman99}. We note that source-specific extraction
1076: parameters (event class, source position in the field of view, energy
1077: band etc.) which maximize reduced $\chi^2$ can significantly improve
1078: detection significance. For example, including only event class G in
1079: the timing analysis of Crab and Geminga halves the number of pulsed
1080: counts while doubling the $\chi^2$ values.
1081:
1082: Despite the very satisfactory matching of the pulsar spin
1083: parameters found in radio (or X-ray) and gamma-ray (supporting the
1084: clock stability and the correctness of the SSB transformations),
1085: possible systematic time shifts in the \emph{AGILE} event lists could
1086: be in principle affecting phasing and must be checked. For example, an
1087: hypothetical constant discrepancy of t$_{err}$ of the on-board time
1088: with respect to UTC would result in a phase shift
1089: $\Phi_{err}=(t_{err}~\rm{mod}~P)/P$, where $P$ is the pulsar period.
1090: The availability of radio observations bracketing the time span of the
1091: gamma-ray observations (or of X-ray observations very close to the
1092: gamma-ray observations for the case of Geminga) allowed us to also
1093: perform accurate phasing of multi-wavelength light-curves. In doing
1094: that, radio ephemeris reference epochs were set to the main peak of
1095: radio light-curves at phase $\Phi_{peak}=0$. In view of
1096: Equations (\ref{phasing2}) and (\ref{phasing3}), this is achieved by
1097: setting $\Phi_0=-\nu\sum_{k=1}^{N}b_k$ (typically $\Phi_0<10^{-2}$).
1098: We found that the phasing of the \emph{AGILE} light-curves of the four
1099: pulsars (radio/X-rays/gamma-ray peaks phase separations) is consistent
1100: with EGRET measurements
1101: \citep[][see \S~\ref{discuss} for details]{fierro98,tbb96,jackson05}
1102: implying no evidence of systematic errors in absolute timing with an
1103: upper limit $t_{err}<1$ ms.
1104:
1105:
1106: %The Crab SuperAGILE light-curve (Figure \ref{crab}) was
1107: %produced with the same folding method reported in
1108: %\S~\ref{timinganalysis}, yielding 63,700$\pm$8,700 pulsed counts
1109: %($\sim$3\% of the total flux).
1110: %It is shown in
1111: %Figure \ref{crab} and discussed in \S~\ref{section-tctr}.
1112: Comparison with the SuperAGILE light-curve peak (see \S~\ref{sec:radio})
1113: is also interesting (SuperAGILE on-board time processing is more complex
1114: than that of the GRID). The Crab SuperAGILE light-curve (Figure \ref{crab})
1115: was produced with the same folding method reported in \S~\ref{timinganalysis},
1116: yielding $63,700\pm8,700$ pulsed counts ($\sim$3\% of the total counts
1117: including background). Inspection of Figure \ref{crab} shows that the
1118: X-ray peaks are aligned with the $E>100$ MeV data within
1119: $\Delta\phi\sim400$ $\mu$s (a value obtained fitting the peaks with Gaussians)
1120: providing an additional test of the \emph{AGILE} phasing accuracy.
1121:
1122: The effective time resolution of \emph{AGILE} light-curves results
1123: from the combination of the different steps involved in the processing
1124: of gamma-ray photon arrival times. The on-board time tagging accuracy
1125: is a mere $\sim$1 $\mu$s, with negligible dead time. For comparison,
1126: the corresponding EGRET time tagging accuracy was $\sim$100 $\mu$s. The
1127: precise GPS space-time positioning of \emph{AGILE} spacecraft
1128: \citep{argan04} allows for the transformation from UTC to Solar System
1129: barycenter time-frame (TDB) with only a moderate loss ($\lesssim$10 $\mu$s)
1130: of the intrinsic instrumental time accuracy. The innovative folding
1131: technique described in \S~\ref{timinganalysis}, accounting for pulsar
1132: timing noise, also reduces smearing effects in the light-curves, fully
1133: exploiting all the information from contemporary radio observations.
