0801.2589/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: \newcommand{\msun}{\mbox{M$_{\odot}$}}
3: \newcommand{\rsun}{\mbox{R$_{\odot}$}}
4: \newcommand{\aopx}{\mbox{$\Delta_{\pi \rm M}$}}
5: \newcommand{\shap}{\mbox{$\Delta_{\rm S}$}}
6: \newcommand{\psr}{\mbox{PSR J0437$-$4715}}
7: 
8: \newcommand{\pb}{\mbox{$P_{\rm b}$}}
9: \newcommand{\pbdot}{\mbox{$\dot{\pb}$}}
10: 
11: \newcommand{\wvect[1]}{{\bsf #1}}
12: \newcommand{\bvect[1]}{{\bsf #1}$_{\bm 0}$}
13: 
14: \shorttitle{High-Precision Timing of \psr}
15: \shortauthors{Verbiest et al.}
16: 
17: \begin{document}
18: 
19: \title{Precision timing of \psr: an accurate pulsar distance,
20:   a high pulsar mass and a limit on the variation of Newton's
21:   gravitational constant}
22: 
23: \author{J. P. W. Verbiest,\altaffilmark{1,2}
24:   M. Bailes,\altaffilmark{1} W. van Straten,\altaffilmark{1}
25:   G. B. Hobbs,\altaffilmark{2} R. T. Edwards,\altaffilmark{2}
26:   R. N. Manchester,\altaffilmark{2} N. D. R. Bhat,\altaffilmark{1}
27:   J. M. Sarkissian,\altaffilmark{2} B. A. Jacoby\altaffilmark{3} and
28:   S. R. Kulkarni\altaffilmark{4}} 
29: 
30: \altaffiltext{1}{Centre for Astrophysics and Supercomputing, Swinburne
31:   Unversity of Technology, P.O. Box 218 Hawthorn, VIC 3122, Australia}
32:   \altaffiltext{2}{Australia Telescope National Facility -- CSIRO,
33:   P.O. Box 76, Epping, NSW 1710, Australia} \altaffiltext{3}{Naval
34:   Research Laboratory, 4555 Overlook Avenue, SW, Washington, DC 20375}
35:   \altaffiltext{4}{Robinson Laboratory, California Institute of
36:   Technology, Mail Code 105-24, Caltech Optical Observatories,
37:   Pasadena, CA 91125}
38: 
39: \begin{abstract}
40: Analysis of ten years of high-precision timing data on the millisecond
41: pulsar \psr\ has resulted in a model-independent kinematic distance
42: based on an apparent orbital period derivative, \pbdot, determined at
43: the $1.5\%$ level of precision ($D_{\rm k} = 157.0\pm 2.4$\,pc),
44: making it one of the most accurate stellar distance estimates
45: published to date.  The discrepancy between this measurement and a
46: previously published parallax distance estimate is attributed to
47: errors in the DE200 Solar System ephemerides. The precise measurement
48: of \pbdot\ allows a limit on the variation of Newton's gravitational
49: constant, $|\dot{G}/G| \leq 23 \times 10^{-12}$\,yr$^{-1}$.  We also
50: constrain any anomalous acceleration along the line of sight to the
51: pulsar to $|a_{\odot}/c| \leq 1.5\times 10^{-18}$\,s$^{-1}$ at $95\%$
52: confidence, and derive a pulsar mass, $m_{\rm psr} = 1.76 \pm
53: 0.20\,M_{\odot}$, one of the highest estimates so far obtained.
54: \end{abstract}
55: 
56: \keywords{stars: distances --- pulsars: individual (PSR J0437$-$4715) ---  stars:
57:   neutron}
58: 
59: \section{Introduction}\label{Introduction}
60: 
61: In \citeyear{jlh+93}, \citeauthor{jlh+93} reported the discovery of
62: \psr, the nearest and brightest millisecond pulsar known. Within a
63: year, the white dwarf companion and pulsar wind bow shock were
64: observed \citep{bbb93} and pulsed X-rays were detected
65: \citep{bt93a}. The proper motion and an initial estimate of the
66: parallax were later presented along with evidence for secular change
67: in the inclination angle of the orbit due to proper motion
68: \citep{sbm+97}. Using high time resolution instrumentation, the
69: three-dimensional orbital geometry of the binary system was
70: determined, enabling a new test of general relativity
71: \citep[GR;][]{vbb+01}. Most recently, multi-frequency observations
72: were used to compute the dispersion measure structure function
73: \citep{yhc+07}, quantifying the turbulent character of the
74: interstellar medium towards this pulsar.
75: 
76: The high proper motion and proximity of \psr\ led to the prediction
77: \citep{bb96} that a distance measurement independent of parallax would
78: be available within a decade, when the orbital period derivative
79: (\pbdot) would be determined to high accuracy. Even if the predicted
80: precision of $\sim$1\% would not be achieved, such a measurement would
81: be significant given the strong dependence of most methods of distance
82: determination on relatively poorly constrained models and the
83: typically large errors on parallax measurements. Even for nearby
84: stars, both the Hubble Space Telescope and the Hipparcos satellite
85: give typical distance errors of $3\%$ \citep{val07} and so far only
86: two distances beyond $100$\,pc have been determined at $\sim1\%$
87: uncertainty \citep{tlmr07}. This kinematic distance is one of the few
88: model-independent methods that does not rely upon the motion of the
89: Earth around the Sun.
