What Goes into a Pulsar Timing Model? David Nice Physics Department, Princeton University Pulsar Timing Array: A Nanohe

1. What Goes into a Pulsar Timing Model? David Nice Physics Department, Princeton University Pulsar Timing Array: A Nanohertz Gravitational Wave Telescope Center for Gravitational Wave Physics Penn State University 21 July 2005

2. Observing The Pulsar Signal

3. Observing The Pulsar Signal Propagation of gravitational wave Basic idea: Pulses travel from pulsar to telescope. The passage of a gravitational wave perturbs the time-of-flight of the pulse. The perturbation is complicated! (GW period << pulse travel time) Search for patterns in pulse ‘arrival times’ indicative of gravitational waves. Many other phenomena affect measured arrival times. Analyzing all these phenomena is called ‘pulse timing’ and is the subject of this talk.

4. Observing The Pulsar Signal PSR J1713+0747Splaver et al. 2005 ApJ 620: 405 astro-ph/0410488

5. A long-wavelength gravitational wave might continuously increase the proper distance traveled by a pulsar over the entire observation program. The plot shows an increase in travel time of 2 s over 6 years, a distance of only 600 m (compared to pulsar distance of 1.1 kpc, for l/l=610-16). Exactly the same thing would be observed if the pulsar rotation period were slightly longer, 4.57013652508278 ms instead of 4.57013652508274 ms (1 part in 1014). The period is known only from timing data  always need to fit out a linear term in timing measurements to find pulse period  a perturbation due to gravitational waves which is linear in time cannot be detected

6. Pulsar rotation slows down over time due to magnetic dipole rotation. The ‘spin-down’ rate is not known a priori. The plot shows what the residuals of J1713+0747 look like if we forget to include spin-down. As the pulsar slows down, pulses are delayed by an amount quadratic in time. The spin-down rate is known only from timing data  always need to fit out a quadratic term in timing measurements to find pulse period  a perturbation due to gravitational waves which is quadratic in time cannot be detected

7. rotation period rotation period derivative

8. sun

9. sun Delays of ~500 s due to time-of-flight across the Earth’s orbit. The amplitude and phase of this delay depend on the pulsar position. Position known only from timing data  always need to fit annual terms out of timing solution  a perturbation due to gravitational waves with ~1 yr period cannot be detected

10. sun Other astrometric phenomena: Proper Motion

11. sun Curved wavefronts Other astrometric phenomena: Proper Motion Parallax

12. sun Measurement of a pulse time of arrival at the observatory is a relativistic event. It must be transformed to an inertial frame: that of the solar system barycenter. Time transfer: Observatory clock  GPS  UT  TDB Position transfer: For Earth and Sun positions, use a solar system ephemeris, e.g., JPL ‘DE405’ For earth orientation (UT1, etc.), use IERS bulletin B

13. PSR J1713+0747 analyzed using DE 405 solar system ephemeris PSR J1713+0747 analyzed using previous-generation DE 200 solar system ephemeris. ~1s timing errors  300 m errors in Earth position.

14. sun rotation period rotation period derivative position proper motion parallax

16. & w Precession 5 2 - æ ö é ù P 1 G 3 3 ( ) m m + b = 3 ç ÷ 1 2 ê ú 2 3 2 1 e c - è ø ë û p Shapiro Delay G [ ] ( ) j - j t 2 m ln 1 sin i sin = - D c 2 0 3 Grav Redshift/Time Dilation v r 1 2 m ( ) m +2m æ ö G P 3 3 2 1 2 e  b = ç ÷ ( ) 4 m m c 2 2 p è ø + 3 1 2 Gravitational Radiation 5 - 5 æ ö æ ö æ ö G P 73 37 1 m m 3 192 3 p & 2 4 + + P b 1 2 = - 1 e e ç ÷ ç ÷ ç ÷ b 7 ( ) 1 5 c 5 2 p ( 1 - e 2 ) m m è ø è ø è ø 24 96 2 + 3 1 2

17. PSR J1713+0747 Shapiro Delay

18. Apparent inverse correlation between orbital period and pulsar mass. (Nice et al., ApJ, submitted)

19. sun rotation period rotation period derivative Keplerian orbital elements relativistic orbital elements kinematic perturbations of orbital elements (secular and annual phenomena) position proper motion parallax

20. Interstellar Dispersion 431 MHz 430 MHz column density of electrons: DM =  ne(l) dl excess propagation time: t (sec) = DM / 2.4110–4 [f(MHz)]2

21. DM Variations in PSR J0621+1002 timing (Splaver et al., ApJ 581: 509, astro-ph/0208281)

22. Figure 1. Polar plots of solar wind speed as a function of latitude for Ulysses' first two orbits. Sunspot number (bottom panel) shows that the first orbit occurred through the solar cycle declining phase and minimum while the second orbit spanned solar maximum. Both are plotted over solar images characteristic of solar minimum (8/17/96) and maximum (12/07/00); from the center out, these images are from the Solar and Heliospheric Observatory (SOHO) Extreme ultraviolet Imaging Telescope (Fe XII at 195 Å), the Mauna Loa K-coronameter (700950 nm), and the SOHO C2 Large Angle Spectrometric Coronagraph (white light) Figure 2. Twelve-hour running averaged solar wind proton speed, scaled density and temperature, and alpha particle to proton ratio as a function of latitude for the most recent part of the Ulysses orbit (black line) and the equivalent portion from Ulysses' first orbit (red line). D J McComas et al 2003. Geophys Res Lett 30, 1517

23. sun rotation period rotation period derivative Keplerian orbital elements relativistic orbital elements kinematic perturbations of orbital elements (secular and annual phenomena) dispersion measure dispersion meas. variations position proper motion parallax solar electron density

24. The examples so far are from 6 years of 1713+0747 data.If all the phenomena discussed so far are removed from those pulse arrival times, the remaining residuals look nearly flat. But: incorporating two years of additional data (taken several years earlier), shows that the residuals are not flat over longer time scales. This is a common feature of pulsar timing data, called `timing noise.’ Timing noise is probably indicative of irregularities in pulsar rotation; its physical origin remains unclear.

25. The z timing noise statistic applied to J1713+0747

26. Timing noise and DM variations of the original millisecond pulsar, B1937+21

27. Timing noise in several young pulsars.

28. Single Binary High Eccentricity Low Eccentricity

29. Single Binary High Eccentricity Low Eccentricity Birth

30. Single Binary High Eccentricity Low Eccentricity Birth Death

31. Single Binary High Eccentricity Low Eccentricity Birth Death Spin-up Rebirth

32. Single Binary High Eccentricity Low Eccentricity

33. Single Binary High Eccentricity Low Eccentricity Smaller MeasurementUncertainty

34. Single Binary High Eccentricity Low Eccentricity Less Timing Noise Smaller MeasurementUncertainty

35. sun rotation period rotation period derivative Keplerian orbital elements relativistic orbital elements kinematic perturbations of orbital elements (secular and annual phenomena) dispersion measure dispersion meas. variations position proper motion parallax solar electron density Fit for all of these simultaneously to find the best pulsar timing solution (‘tempo’), then examine the residuals for signs of gravitational waves.