1134: In summary, the effective time resolution of the current \emph{AGILE}
1135: pulsar light-curves (and then multi-wavelength phasing accuracy
1136: assuming $t_{err}=1$ $\mu$s) is mainly limited by the available count
1137: statistics and can be estimated by:
1138: \begin{equation}
1139: \Delta t={{P}\over{N}}={{\sigma^2(C_p+2B)}\over{C_p^2}}
1140: \end{equation}
1141: \noindent
1142: where $P$ is pulsar period, $N$ is the number of bins in the
1143: light-curve histogram, $\sigma$ is the signal-to-noise ratio, $C_p$
1144: are the pulsed counts and $B$ are the background counts. In order to
1145: keep the average signal-to-noise ratio of light-curve bins (during the
1146: on-pulse phase) at a reasonable level ($>$3\,$\sigma$), the resulting
1147: effective time resolution is constrained to 200--500 $\mu$s. At
1148: present, the best effective time resolution ($\sim$200 $\mu s$) is
1149: obtained for the 400-bin light-curve of Vela (G+L class selection)
1150: although a 100-bins light-curve (Figure \ref{vela}) is better suited
1151: to study pulse shapes and to search for possible weak features. The
1152: effective time resolution will obviously improve with exposure time
1153: $\Delta t \propto T_{exp}^{-1}$ and a resolution $\lsim$50 $\mu$s is
1154: expected after two years of \emph{AGILE} observations of Vela.
1155:
1156: \section{Discussion}
1157: \label{discuss}
1158:
1159: With about 10 years since the last gamma-ray observations
1160: of Crab, Vela, Geminga \citep{fierro98} and B\,1706--44
1161: \citep[][]{tbb96} by CGRO, the improved time resolution of
1162: \emph{AGILE} and the much longer observation campaigns in progress are
1163: now offering the possibility of both to search for new features in the
1164: shape of the light-curves of these gamma-ray pulsars and to
1165: investigate the possible occurrence of variations in the gamma-ray
1166: pulsed flux parameters.
1167:
1168: After nine months of observations in the frame of the Science
1169: Verification Phase (2007 July--November) and of the Scientific
1170: Pointing Program AO\,1 (pointings 1--7, 2007 December 1--2008 April 10),
1171: \emph{AGILE} reached an exposure (E$>$100 MeV) of the Vela region $\gtrsim$$10^9$
1172: cm$^2$ s ($\sim$10,000 pulsed counts from Vela), comparable to that of
1173: the nine-year life of EGRET (although \emph{AGILE} data have a higher
1174: residual particle background), and an even better exposure ($1.5\times10^9$ cm$^2$ s) in
1175: the core region of the Galactic Plane ($l=310\degr$--$340\degr$)
1176: corresponding to the Southern Hemisphere . In fact, \emph{AGILE}
1177: observed $\sim$2,400 pulsed counts from PSR B1706$-$44 up to date, a
1178: factor 1.5 better counts statistics than EGRET for this source.
1179: For Crab and Geminga, an exposure level comparable with that obtained
1180: by EGRET will be reached at the end of the AO\,1 pointing number 15 in the
1181: Anti-Center region (October, 2008).
1182:
1183: \subsection{The gamma-ray light-curves}
1184: \label{lc}
1185:
1186: The plots shown in Figures \ref{vela}, \ref{geminga}, \ref{crab}, and
1187: \ref{J1709} allow us to start assessing new features in
1188: gamma-ray pulsar light-curves. Narrower and better resolved main peaks
1189: are revealed, together with previously unknown secondary features, to be
1190: confirmed when more count statistics (and an improved particle background
1191: filtering) will be available.
1192:
1193: The Vela light-curves for different energy bands are shown in Figure
1194: \ref{vela}. A Gaussian fit to the Vela main peak (P1) at $E>100$ MeV
1195: provides a FWHM of $0.018\pm 0.002$ centered at $\phi=0.1339\pm0.0007$
1196: in phase consistent with the EGRET observations (Kanbach et al., 1994).