90: 
91: As demonstrated by \citet{dt91}, \pbdot\ can also be used to constrain
92: the variation of Newton's gravitational constant. The best such limit
93: from pulsar timing to date \citep[$|\dot{G}/G| = (4\pm5)\times
94: 10^{-12}$\,yr$^{-1}$ from PSR B1913+16]{tay93} is compromised due to
95: the poorly constrained equation of state for the neutron star
96: companion \citep{nor90}. The slightly weaker but more reliable limit
97: of $|\dot{G}/G| = (-9\pm 18) \times 10^{-12}$\,yr$^{-1}$ \citep[from
98: PSR B1855+09, which has a white dwarf companion]{ktr94} should
99: therefore be considered instead. A more stringent limit can be
100: obtained from the \citet{sns+05} timing of PSR J1713+0747:
101: $|\dot{G}/G| = (2.5\pm 3.3) \times 10^{-12}$\,$yr^{-1}$ (at 95\%
102: certainty). This limit is, however, based upon the formal errors of
103: \pbdot, $P_{\rm b}$ and parallax, which are easily underestimated by
104: standard methodologies, as we shall demonstrate later. Because of this
105: we believe the \citet{sns+05} limit is probably underestimated, but
106: still of relevance.  However, none of these limits are as strong as
107: that put by lunar laser ranging \citep[LLR;][]{wtb04}: $\dot{G}/G =
108: (4\pm 9) \times 10^{-13}$\,yr$^{-1}$. Besides limiting alternative
109: theories of gravity, bounds on $\dot{G}$ can also be used to constrain
110: variations of the Astronomical Unit ($AU$). Current planetary radar
111: experiments \citep{kb04} have measured a significant linear increase
112: of $d AU/dt =0.15\pm 0.04$\,m yr$^{-1}$, which may imply $\dot{G}/G =
113: (-10 \pm 3)\times 10^{-13}$\,yr$^{-1}$, just beyond the sensitivity of
114: the limits listed above.
115: 
116: As mentioned before, the equation of state for dense neutron star
117: matter is very poorly constrained.  Pulsar mass determinations can
118: probe the range of permissible pulsar masses and thereby limit
119: possible equations of state \citep{lp07}.  Presently, only the pulsars
120: NGC 6440B, Terzan 5\,I and Terzan 5\,J have predicted masses higher
121: than the typical value of $1.4\,M_{\odot}$\citep{frb+07,rhs+05};
122: however, as discussed in more detail in Section \ref{Mass}, such
123: predictions do not represent objective mass estimates.
124: 
125: The remainder of this paper is structured as follows: Section
126: \ref{Obs} describes the observations, data analysis and general timing
127: solution for \psr. Section \ref{Dist} describes how the measurement of
128: \pbdot\ leads to a new and highly precise distance. In Section
129: \ref{GandA}, this measurement is combined with the parallax distance
130: to derive limits on $\dot{G}$ and the Solar System
131: acceleration. Section \ref{Mass} presents the newly revised pulsar
132: mass and our conclusions are summarised in Section \ref{Conc}.
133: 
134: \section{Observations and Data Reduction}\label{Obs}
135: 
136: Observations of \psr\ were made over a time span of ten years (see
137: Figure \ref{Fig::Res}), using the Parkes 64-m radio telescope, two
138: 20\,cm receiving systems (the central beam of the Parkes multi-beam
139: receiver \citep{swb96} and the H-OH receiver) and four generations of
140: digital instrumentation (see Table \ref{tbl::Obs}): the Fast Pulsar
141: Timing Machine (FPTM), the S2 VLBI recorder, and the
142: Caltech-Parkes-Swinburne Recorders (CPSR and CPSR2).  The FPTM is an
143: autocorrelation spectrometer, whereas the three other instruments are
144: baseband data recording and processing systems that employ
145: phase-coherent dispersion removal.
146: 
147: \subsection{Arrival Time Estimation}
148: For the FPTM, S2, and CPSR backends, the uncalibrated polarization
149: data were combined to form the polarimetric invariant interval
150: \citep{bri00} and each observation was integrated in time and
151: frequency before pulse arrival times were calculated through standard
152: cross-correlation with an instrument-dependent template profile.  For
153: the CPSR2 data, the technique described by \citet{van04a} was used to
154: calibrate 5 days of intensive \psr\ observations made on 2003 July 19
155: to 21, 2003 August 29, and 2005 July 24.  The calibrated data were
156: integrated to form a polarimetric template profile with an integration
157: length of approximately 40 hours and frequency resolution of 500 kHz.
158: This template profile and Matrix Template Matching
159: \citep[MTM,][]{van06} were used to calibrate the three years of CPSR2
160: data.  An independent MTM fit was performed on each five-minute
161: integration, producing a unique solution in each frequency channel, as
162: shown in Figure 2 of \citet{van06}. The calibrated data were then
163: integrated in frequency to produce a single full-polarization profile
164: at each epoch.  MTM was then used to derive time-of-arrival (TOA)
165: estimates from each calibrated, five-minute integration.  The
166: application of MTM during the calibration and timing stages reduced
167: the weighted rms of the CPSR2 post-fit timing residuals by a factor of
168: two.  All the data reduction described above was performed using the
169: \textsc{psrchive} software package \citep{hvm04}.
170: 
171: \subsection{Timing Analysis}
172: Most data were recorded at a wavelength of 20\,cm; however, in the
173: final three years, simultaneous observations at 10 and 50\,cm were
174: used to measure temporal variations of the interstellar dispersion
175: delay \citep[corrections for these variations were implemented in a
176: way similar to that of][]{yhc+07}. A linear trend of these delays was
177: also obtained for the year of FPTM data, using data at slightly
178: different frequencies close to $1400$\,MHz.
179: 
180: The arrival times were analyzed using the \textsc{Tempo2} pulsar
181: timing software package \citep{hem06,ehm06} and consistency with the
182: earlier program, \textsc{Tempo}, was verified.  The timing model (see
183: Table \ref{Model}) is based on the relativistic binary model first
184: derived by \citet{dd86}, and expanded to contain the geometric orbital
185: terms described by \citet{kop95,kop96}. The model is optimised through
186: a standard weighted least-squares fit in which all parameters are
187: allowed to vary, including the unknown time delays between data from
188: different instruments, but excluding the mean value of dispersion
189: measure, which is determined from the simultaneous CPSR2, 64\,MHz-wide
190: bands centred at 1341 and 1405\,MHz.
191: 
192: A major difference between our implementation of solutions for the
193: orbital angles $\Omega$ and $i$ and previous efforts
194: \citep{vbb+01,hbo06} is that they were implemented as part of the
195: standard fitting routine.  This ensures any covariances between these
196: and other parameters (most importantly the periastron advance and
197: companion mass, see Table \ref{Model} and Section \ref{Mass}) are
198: properly accounted for, thereby yielding a more reliable measurement
199: error. The previous works 
200: mentioned above derived these effects from an independent mapping of
201: $\chi^2$ space, leaving the errors of other parameters unaffected.