1197: The apex of the main peak is resolved by \emph{AGILE} with a width of
1198: $\sim$0.8 ms and its apparent trail ($\phi=0.13$--0.3) could be
1199: due to the occurrence of one or more secondary peaks. In
1200: fact, marginal evidence of a relatively narrow lower peak (P3) at
1201: $\phi\sim0.25$ is present in the $E<$100 MeV (3.3$\sigma$ fluctuation
1202: with respect to the average interpulse rate) and in the $E>$1 GeV (3.8$\sigma$)
1203: light-curves. P3 is located at the phase of the optical peak 1, and also at
1204: the phase of a bump predicted in a two-pole caustic model
1205: \citep{dyks04}, due to overlapping field lines from opposite poles
1206: near the light cylinder. In the outer gap model, this bump is the
1207: first peak in the light-curve, and also comes from very near the light
1208: cylinder. The peak at $\phi=0.5$--0.6 (P2) in the $E>100$
1209: MeV light-curve cannot be satisfactorily fit with a single Gaussian or
1210: a Lorentzian curve
1211: %(the value of the reduced $chi^2$ is
1212: ($\chi^2_r>3$), due the possible presence of a bump at $\phi\sim0.5$
1213: (P2a). A fit with two Gaussian provides $\phi=0.560\pm0.001$ (FWHM
1214: $0.031 \pm 0.003$) and $\phi=0.49\pm0.01$ for the phases of the major
1215: peak (P2b) and lower peak apex (P2a), respectively. The phase
1216: separation between the main gamma-ray peaks $\Delta
1217: \phi=0.426\pm0.002,$ as well as that between the gamma-ray and the
1218: radio peak, are unchanged since EGRET observations \citep{kanbach94,ramanamurthy95}.
1219: In the $E<100$ MeV band a secondary peak+valley structure (P4) appears at
1220: $\phi\sim0.9$ ($\sim$4$\sigma$ level). It is worth noting its
1221: symmetric position around the radio peak with respect to main peak
1222: (P1) and a possible correlation of P4 with features seen in the X-ray
1223: light-curves \citep{manzali07}; P4 also coincides with peak number 3 of the
1224: \emph{RossiXTE} light curve \citep{hsg02}.
1225:
1226: Geminga shows an $E>100$ MeV light-curve (Figure \ref{geminga}) with
1227: properties similar to those of Vela. Apart from the major peaks P2
1228: ($\phi=0.999\pm0.002$, FWHM $0.062\pm0.008$) and P1 (with apex
1229: $\phi=0.507\pm0.004$, FWHM $0.08 \pm 0.01$), secondary peaks are seen
1230: trailing P1 and leading P2: the main peak's P2 leading trail
1231: ($\phi=0.8$--1) could be possibly associated to unresolved multiple
1232: peaks while P1 displays a ``bump'' at
1233: $\phi\sim0.55$--0.6. P1 seems to be characterized by a double structure:
1234: % could be possibly present at its apex:
1235: a fit with a simple Gaussian yields $\chi^2_r\sim1.2$ to be
1236: compared with a double Gaussian model providing $\chi^2_r<1$. This
1237: feature seems still present even considering different observation
1238: block separately. The main-peak separation ($\Delta
1239: \phi=0.508\pm0.007$ at $E>100$ MeV) is greater than the value obtained for Vela and it
1240: decreases slightly with energy. The 2--10 keV X-ray light-curve shows
1241: a peak (P3) in correspondence with a possible excess in the hard
1242: gamma-ray band and a broad top-hat shaped feature partially
1243: overlapping in phase with P1.
1244:
1245: The Crab light-curve for $E>100$ MeV (Figure \ref{crab}) shows a
1246: previously unknown broad feature at $\phi=0.65-0.8$ (P3), in addition
1247: to the main peaks P1 and P2 ($\phi_{P1}=0.999\pm0.002,$ FWHM $0.054
1248: \pm 0.005$; $\phi_{P2}=0.382\pm0.008,$ FWHM $0.14 \pm 0.04$). The
1249: probability of P3 being a background fluctuation is of $\sim$$10^{-4}$
1250: (3.7$\sigma$). P2 could be possibly resolved in two sub-peaks in
1251: future longer observations. P3 is coincident with the feature HFC2
1252: that appears in the radio profile above 4 GHz \citep{moffett96}. From
1253: the polarization of this component, \citet{moffett99} suggest that
1254: this peak may come from a lower emission altitude, near the polar
1255: cap. P3 is actually at a phase that could plausibly come from low
1256: altitude cascades in the slot gap model
1257: \citep[][Figure~2]{muslimov04}, the pairs from which may be also
1258: causing the HFC2 radio component(s), while P1 and P2 comes from the
1259: high-altitude slot gap.
1260:
1261: According to the observations of \emph{SAS-2}, \emph{COS B} and
1262: \emph{CGRO}/EGRET, the ratio P2/P1 of the main peak intensities could
1263: present a variability pattern (possibly ascribed to
1264: the nutation of the neutron star) that can be fit with a sinusoid with a
1265: period of $\sim$13.5 years \citep[][]{kanbach90,ramanamurthy95}
1266: although this is not required by EGRET data alone \citep{tompkins97}. We
1267: observed a P2/P1 intensity ratio $0.66\pm0.10$ in good agreement with
1268: the value of $\sim$0.59 predicted for 54,350 MJD (for the energy range
1269: 50 MeV--3 GeV).
1270: %This interesting periodicity could be attributed to
1271: %the nutation of the neutron star.
1272: Unfortunately, our P2/P1 value is
1273: similar - within the errors - to the EGRET determination ($\sim$0.5):
1274: then an unambiguous assessment of the origin of this possible phenomenology
1275: will require measurements close to the epoch (56,150 MJD)
1276: corresponding to the predicted maximum or the intensity ratio P2/P1
1277: ($\sim$1.4). Variability should also be invoked to explain the
1278: possible detection of P3, which was never seen before in the EGRET
1279: database in spite of an overall exposure comparable to that reached by
1280: \emph{AGILE} so far. We note that the main peak intensity ratios
1281: computed for Vela ($\rm{P2/P1}= 0.91\pm0.07$) and Geminga ($\rm{P2/P1}=
1282: 0.8\pm0.1$) do not yield evidence of significant variations with
1283: respect to past observations \citep{ramanamurthy95}.
1284:
1285: The \emph{AGILE} light-curves of PSR~B\,1706$-$44 are shown in Figure
1286: \ref{J1709}. The broad-band light-curve ($E>30$ MeV) clearly shows two
1287: peaks ($\phi_{P1}=0.211\pm0.007,$ $\phi_{P2}=0.448\pm0.005$) bracketing
1288: considerable bridge emission (contributing to $>$50\% of the pulsed
1289: counts) while in the 0.1--1 GeV range the peaks
1290: %apparently fade in
1291: cannot be discerned from the bridge emission and the pulsar profile presents
1292: an unresolved broad
1293: ($\Delta \phi=0.3-0.4$) single peak.
1294: %feature of enhanced emission.
1295: PSR B\,1706--44
1296: is a young ($\sim2\times10^4$ yr) and energetic ($3.4\times10^{36}$
1297: erg s$^{-1}$) 102.5 ms pulsar \citep{jlm92} with emission properties similar
1298: to Vela \citep{becker02}. Double peaked PSR~B\,1706--44 is then in
1299: fact ``Vela-like'' not only energetically, but also with respect to
1300: the offset between the maxima of the high energy and the radio
1301: profiles, with neither of the two gamma-ray narrow pulses aligned to
1302: the radio peak.
1303:
1304: \emph{AGILE} allowed for a long monitoring of gamma-ray pulsar
1305: light-curves shapes.
1306: %We carefully looked for possible light-curve
1307: %variations by KS tests (Kolmogorof-Smirnoff test statistics [KS]; e.g. \citealt{kendall79}).
1308: We carefully looked for possible lightcurve variations by KS tests
1309: (two-dimensional Kolmogorof-Smirnoff test [KS]; \citealt{peacock83,fasano87}).
1310: For each pulsar, different gamma-ray light-curves (with 10, 20
1311: and 40 bins) were obtained grouping contiguous data set and requiring
1312: at least 30 counts bin$^{-1}$ (300--1000 counts for each light-curve). Each
1313: light-curve was compared by KS test with each other and with the
1314: average shape corresponding to the entire data set. No pulse shape
1315: variation was detected with a significance $>$3\,$\sigma$ on
1316: timescales ranging from 1 days (Vela) to few months (Crab,
1317: Geminga and PSR~B\,1706$-$44).
1318: %The limited statistics of the counts prevented to search for
1319: %variability in the light curves of PSR~B\,1706$-$44 with the database
1320: %available so far.
1321:
1322: \subsection{Implications for the emission models}
1323: \label{models}
1324:
1325: Pulsars derive their emitting power from rotational energy loss owing
1326: to the relativistic acceleration (up to Lorentz factor
1327: $\Gamma\sim10^5$--$10^7$) of charged particles by very high electric
1328: potentials induced by the rotating magnetic fields. The charge density
1329: that builds up in a neutron star magnetosphere is able to short out
1330: the electric field parallel to the magnetic field everywhere except a
1331: few locations of non-force-free ``gaps''. It is unclear whether these
1332: acceleration regions can form in the strong field near ($\lesssim$1
1333: stellar radii) the neutron star surface (polar cap/low-altitude slot
1334: gap model: \citet{daugherty96} and \citet{muslimov03}; high-altitude
1335: slot gap model: \citet{muslimov04} and \citet{harding08}), in the
1336: outer magnetosphere near the speed of light cylinder (outer gap model:
1337: \citet{cheng86,romani96,hirotani01,takata07}) or even beyond in the
1338: wind zone \citep{petri05}.