202: 
203: As can be seen from Figure \ref{Fig::Res}, there are significant
204: low-frequency structures present in the timing residual data. Since
205: the standard least-squares fitting routine used in \textsc{Tempo2}
206: does not account for the effect of such correlations on parameter
207: estimation, we performed a Monte-Carlo simulation where data sets with
208: a post-fit power spectrum statistically consistent with that of the
209: \psr\ data were used to determine the parameter estimations
210: uncertainties in the presence of realistic low frequency noise. These
211: errors, as well as the factors by which the original errors were
212: underestimated, are shown in Table \ref{Model}. As an example, the
213: distribution of derived pulsar masses from the Monte-Carlo simulation
214: is given in Figure \ref{PSRMass}. Because of the dispersion measure
215: corrections implemented in the final three years of data, one can
216: expect the spectrum of these most precise data points to contain less
217: low-frequency noise than the ten year data set as a whole. We
218: therefore expect the errors resulting from this analysis to be
219: slightly overestimated. Ongoing research into extending the fitting
220: routine with reliable whitening
221: schemes to avoid the spectral leakage and hence improve the
222: reliability of the measured parameters, is expected to reduce these
223: errors by factors of around two. All errors given in this paper are
224: those resulting from the Monte-Carlo simulations, unless otherwise
225: stated. The simulations also showed that any biases resulting from the
226: red noise are statistically negligible for the reported parameters. (A
227: full description of this Monte-Carlo technique and the whitening
228: schemes mentioned will be detailed in a future publication.)
229: 
230: \subsection{Solar System Ephemerides}\label{SSE}
231: 
232: Pulsar timing results are dependent on accurate ephemerides for the
233: Solar System bodies. The results presented in this paper were obtained
234: using the DE405 model \citep{sta04b} and, for comparison, selected
235: parameters obtained with the earlier DE200 model are shown in Table
236: \ref{DEModels}. The greatly reduced $\chi^2$ indicates that the newer
237: Solar System ephemerides are superior to the earlier DE200,
238: reinforcing similar conclusions of other authors
239: \citep{sns+05,hbo06}. We notice the parallax value changes by more
240: than $10\,\sigma$, and that the different derived values are closely
241: correlated with the ephemeris used. Although the effect is not as
242: dramatic as it appears because of the under-estimation of the
243: \textsc{Tempo2} errors, the fact that the DE405 results agree much
244: better with the more accurate kinematic distance (discussed in the
245: next Section), strongly suggests that the differences are due to the
246: ephemeris used and confirms that the DE405 ephemeris is superior.
247: Finally, we note that the DE405 measurement of $\dot{\omega}$ ($0.016
248: \pm 0.008\,^{\circ}{\rm yr}^{-1}$) is consistent with the GR
249: prediction for this system ($0.0172 \pm 0.0009\,^{\circ}{\rm
250: yr}^{-1}$).
251: 
252: \section{Kinematic Distance}\label{Dist}
253: 
254: As shown in Figure \ref{Fig::Pbdot}, the long-term timing history
255: enables precise measurement of the orbital period derivative, \pbdot $
256: = (3.73\pm0.06)\times 10^{-12}$.  This observed value represents a
257: combination of phenomena that are intrinsic to the binary system and
258: dynamical effects that result in both real and apparent accelerations
259: of the binary system along the line of sight \citep{bb96}; i.e.
260: \begin{equation}\label{PbdotEq::Basis}
261: \dot{P}_{\rm b}^{\rm obs} =
262: \dot{P}_{\rm b}^{\rm int} +
263: \dot{P}_{\rm b}^{\rm Gal} +
264: \dot{P}_{\rm b}^{\rm kin}
265: \end{equation}
266: where ``obs'' and ``int'' refer to the observed and intrinsic values;
267: ``Gal'' and ``kin'' are the Galactic and kinematic contributions.
268: 
269: Intrinsic orbital decay is a result of energy loss typically due to
270: effects such as atmospheric drag and tidal dissipation; however, in a
271: neutron star--white dwarf binary system like \psr, energy loss is
272: dominated by quadru\-po\-lar gravitational wave emission. For this
273: system, GR predicts \citep{tw82} $\dot P_{\rm b}^{\rm
274: GR}=-4.2\times10^{-16}$, two orders of magnitude smaller than the
275: uncertainty in the measured value of \pbdot.
276: 
277: Galactic contributions to the observed orbital period derivative
278: include differential rotation and gravitational acceleration
279: \citep{dt91}.  The differential rotation in the plane of the Galaxy is
280: estimated from the Galactic longitude of the pulsar and the
281: Galactocentric distance and circular velocity of the Sun. Acceleration
282: in the Galactic gravitational potential varies as a function of height
283: above the Galactic plane \citep{hf04b}, which may be estimated using
284: the parallax distance and the Galactic latitude of the
285: pulsar. Combining these terms gives $\dot{P}_{\rm b}^{\rm Gal} = (-1.8
286: -0.5) \times 10^{-14} = -2.3\times10^{-14}$, which is of the same
287: order as the current measurement error.
288: 
289: Given the negligible intrinsic contribution, Equation
290: \ref{PbdotEq::Basis} can be simplified and rewritten in terms of the
291: dominant kinematic contribution known as the Shklovskii effect
292: \citep{shk70}, an apparent acceleration resulting from the non-linear
293: increase in radial distance as the pulsar moves across the plane
294: perpendicular to the line of sight; quantified by the proper motion,
295: $\mu$, and distance $D$ from the Earth:
296: %
297: \begin{equation}\label{PbdotEq::Shk}
298: \dot{P}_{\rm b}^{\rm obs} -
299: \dot{P}_{\rm b}^{\rm Gal} \simeq
300: \dot{P}_{\rm b}^{\rm kin} = \frac{\mu^2D}{c} P_{\rm b},
301: \end{equation}
302: where $c$ is the vacuum speed of light. Using the measured values of
303: $\mu$, \pb, and \pbdot, Equation \ref{PbdotEq::Shk} is used to derive
304: the kinematic distance \citep{bb96}: $D_k = 157.0 \pm 2.4$ pc.  This
305: distance is consistent with the one derived from parallax ($D_{\rm \pi}
306: = 150\pm12
307: $\, pc -- see also Figure \ref{Fig::Px}) and is, with a relative error
308: of $1.5\%$, comparable in precision to the best parallax measurements
309: from VLBI \citep{tlmr07} and better than typical relative errors
310: provided by the Hipparcos and Hubble space telescopes \citep{val07}.
311: 
312: Given the dependence of parallax distances on ephemerides, as
313: described in Section \ref{SSE}, it is interesting to note the
314: robustness of $D_{\rm k}$.  Also, Table \ref{Model} shows that the
315: presence of red noise corrupts the parallax error by a factor of 7.9,
316: whereas \pbdot\ is only affected by a factor of 2.5.  These facts
317: clearly indicate the higher reliability of $D_{\rm k}$ as compared to
318: $D_{\rm \pi}$.
319: 
320: \section{Limits on \pbdot\ Anomalies: $\dot{G}$ and the
321:   Acceleration of the Solar System}\label{GandA}
322: 
323: Any anomalous orbital period derivative can be constrained by
324: substituting the parallax distance into Equation 2, yielding
325: \begin{eqnarray}
326:   \Big(\frac{\dot{P}_{\rm b}}{P_{\rm b}}\Big)^{\rm excess} & = &
327:   \big(\dot{P}_{\rm b}^{\rm obs} - \dot{P}_{\rm b}^{\rm Gal} -
328:   \dot{P}_{\rm b}^{\rm kin}\big)/P_{\rm b}\nonumber\\ 
329:   & = & (3.2 \pm 5.7) \times 10^{-19}\, {\rm s}^{-1}.
330: \end{eqnarray}
331: in which the error is almost exclusively due to the parallax
332: uncertainty. Following \citet{dt91}, this can be translated into a
333: limit on the time derivative of Newton's gravitational constant (given
334: are $95\%$ confidence levels):
335: \begin{equation}
336: \frac{\dot{G}}{G} = -\frac{1}{2}\Big(\frac{\dot{P_{\rm b}}}{P_{\rm
337:     b}}\Big)^{\rm excess} = (-5\pm 18) \times 10^{-12}\, {\rm yr}^{-1}
338: \end{equation}
339: 
340: This limit is of the same order as those previously derived from
341: pulsar timing (see Section \ref{Introduction}), but a currently
342: ongoing VLBI campaign on this pulsar is expected to improve
343: significantly on our parallax measurement, and this should improve our
344: limit to close to that put by LLR \citep[\mbox{($4\pm 9) \times
345: 10^{-13}$\, yr$^{-1}$};][]{wtb04}.  The LLR experiment is based on a
346: complex $n$--body relativistic model of the planets that incorporates
347: over 140 estimated parameters, such as elastic deformation, rotational
348: dissipation and two tidal dissipation parameters.  In contrast, the
349: \psr\ timing result is dependent on a different set of models and
350: assumptions, and therefore provides a useful independent confirmation
351: of the LLR result.
352: 
353: A recent investigation into the possible causes of a measured
354: variability of the Astronomical Unit \citep[$AU$;][]{kb04} has refuted
355: all but two sources of the measured value of $d AU/dt = 0.15 \pm
356: 0.04\, {\rm m/yr}$. \citet{kb04} state that the measured linear
357: increase in the $AU$ would be due to either systematic effects or to a
358: time-variation of $G$ at the level of $\dot{G}/G = (-10\pm 3) \times
359: 10^{-13}\, {\rm yr}^{-1}$, comparable to, but inconsistent with, the
360: LLR limit.
361: 
362: The anomalous \pbdot\ measurements of a number of millisecond pulsars
363: have also been used to place limits on the acceleration of the Solar
364: System due to any nearby stars or undetected massive planets
365: \citep{zt05}. The \psr\ data set limits any anomalous Solar System
366: acceleration to $ | a_{\odot}/c | \leq 1.5\times 10^{-18}\, {\rm
367: s}^{-1}$ in the direction of the pulsar with $95$\% certainty.  This
368: rules out any Jupiter-mass planets at distances less than $117\,$AU
369: along the line of sight, corresponding to orbital periods of up to
370: 1270\,years. Similarly, this analysis excludes any Jupiter-mass
371: planets orbiting \psr\ between $\sim$5 and $117\,$AU along the line of
372: sight. \citet{zt05} also compared the sensitivity of this limit to
373: that of optical and infra-red searches for trans-Neptunian objects (TNOs) and
374: concluded that beyond $\sim 300$\,AU the acceleration limit becomes
375: more sensitive than the alternative searches. At a distance of
376: $300\,$AU from the Sun, the $95\%$ confidence upper limit on the mass
377: of a possible TNO (in the direction of the pulsar) is $6.8$ Jupiter
378: masses. The precise VLBI measurement of parallax mentioned above
379: might decrease this to close to one Jupiter mass.
380: 
381: \section{Pulsar Mass}\label{Mass}
382: 
383: A combination of the mass function and a measurement of the Shapiro
384: delay range can be used to obtain a measurement of the pulsar
385: mass. Using this method, \citet{vbb+01} derived a mass for \psr\ of
386: $1.58 \pm 0.18\,M_{\odot}$ whereas \citet{hbo06} obtained $1.3 \pm
387: 0.2\,M_{\odot}$. It should be noted, however, that these values
388: resulted from a model that incorporated geometric parameters first
389: described by \citet{kop95,kop96}, but covariances between these and
390: other timing parameters (most importantly the companion mass or
391: Shapiro delay range) were not taken into account. Whilst the length of
392: the data sets used by these authors were only a few years, it can also
393: be expected that some spectral leakage from low-frequency noise was
394: unaccounted in the errors of these previously published values. As
395: described in Section \ref{Obs}, the Monte-Carlo simulations and
396: extended fitting routines implemented for the results reported in this
397: paper do include these covariances and spectral leakage; it can
398: therefore be claimed that the current estimates (at $68\%$ confidence)
399: of $m_{\rm c} = 0.254 \pm 0.018\,M_{\odot}$ and $m_{\rm psr} = 1.76
400: \pm 0.20\,M_{\odot}$, for the white dwarf companion and pulsar
401: respectively, reflect the measurement uncertainty more realistically
402: than any previous estimate.  The distribution of $m_{\rm psr}$ that
403: follows from the 5000 Monte-Carlo realizations is shown in Figure
404: \ref{PSRMass}, together with a Gaussian with mean $1.76$ and standard
405: deviation $0.20$. This demonstrates the symmetric distribution of the
406: pulsar mass likelihood distribution, induced by the precise
407: determination of the orbital inclination angle.