1339:
1340: In polar cap/low altitude slot gap models, gamma-rays result from
1341: magnetic pair cascades
1342: %(not photon-photon)
1343: induced by curvature or
1344: inverse-Compton photons. The spectrum is dominated by synchrotron
1345: radiation of pairs at lower energies ($\lesssim$1 GeV) and by
1346: curvature radiation at higher energies ($>$1 GeV). In the
1347: high-altitude slot gap, gamma-rays result from curvature radiation and
1348: synchrotron radiation of primary electrons (no cascade). In outer gap
1349: models, gamma-rays result from photon-photon pair cascades induced by
1350: curvature radiation. The spectrum is dominated by pair synchrotron
1351: radiation below 20 MeV, curvature radiation above 20 MeV and inverse
1352: Compton at 1--10 TeV.
1353:
1354: The size and spectrum of the emitting regions are then directly
1355: related to the intensity and location in the magnetosphere of the
1356: accelerating electric fields. Strong fields imply thin accelerator
1357: gaps, while weaker fields are associated to thick gaps: the
1358: acceleration zone grows bigger as the particle must accelerate over
1359: larger distance to radiate pair-production photons. On an
1360: observational point of view, it is in turn expected that acceleration
1361: gap sizes are related to the width of light-curve peaks. The highly
1362: relativistic particles emit photons at very small ($\sim$$1/\Gamma$)
1363: angles to the open magnetic field lines. The theoretical width of a
1364: light-curve peak associated to an infinitely small gap would be then
1365: $\Delta t = P/(2 \pi \Gamma)$, a value typically smaller than 1
1366: $\mu$s. Therefore, the width of the apex of the peak can be related to
1367: the core gap size. For example, the $\Delta t\sim800$ $\mu$s width of
1368: the Vela pulsar peak (P1) resolved by \emph{AGILE} implies a projected
1369: gap core width of $\sim$1 km for gap height of $\sim$1
1370: stellar radii. Broader peaks as P2 and P3 in the Crab light-curve
1371: involves instead magnetospheric region tens of kilometers long. The
1372: relation of the width of the peaks to the acceleration gap size works
1373: for the low altitude emission models but is complicated by
1374: relativistic effects (aberration, retardation) in the high-altitude
1375: emission models, where the peaks are formed by caustics
1376: \citep{dyks03}. The peak widths will still depend on the width of the
1377: acceleration (outer or slot) gap, but one must perform a comparison
1378: with detailed models to constrain the physical size of the gap.
1379:
1380: The likely presence of multiple contiguous peaks as P2a-b in the Vela
1381: light-curve, P1 in Geminga and possibly within the wide P2 broad peak
1382: in Crab could be related to short-term oscillation of gap
1383: locations. Light-curve variations on timescales $\gtrsim$1 days are
1384: anyway excluded by the KS tests described above. Alternatively, the
1385: apparent superimposition of different gaps placed at different heights
1386: in the magnetosphere could be another plausible explanation for
1387: multiple-contiguous peaks, while the hypothesis of jumps in the
1388: acceleration electric fields strength would lack of evident physical
1389: justification. In general, the presence of multiple (contiguous or
1390: not) structures in the light-curves is difficult to fit in a scenario
1391: alternatively involving polar cap, outer gap or wind zone models
1392: exclusively. The particle acceleration may well be simultaneously
1393: occurring in all the regions predicted by these models.
1394: %Anyway, for example
1395: Indeed, a three peak light-curve can still be explained invoking
1396: polar cap gaps alone provided that both polar caps can cross the line-of-sight.