408: 
409: We also note that the new mass measurement of \psr\ is the highest
410: obtained for any pulsar to date. Distinction needs to be made between
411: the objective mass estimate presented in this paper and the subjective
412: mass predictions presented in \citet{rhs+05} and \citet{frb+07}. The
413: pulsar mass confidence interval presented in this paper is derived
414: from the measurement uncertainties of all relevant model parameters,
415: including the well-determined orbital inclination angle, $i$. In
416: contrast, $i$ is unknown in the Terzan 5\,I and J \citep{rhs+05} and
417: PSR J1748$-$2021B \citep{frb+07} binary systems, and the posterior
418: probability intervals for the pulsar masses presented in these works
419: are based upon the prior assumption of a uniform distribution of $\cos
420: i$. These fundamental differences must be accounted for in any
421: subsequent hypothesis testing. 
422: Consequently, \psr\ is currently the only pulsar to provide reliable
423: constraints on equations of state based on hyperons and Bose-Einstein
424: condensates as described by \citet{lp07}. Simulations with
425: \textsc{Tempo2} indicate that a forthcoming observational campaign
426: with a new generation of backend systems can be expected to increase
427: the significance of this measurement by another factor of about two in
428: the next year.
429:  
430: \section{Conclusions}\label{Conc}
431: 
432: We have presented results from the highest-precision long-term timing
433: campaign to date. With an overall residual rms of 199\,ns, the
434: 10\,years of timing data on \psr\ have provided a precise measurement
435: of the orbital period derivative, \pbdot, leading to the first
436: accurate kinematic distance to a millisecond pulsar: $D_{\rm k} =
437: 157.0\pm 2.4$\,pc. Application of this method to other pulsars in the
438: future can be expected to improve distance estimates to other binary
439: pulsar systems \citep{bb96}.
440: 
441: Another analysis based on the \pbdot\ measurement places a limit on
442: the temporal variation of Newton's gravitational constant. We find a
443: bound comparable to the best so far derived from pulsar timing:
444: $\dot{G}/G = (-5\pm18)\times 10^{-12}\,{\rm yr}^{-1}$.  An ongoing
445: VLBI campaign on this pulsar is expected to improve this limit,
446: enabling an independent confirmation of the LLR limit.
447: 
448: Previous estimates of the mass of \psr\ have been revised upwards to
449: $m_{\rm psr} = 1.76 \pm 0.20\, M_{\odot}$, which now makes it one of
450: the few pulsars with such a heavy mass measurement.  A new generation
451: of backend instruments, dedicated observing campaigns, and data
452: prewhitening techniques currently under development should decrease
453: the error in this measurement enough to significantly rule out various
454: equations of state for dense nuclear matter.
455: 
456: \acknowledgments
457: 
458: The Parkes Observatory is part of the Australia Telescope which is
459: funded by the Commonwealth of Australia for operation as a National
460: Facility managed by CSIRO.  We thank the staff at Parkes Observatory
461: for technical assistance during regular observations. B.A.J holds a
462: National Research Council Research Associateship Award at the Naval
463: Research Laboratory (NRL). Basic research in radio astronomy at NRL is
464: supported by the Office of Naval Research. The authors wish to express
465: their gratitude to William Coles from the University of California at
466: San Diego, for extensive discussion and help with the Monte-Carlo
467: error analysis. Finally, we thank the referee for valuable and
468: inspiring comments.
469: 
470: {\it Facilities:} \facility{Parkes}
471: 
472: \begin{thebibliography}{}
473: 
474: \bibitem[Becker \& Tr\"umper (1993)]{bt93a}
475: Becker, W. and Tr\"umper, J. 1993, Nature, 365, 528.
476: 
477: \bibitem[Bell \& Bailes (1996)]{bb96}
478: Bell, J.~F. and Bailes, M. 1996, \apj, 456, L33.
479: 
480: \bibitem[Bell, Bailes \& Bessell (1993)]{bbb93}
481: Bell, J.~F., Bailes, M., and Bessell, M.~S. 1993, Nature, 364, 603.
482: 
483: \bibitem[Britton (2000)]{bri00}
484: Britton, M.~C. 2000, \apj, 532, 1240.
485: 
486: \bibitem[Damour \& Deruelle (1986)]{dd86}
487: Damour, T. and Deruelle, N. 1986, Ann. Inst. H. Poincar\'e (Physique
488: Th\'eorique), 44, 263.
489: 
490: \bibitem[Damour \& Taylor (1991)]{dt91}
491: Damour, T. and Taylor, J.~H. 1991, \apj, 366, 501.
492: 
493: \bibitem[Edwards, Hobbs \& Manchester (2006)]{ehm06}
494: Edwards, R.~T., Hobbs, G.~B., and Manchester, R.~N. 2006, \mnras, 372, 1549.
495: 
496: \bibitem[Foster \& Backer (1990)]{fb90}
497: Foster, R.~S. and Backer, D.~C. 1990, \apj, 361, 300.
498: 
499: \bibitem[{Freire} { et al.} (2007)]{frb+07}
500:   {Freire}, P.~C.~C., {Ransom}, S.~M., {Begin}, S., {Stairs},
501:   I.~H., {Hessels}, J.~W.~T., {Frey}, L.~H., and {Camilo}, F. 2007,
502:   ArXiv e-prints, 0711.0925.
503: 
504: \bibitem[Hobbs (2005)]{hob05}
505: Hobbs, G. 2005, Proc. Astron. Soc. Austral., 22, 179.
506: 
507: \bibitem[Hobbs, Edwards \& Manchester (2006)]{hem06}
508: Hobbs, G.~B., Edwards, R.~T., and Manchester, R.~N. 2006, \mnras, 369, 655.
509: 
510: \bibitem[{Holmberg} \& {Flynn} (2004)]{hf04b}
511: {Holmberg}, J. and {Flynn}, C. 2004, \mnras, 352, 440.
512: 
513: \bibitem[Hotan, Bailes \& Ord (2006)]{hbo06}
514: Hotan, A.~W., Bailes, M., and Ord, S.~M. 2006, \mnras, 369, 1502.
515: 
516: \bibitem[{Hotan}, {van Straten} \& {Manchester} (2004)]{hvm04}
517: {Hotan}, A.~W., {van Straten}, W., and {Manchester}, R.~N. 2004, Proc. Astron.
518:   Soc. Austral., 21, 302.
519: 
520: \bibitem[{Jenet} { et al.}  (2006)]{jhv+06}
521: {Jenet}, F.~A. { et al.}  2006, \apj, 653, 1571.
522: 
523: \bibitem[Johnston { et al.}  (1993)]{jlh+93}
524: Johnston, S. { et al.}  1993, Nature, 361, 613.
525: 
526: \bibitem[Kaspi, Taylor \& Ryba (1994)]{ktr94}
527: Kaspi, V.~M., Taylor, J.~H., and Ryba, M. 1994, \apj, 428, 713.
528: 
529: \bibitem[Kopeikin (1995)]{kop95}
530: Kopeikin, S.~M. 1995, \apj, 439, L5.
531: 
532: \bibitem[Kopeikin (1996)]{kop96}
533: Kopeikin, S.~M. 1996, \apj, 467, L93.
534: 
535: \bibitem[{Krasinsky} \& {Brumberg} (2004)]{kb04}
536: {Krasinsky}, G.~A. and {Brumberg}, V.~A. 2004, Celestial Mechanics and
537:   Dynamical Astronomy, 90, 267.
538: 
539: \bibitem[Lattimer \& Prakash (2007)]{lp07}
540: Lattimer, J.~M. and Prakash, M. 2007, Phys. Rep., 442, 109.
541: 
542: \bibitem[Nordtvedt (1990)]{nor90}
543: Nordtvedt, K. 1990, Phys. Rev. Letters, 65, 953.
544: 
545: \bibitem[{Ransom} { et al.} (2005)]{rhs+05}
546: {Ransom}, S.~M., {Hessels}, J.~W.~T., {Stairs}, I.~H., {Freire}, P.~C.~C.,
547:   {Camilo}, F., {Kaspi}, V.~M., and {Kaplan}, D.~L. 2005, Science, 307, 892.
548: 
549: \bibitem[Sandhu (2001)]{san01}
550: Sandhu, J.~S. 2001.
551: \newblock PhD thesis, California Institute of Technology.
552: 
553: \bibitem[Sandhu { et al.}  (1997)]{sbm+97}
554: Sandhu, J.~S., Bailes, M., Manchester, R.~N., Navarro, J., Kulkarni, S.~R., and
555:   Anderson, S.~B. 1997, \apjlett, 478, L95.
556: 
557: \bibitem[Shklovskii (1970)]{shk70}
558: Shklovskii, I.~S. 1970, Soviet Astron. AJ, 13, 562.
559: 
560: \bibitem[Splaver { et al.}  (2005)]{sns+05}
561: Splaver, E.~M., Nice, D.~J., Stairs, I.~H., Lommen, A.~N., and Backer, D.~C.
562:   2005, \apj, 620, 405.
563: 
564: \bibitem[Standish (2004)]{sta04b}
565: Standish, E.~M. 2004, \astap, 417, 1165.
566: 
567: \bibitem[Staveley-Smith { et al.} (1996)]{swb96}
568: {Staveley-Smith}, L. { et al.}  1996, Publications of the Astronomical Society
569:   of Australia, 13, 243.
570: 
571: \bibitem[Stinebring { et al.}  (1990)]{srtr90}
572: Stinebring, D.~R., Ryba, M.~F., Taylor, J.~H., and Romani, R.~W. 1990, Phys.
573:   Rev. Letters, 65, 285.
574: 
575: \bibitem[{Taylor} (1993)]{tay93}
576: {Taylor}, J.~H. 1993, in { Particle Astrophysics}, ed.\ G. {Fontaine} and J.
577:   {Tran Thanh van}, 367.
578: 
579: \bibitem[Taylor \& Weisberg (1982)]{tw82}
580: Taylor, J.~H. and Weisberg, J.~M. 1982, \apj, 253, 908.
581: 
582: \bibitem[Thorsett \& Chakrabarty (1999)]{tc99}
583: Thorsett, S.~E. and Chakrabarty, D. 1999, \apj, 512, 288.
584: 
585: \bibitem[{Torres} { et al.}  (2007)]{tlmr07}
586: {Torres}, R.~M., {Loinard}, L., {Mioduszewski}, A.~J., and {Rodriguez}, L.~F.
587:   2007, ArXiv e-prints, 708.
588: 
589: \bibitem[{Valls-Gabaud} (2007)]{val07}
590: {Valls-Gabaud}, D. 2007, in { IAU Symposium}, volume 240 of { IAU Symposium},
591:   281.
592: 
593: \bibitem[van Straten (2003)]{van03a}
594: van Straten, W. 2003.
595: \newblock PhD thesis, Swinburne University of Technology.
596: 
597: \bibitem[van Straten (2004)]{van04a}
598: van Straten, W. 2004, \apjsupp, 152, 129.
599: 
600: \bibitem[van Straten (2006)]{van06}
601: van Straten, W. 2006, \apj, 642, 1004.
602: 
603: \bibitem[van Straten { et al.}  (2001)]{vbb+01}
604: van Straten, W., Bailes, M., Britton, M., Kulkarni, S.~R., Anderson, S.~B.,
605:   Manchester, R.~N., and Sarkissian, J. 2001, Nature, 412, 158.
606: 
607: \bibitem[Williams, Turyshev \& Boggs (2004)]{wtb04}
608: Williams, J.~G., Turyshev, S.~G., and Boggs, D.~H. 2004, Phys. Rev. Letters,
609:   93(26), 261101.
610: 
611: \bibitem[You { et al.}  (2007)]{yhc+07}
612: You, X.-P. { et al.}  2007, \mnras, 378, 492.
613: 
614: \bibitem[Zakamska \& Tremaine (2005)]{zt05}
615: Zakamska, N.~L. and Tremaine, S. 2005, \aj, 130, 1939.