1397:
1398: Wherever in the magnetosphere a gap can form, a multi-frequency
1399: emission may occur along an hollow cone (due to magnetosphere
1400: symmetries) or any other suitable surface: light-curve peaks are
1401: generated when the viewing angle from any given location on this
1402: surface to the observer also crosses the gap. The pulse spectrum slope
1403: depends on the accelerating field as well as on the magnetic fields,
1404: which strongly affect the synchrotron emission efficiency and the pair
1405: attenuation along the cascade path: the stronger the magnetic field
1406: is, the steeper the spectrum is. An outer gap can in principle
1407: generate pulse spectra extending up to tens of GeV \citep{zc00}, while
1408: gaps close to surface high magnetic fields can produce steep spectra
1409: \citep[photon index $\lesssim$3;][]{ramanamurthy96} and cut-off in the
1410: tens of MeV range \citep[PSR\,1509--58,][]{harding97}. The absence of
1411: corresponding gamma-ray pulse in phase with radio main peaks in Vela
1412: and PSR~B\,1706--44 could be then ascribed to gap-dependent spectral
1413: slope and not only to beam angle and viewing geometry differences.
1414:
1415: The multiplicity and variety of features seen in \emph{AGILE}
1416: light-curves can pave the way to a parameterized standard
1417: model (e.g. with adjustable accelerating electric fields strength,
1418: and location in the magnetosphere) for pulsars gaps and their
1419: corresponding observed high-energy
1420: pulses. In this perspective, the \emph{AGILE} light-curves time
1421: resolution, currently limited only by the (continuously increasing) source
1422: counts statistics, will eventually yield a pulsar gaps
1423: map by coupling timing analysis and phase resolved spectral analysis.
1424:
1425: \subsection{The Vela glitch of August 2007}
1426: \label{velaglitch}
1427:
1428: During early \emph{AGILE} observations, Vela experienced a weak
1429: glitch clearly detected in radio as a discontinuity in the pulsar's
1430: spin parameters.
1431:
1432: Glitches are small ($\Delta\nu/\nu\sim10^{-9}$--$10^{-6}$) and sudden
1433: ($\lesssim$1 day) discontinuous increases in the pulsar frequency,
1434: often followed by a recovery (1--100 days) to the pre-glitch
1435: frequency. About $\sim$6\% of pulsars are known to have shown
1436: glitches\footnote{See e.g. http://www.atnf.csiro.au/research/pulsar/psrcat/.}),
1437: with a higher incidence of events in younger pulsars. This phenomenon
1438: is potentially a very promising tool for probing the physics of the
1439: neutron star interiors \citep{lyne00}. Although no general consensus
1440: has been reached to date about the origin of the glitches, many models
1441: are based on the exchange of angular momentum between the superfluid
1442: neutron star core and its normal, solid crust \citep{ruderman76,ruderman91,alpar84,aaps84}.
1443: This angular momentum transfer may excite starquake waves,
1444: propagating toward the neutron star surface. Since the magnetic field
1445: frozen in the crust is ``shaken'', the resulting oscillating
1446: electromagnetic potential could generate strong electric fields
1447: parallel to the magnetic field, which in turn would accelerate
1448: particles to relativistic energies, possibly emitting a burst of
1449: high-energy radiation.
1450:
1451: Since the first observation of a pulsar glitch in 1969
1452: \citep{radhakrishnan69}, Vela has shown $\sim$10 major glitches.
1453: Due to its large field of view, the quest for possible gamma-ray
1454: bursting behavior due to a glitch is then an effective opportunity for
1455: \emph{AGILE}. Despite the fact that the August 2007 glitch is a weak one
1456: ($\Delta\nu/\nu\sim10^{-9}$), it is worthwhile to
1457: %investigate about possible sign of it
1458: search for a signal in the \emph{AGILE} data.
1459:
1460: The characteristic energy of a pulsar glitch can be roughly estimated from the
1461: associated pulsar frequency jump $E_{\rm glitch}=\Delta E_{\rm rot}=4\pi^2
1462: I\nu\Delta\nu$, where $I$ ($\sim$10$^{45}$ g cm$^2$) is the neutron
1463: star momentum of inertia. The corresponding expected gamma-ray counts
1464: would be:
1465: \begin{equation}
1466: C_{\gamma}^{\rm glitch}=\eta{{ E_{\rm glitch}A_{\rm eff}}\over{4\pi d^2
1467: E_{\gamma}}} \simeq 10^{11} \eta {\Delta\nu\over\nu}
1468: \label{eqn:cglitch}
1469: \end{equation}
1470: \noindent
1471: where $\eta=[0,1]$ is the unknown conversion efficiency of the glitch
1472: energy to gamma-ray emission, $d$ ($\sim$0.3 kpc) is the pulsar
1473: distance, $E_{\gamma}$ ($\sim$300 MeV) is the average gamma-ray photon
1474: energy assuming a spectral photon index $\Gamma=-2$, and A$_{\rm eff}$ is
1475: the \emph{AGILE} effective area.