616: 
617: \end{thebibliography}
618: 
619: \clearpage
620: 
621: \begin{figure}
622: \includegraphics[width=8.2cm]{f1.eps}
623: \caption{Combined 20\,cm post-fit timing residuals for new and
624:   archival \psr\ timing data. Vertical dashed lines separate the
625:   different instruments.\label{Fig::Res}}
626: \end{figure}
627: 
628: \begin{figure}
629: \includegraphics[width=6.15cm,angle=-90.0]{f2.eps}
630: \caption{Pulsar mass probability distribution. The solid line shows the
631:   histogram of 5000 pulsar masses derived from a Monte-Carlo
632:   simulation with power spectrum and sampling equal to that of the
633:   \psr\ data set. The dashed line is a Gaussian distribution with a
634:   mean value of $m_{\rm psr} = 1.76\,M_{\odot}$ and standard deviation
635:   of $0.20\,M_{\odot}$. 
636:   \label{PSRMass}}
637: \end{figure}
638: 
639: \begin{figure}
640: \includegraphics[width=8.2cm]{f3.eps}
641: \caption{Variations in epoch of periastron passage ($T_{\rm 0}$) due
642:   to apparent orbital period increase. A steady increase in orbital
643:   period is equivalent to a quadratic increase in $T_{\rm 0}$ relative
644:   to periastron times for a constant orbital period.  For this plot,
645:   $T_{\rm 0}$ was measured on data spans of up to 120 days with a
646:   model having no orbital period derivative. The formal one-$\sigma$
647:   measurement errors reported by \textsc{Tempo2} are shown by vertical
648:   error bars and the epochs 
649:   over which the measurements were made are shown by horizontal bars. As
650:   the mean measurement time was determined through a weighted average
651:   of the data contained in the fit, these horizontal bars need not be
652:   centred at the mid time associated with the measurement.  
653:   The parabola shows the effect of the \pbdot\
654:   value obtained from a fit to the data shown in Figure
655:   \ref{Fig::Res}.\label{Fig::Pbdot}}
656: \end{figure}
657: 
658: \begin{figure}
659: \includegraphics[width=8.2cm]{f4.eps}
660: \caption{Parallax signature of \psr. Top: Timing residuals for \psr\
661:   as a function of day of year (starting on 18 November), without
662:   parallax but with all remaining parameters at their best-fit
663:   values. The smooth curve represents the model fit of a parallax of
664:   6.65 mas. Bottom: The same timing residuals with parallax included
665:   in the model. The overall rms for the top and bottom plots is $524$
666:   and $199$\,ns respectively. 
667:   The double-humped signature specific
668:   to parallax originates from the delay in pulse time-of-arrival (TOA)
669:   as 
670:   the Earth orbits the Sun and samples different parts of the curved
671:   wave-front originating at the pulsar.\label{Fig::Px}}
672: \end{figure}
673:  
674:  \clearpage
675:  
676: \begin{deluxetable}{cclrrccl}
677: \tablecaption{Characteristics of the timing data for the four
678:   instruments used\label{tbl::Obs}}
679: \tablewidth{0pt}
680: \tablehead{
681: \colhead{Backend} & \colhead{Date range} & \colhead{Ref.} &
682: \colhead{Bandwidth} & \colhead{RMS} & \colhead{Observation} &
683: \colhead{Number of} & \colhead{TOA} \\
684: \colhead{} & \colhead{} & \colhead{} &
685: \colhead{} & \colhead{Residual} & \colhead{length\tablenotemark{a}} &
686: \colhead{TOAs} & \colhead{error\tablenotemark{a}} 
687: 
688: }
689: \tablecolumns{8}
690: \startdata
691: FPTM & 1996 Apr -- 1997 May & 1,2 & $256\,$MHz &
692:   $368\,$ns & $10\,$min & 207 & $500\,$ns \\
693: S2 & 1997 Jul -- 1998 Apr & 3 & $16\,$MHz & $210\,$ns
694:   & $120\,$min & 117 & $160\,$ns \\
695: CPSR & 1998 Aug -- 2002 Aug & 3 & $20\,$MHz &
696:   $218\,$ns & $15\,$min & 1782 & $250\,$ns \\
697: CPSR2 & 2002 Nov -- 2006 Mar & 4 & $2\times
698: 64\,$MHz\tablenotemark{b} &
699:   $164\,$ns & $60\,$min & 741 & $150\,$ns \\
700: \enddata
701: \tablenotetext{a}{Displayed are typical values only.}
702: \tablenotetext{b}{CPSR2 records two adjacent $64\,$MHz bands
703:   simultaneously at 20\,cm.}
704: \tablerefs{
705:   (1) \citet{sbm+97}; (2) \citet{san01}; (3) \citet{van03a}; (4)
706:   \citet{hbo06}.}
707: \end{deluxetable}
708: 
709: \clearpage
710: 
711: \begin{deluxetable}{llrcr}
712: \tablecaption{\psr\ timing model
713:   parameters\tablenotemark{a}\label{Model}} 
714: \tablewidth{0pt}
715: \tablehead{
716:   \colhead{Parameter Name} & \colhead{Parameter} &
717:   \colhead{\textsc{Tempo2}}& \colhead{Monte-Carlo} & \colhead{Error}\\ 
718:   \colhead{and Units} & \colhead{Value} &
719:   \colhead{Error\tablenotemark{b}} & \colhead{Error\tablenotemark{b}}
720:   & \colhead{Ratio} 
721: }
722: \tablecolumns{2}
723: \startdata
724: \cutinhead{Fit and Data Set}
725: MJD range \dotfill & 50191.0--53819.2 & & & \\
726: Number of TOAs \dotfill & 2847 & & & \\
727: Rms timing residual ($\mu$s) \dotfill & 0.199 & & & \\
728: \cutinhead{Measured Quantities}
729:   Right ascension, $\alpha$ (J2000)       \dotfill &