1476:
1477: Even in the virtual limit assumption that the entire glitch energy could
1478: be driven into gamma-ray emission, a weak glitch with a frequency
1479: shift of ${\Delta\nu/\nu}= 1.3\times10^{-9},$ as that observed in
1480: August 2007, cannot produce a strong signal
1481: ($C_{\gamma}^{\rm glitch}<100$--200 counts), if the core fluence is spread
1482: in $\sim$1 day. In fact, no excess on daily timescales was detected,
1483: although for much shorter timescales of 3--6 minutes a
1484: $>$5\,$\sigma$ excess ($\sim$15 counts) in the photon counts is
1485: seen at $\sim$54,312.693 MJD (Figure \ref{glitch}).
1486: \begin{figure} % figure 9
1487: \resizebox{\hsize}{!}{\includegraphics[angle=00,width=15cm]{f9.eps}}
1488: \caption{\label{glitch} Unfolded Vela pulsar light-curve (4.5 min binsize, $E>50$ MeV, G-class events).
1489: Dashed lines bracket glitch epoch uncertainty range ($54,313\pm3$ MJD).
1490: A $>$5$\sigma$ count excess at $\sim$54,312.693 MJD could be associated to
1491: gamma-ray bursting emission from the glitch.
1492: %...figure probably not to be included in the final draft...
1493: }
1494: \end{figure}
1495:
1496: On the other hand, stronger Vela glitches, as that of 1988 Christmas
1497: \citep{mcculloch90} with a frequency shift of ${\Delta\nu/\nu} =
1498: 2\times10^{-6}$, could in principle produce more significant transient
1499: gamma-ray emission. Typical count-rate from Vela is $\sim$100--200
1500: counts day$^{-1}$, then a fluence of $\gtrsim$1000 counts in $\sim$1-day
1501: or less from the hypothetical gamma-ray glitch burst should be easily
1502: detectable. According to Equation (\ref{eqn:cglitch}), such a flux could
1503: arise from a glitch with ${\Delta\nu/\nu}\sim 10^{-7}$ (typical Vela
1504: glitch size), converting a relatively small fraction ($\eta\sim0.1$)
1505: of its energy in gamma-rays. The chance occurrence of a strong Vela
1506: glitch in the \emph{AGILE} field of view over three years of mission
1507: operations is of $\sim$20\%.
1508:
1509: \section{Conclusions}
1510: \label{conclusions}
1511:
1512: \emph{AGILE} collected $\sim$15,000 pulsed counts from known gamma-ray pulsars
1513: during its first 9 months of operations. The \emph{AGILE} PDHU clock
1514: stability, coupled with the exploitation of pulsar timing noise
1515: information,
1516: % in the folding process
1517: allows for pulsar period fitting
1518: with discrepancies with respect to radio measurements at the level of
1519: the period search resolution ($\sim$$10^{-12}$ s) over the long
1520: gamma-ray data span ($>$6 months). Thanks to \emph{AGILE} GPS-based high
1521: time tagging accuracy ($\sim$1 $\mu$s), the effective time resolution
1522: of \emph{AGILE} light-curves is limited only by the count statistics
1523: at current level of exposure ($\sim$$1.2\times10^9$ cm$^2$ s for the Vela
1524: region). The best effective time resolution obtained for Vela
1525: observations is of $\sim$200 $\mu$s for a signal-to-noise
1526: $>$3\,$\sigma$ in the on-pulse light-curve bins. An improved effective
1527: time resolution $\lesssim$ 50$\mu$s is expected after three years of
1528: \emph{AGILE} observations.
1529:
1530: \emph{AGILE} multi-wavelength phasing of the four gamma-ray pulsars is
1531: consistent with the results obtained by \emph{EGRET}, although
1532: the high resolution \emph{AGILE} light-curves show narrower and
1533: structured peaks and new interesting features. In particular, a third
1534: peak is possibly detected at $\sim$3.7$\sigma$ level in the Crab light-curve
1535: and several interesting features seem present in the Vela light-curves.
1536: %In next months, the improved gamma-rays count statistics (with the ongoing
1537: %simultaneous radio monitoring campaigns) will discriminate
1538: %new features from background fluctuations with higher confidence.