730:   04$^{\mathrm h}$37$^{\mathrm m}$15\fs8147635 & 3 & 29 & 9.8\\
731:   Declination, $\delta$ (J2000)           \dotfill & 
732:   $-$47\degr15\arcmin08\farcs624170 & 3 & 34 & 11\\
733:   Proper motion in $\alpha$, $\mu_\alpha \cos{\delta}$ (mas yr$^{-1}$)
734:   \dotfill &  121.453 & 1 & 10 & 8.7 \\ 
735:   Proper motion in $\delta$, $\mu_\delta$ (mas yr$^{-1}$) \dotfill &
736:   $-$71.457 & 1 & 12 & 9.0\\ 
737:   Annual parallax, $\pi$ (mas)  \dotfill & 6.65 & 7 & 51 & 7.9 \\
738:   Dispersion measure, $DM$ (cm$^{-3}$ pc) \dotfill & 2.64476 & 7 &
739:   \tablenotemark{d} & \tablenotemark{d} \\
740:   Pulse period, $P$ (ms)                    \dotfill &
741:   5.757451924362137 & 2 & 99 & 47\\ 
742:   Pulse period derivative, $\dot{P}$ (10$^{-20}$) \dotfill &
743:   5.729370 & 2 & 9 & 4.8 \\
744:   Orbital period, $P_{\rm b}$ (days)        \dotfill &
745:   5.74104646\tablenotemark{c} & 108 & 200 & 1.9 \\
746:   Orbital period derivative, $\dot{P_{\rm b}}$ (10$^{-12}$) \dotfill &
747:   3.73 & 2 & 6 & 2.5\\
748:   Epoch of periastron passage, $T_{0}$ (MJD) \dotfill &
749:   52009.852429\tablenotemark{c} & 582 & 780 & 1.3\\
750:   Projected semi-major axis, $x$ (s)      \dotfill &
751:   3.36669708\tablenotemark{c} & 11 & 14 & 1.4\\
752:   Longitude of periastron, $\omega_0$ (\degr) \dotfill &
753:   1.2224\tablenotemark{c} & 365 & 490 & 1.3 \\
754:   Orbital eccentricity, $e$ (10$^{-5}$) \dotfill & 1.9180 & 3 &
755:   7 & 2.1\\ 
756:   Periastron advance, $\dot \omega$ (\degr\ yr$^{-1}$) \dotfill &
757:   0.01600\tablenotemark{c} & 430 & 800 & 1.8\\ 
758:   Companion mass, $m_{\rm 2}$ (M$_{\odot})$       \dotfill &
759:   0.254\tablenotemark{c} & 14 & 18 & 1.3 \\
760:   Longitude of ascension, $\Omega$ (\degr)  \dotfill &
761:   207.8\tablenotemark{c} & 23 & 69 & 3.0 \\
762:   Orbital inclination, $i$ (\degr)          \dotfill & 137.58 & 6
763:   & 21 & 3.7\\
764: \cutinhead{Set Quantities}
765: Reference epoch for $P$, $\alpha$ & & & & \\
766:   and $\delta$ determination (MJD)
767:   \dotfill & 52005 & & & \\
768:   Reference epoch for DM &  &  & & \\
769:   determination (MJD) \dotfill & 53211 & & & \\
770: \enddata
771: \tablenotetext{a}{These parameters are determined using
772:   \textsc{Tempo2} which uses the International Celestial Reference
773:   System and Barycentric Coordinate Time. As a result this timing
774:   model must be modified before being used with an observing system
775:   that inputs \textsc{Tempo} format parameters. See \citet{hem06} for
776:   more information.}
777: \tablenotetext{b}{Given uncertainties are $1\sigma$ values in the last
778:   digits of the parameter values.}
779: \tablenotetext{c}{Because of large covariances, extra precision is given
780:   for selected parameters.}
781: \tablenotetext{d}{Dispersion measure was determined through alignment
782:   of simultaneous CPSR2 observations centred at 1341\,MHz and
783:   1405\,MHz. The effect of red noise is therefore not applicable.} 
784: \end{deluxetable}
785: 
786: \clearpage
787: 
788: \begin{deluxetable}{lll}
789: \tablewidth{0pt} 
790: \tablecaption{Comparison of DE200 and DE405 results for
791:   \psr\tablenotemark{a}
792:   \label{DEModels}}
793: \tablehead{
794: \colhead{Parameter name} & \colhead{DE200 result} & \colhead{DE405
795:   result}}
796: \startdata
797: Rms residual (ns) \dotfill & 281 & 199 \\
798: Relative $\chi^2$ \dotfill & 2.01  & 1.0 \\
799: Parallax, $\pi$ (mas) \dotfill & 7.84(7)  & 6.65 (7) \\
800: Parallax distance, $D_{\pi}$ (pc) \dotfill & 127.6(11) & 150.4(16) \\
801: Previously published $\pi$ (mas) \dotfill 
802: & 7.19(14)\tablenotemark{1} 
803:  & 6.3(2)\tablenotemark{2} \\
804: Kinematic distance, $D_k$ (pc) \dotfill & 154.5 (10) &
805: 156.0 (10) \\  
806: $D_k$ corrected for Galactic effects (pc) \dotfill & 
807:    155.5 (10) & 157.0 (10) \\
808: Variation of Newton's gravitational & & \\
809: constant, $| {\dot{G} / G} |$ (10$^{-12}$ yr$^{-1}$)\dotfill &
810: $-21.2(22)$\tablenotemark{b}  & $-5.0(26)$\tablenotemark{b} \\
811: Total proper motion, $\mu_{tot}$ (mas yr$^{-1}$) \dotfill & 140.852(1)
812: & 140.915(1) \\ 
813: Companion mass, $m_{\rm c}$ ($M_{\odot}$) \dotfill & 0.263(14)
814:  & 0.254(14) \\ 
815: Pulsar mass, $m_{\rm psr}$ ($M_{\odot}$) \dotfill & 1.85(15)
816:  & 1.76(15) \\ 
817: Periastron advance, $\dot{\omega}$ (\degr\ yr$^{-1}$) \dotfill &
818: 0.020(4)  & 0.016(4) \\ 
819: GR prediction of $\dot{\omega}$ (\degr\ yr$^{-1}$)
820: \dotfill & 0.0178(9)  & 0.0172(9) 
821: \enddata
822: \tablenotetext{a}{Numbers in parentheses represent the formal \textsc{Tempo2}
823:   $1\,\sigma$ uncertainty in the last digits quoted, unless otherwise stated.}
824: \tablenotetext{b}{Given are $2\,\sigma$ errors, i.e. $95\%$ confidence levels.}
825: \tablerefs{
826: (1) \citet{vbb+01}; (2) \citet{hbo06}}
827: \end{deluxetable}
828: \end{document}