1539:
1540: In any case, the highly structured light-curves hint at a complex
1541: scenario for the sites of particle acceleration in the pulsar
1542: magnetospheres, implying different electric gaps with physical
1543: properties probably mostly related to their height above the neutron
1544: star surface. Alternatively, slight spatial oscillations of the gap
1545: locations on timescales $\lsim$1 day could be invoked to explain the
1546: multiple contiguous peaks seen in the light-curves.
1547:
1548: The foreseen balance of count statistics at the end of AO\,1 observing
1549: program will allow for phase-resolved spectral analysis of the
1550: light-curves and, correspondingly, a spatial mapping of the
1551: magnetospheric gaps (their altitude above the neutron star surface
1552: being possibly related to the shape of their spectra).
1553:
1554: We finally note that the timing calibration and tests presented in
1555: this paper pave the way to an effective search for new gamma-ray
1556: pulsars with \emph{AGILE}. The negligible discrepancies among the
1557: radio and the gamma-ray pulsar spin parameters seen in the known
1558: gamma-ray pulsars imply that the direct folding of the \emph{AGILE}
1559: data on new pulsar candidates is a safe and efficient procedure, when
1560: the folding parameters are obtained from radio/X-rays ephemeris having
1561: suitable epoch range of validity. In fact, gamma-ray period search
1562: trials around radio/X-rays solutions will be unnecessary when
1563: simultaneous ephemeris are available, strongly improving the detection
1564: significance for faint sources. According to the predicted gamma-ray
1565: pulsar luminosities ($L_\gamma \propto \dot{E}^{1/2}/d^2$,
1566: e.g. \citealt{pellizzoni04agile}), the detection with \emph{AGILE} of
1567: top-ranked Vela-like pulsars is then expected as soon as exposure
1568: levels $\gtrsim$$10^9$ cm$^2$ s ($E>100$ MeV) will be attained for
1569: clean G-class events.
1570:
1571: \acknowledgments
1572: We acknowledge D.~A.~Smith, D.~J.~Thompson and G.~Tosti of GLAST Team for
1573: useful discussions on multi-wavelength observations of pulsars and for
1574: their comments on the paper draft. JPH was supported by NASA
1575: \emph{XMM-Newton} grants NNX06AH58G and NNX07AU65G. AP and MB received
1576: financial support from the Italian Minister of Research (MIUR) under national
1577: program PRIN-MIUR 2005. The Parkes radio telescope is part of the
1578: Australia Telescope which is funded by the Commonwealth of Australia
1579: for operation as a National Facility managed by CSIRO.
1580: \emph{XMM-Newton} is an ESA science mission with instruments and
1581: contributions directly funded by ESA Member States and NASA.
1582:
1583: \bibliographystyle{apj}
1584: \bibliography{biblio}
1585:
1586:
1587:
1588: %%\begin{deluxetable}{lc} % table 2
1589: %%% \tabletypesize{\scriptsize}
1590: %% \tablecolumns{1}
1591: %%\tablewidth{0pc}
1592: %% \tablecaption{\label{tab:geminga}Geminga X-ray ephemeris obtainded
1593: %% from the timing analysis of \emph{XMM-Newton} observations in the epoch range
1594: %%52,368--54,170 MJD (see \S~\ref{sec:radio}).}
1595: %%\tablehead{\colhead{Parameter} & \colhead{Value}}
1596: %%\startdata
1597: %%%PSRJ & 0633+1746 \\
1598: %%RA (J2000) & $06^{\rm h}33^{\rm m}54\fs1$ \\
1599: %%DECJ (J2000) & $17\degr46\arcmin12\farcs9$ \\
1600: %%F0 (s$^{-1}$) & 4.2175868009(1) \\
1601: %%F1 (s$^{-2}$) & $-1.95212(1)\times10^{-13}$ \\
1602: %%PEPOCH (MJD) & 53629.99999972957 \\
1603: %%PEPOCH (MJD) & 53,630.00000027844 \\
1604: %%POSEPOCH (MJD) & 49,793 \\
1605: %%PMRA (mas yr$^{-1}$) & 138(4) \\
1606: %%PMDEC (mas yr$^{-1}$) & 97(4) \\
1607: %%START (MJD) & 52,368.71 \\
1608: %%FINISH (MJD) & 54,170.76 \enddata
1609: %%%\tablenotetext{a}{The data are available in the \emph{XMM-Newton} Science Archive, at http://xmm.%%esac.esa.int/.}
1610: %%\end{deluxetable}
1611: \end{document}
1612: