The timing behaviour of radio pulsars George Hobbs Australia Telescope National Facility

CSIRO. Gravitational wave detection Contents Radio pulsars Pulsar timing A few things that you can do with pulsar timing Young pulsars - a new (predictive?) model for timing noise What has this to do with this conference? - pulsars are compact objects - radio pulsar timing is a powerful technique for studying pulsars - can determine parameters of interest - neutron star masses, rotation rates etc. - can study the pulsar spin-down => implications for internal structure of neutron star.

CSIRO. Gravitational wave detection Let’s start at the beginning 08:35: :10:34.87

CSIRO. Gravitational wave detection Radio pulsars Animation: Michael Kramer

CSIRO. Gravitational wave detection Properties of radio pulsars

CSIRO. Gravitational wave detection Must average many thousands of pulses together to obtain stable profile Must convert to reference frame suitable for the timing model – e.g. solar system barycentre Must convert to arrival times at infinite frequency Must convert to conform with terrestrial time standards Must add extra propagation delays e.g. through the solar system Pulsar timing: The basics (see Hobbs, Edwards & Manchester 2006, MNRAS) Obtain pulse arrival times at observatory Model for pulsar spin down Form timing residuals – how good is the timing model at predicting the arrival times Improve timing model

CSIRO. Gravitational wave detection What can we do with the timing model?

CSIRO. Gravitational wave detection Some examples: pulsar velocities With long data spans can get accurate pulsar proper motions - with a distance estimate can obtain velocities. Mean space velocity ~ 400km/s (Hobbs et al. 2005)

CSIRO. Gravitational wave detection Determine pulsar masses and testing GR Champion et al. (2008) Science: PSR J , NS mass = 1.74 ± 0.04 M o (unusually large) Double pulsar (B) has mass 1.25 M o - significantly smaller (Lyne et al Sci)

CSIRO. Gravitational wave detection What can we do with the timing residuals? Residuals are a measure of unmodelled physics Are these residuals from … The pulsar spin-down Terrestrial time standards Pulse propagation through the interstellar medium Orbital companions to the pulsar Gravitational waves! Errors in the planetary ephemeris …

CSIRO. Gravitational wave detection Spin-down irregularities No angular signature

CSIRO. Gravitational wave detection Terrestrial time standard irregularities Monopolar signature

CSIRO. Gravitational wave detection Errors in the planetary ephemerides - e.g. error in the mass of Jupiter Dipolar signature

CSIRO. Gravitational wave detection What if gravitational waves exist? Quadrapolar signature

CSIRO. Gravitational wave detection The post-fit planet ‘signal’: The effect of fitting CSIRO. Measuring the mass of Jupiter using pulsars Jupiter Mars Simulations of 10 years of pulsar residuals with an RMS of 100ns

CSIRO. Gravitational wave detection Current status (Champion et al. 2009, in prep) Use data from Parkes, Arecibo, Effelsberg and Nancay radio telescopes M Sun Best PublishedThis work Mercury (7)x (2)x10 -7 Venus (4)x (10)x10 -6 Mars (9)x (8)x10 -7 Jupiter* (8)x (4)x10 -4 Saturn (8)x (14)x (11)x10 -4

CSIRO. Gravitational wave detection The timing residuals of young pulsars 76-m Lovell Radio Telescope 366 pulsars with tspan > 10yr Hobbs, Lyne & Kramer (2004) Not high time precision experiments

CSIRO. Gravitational wave detection Pulsar timing residuals (fit for F0 and F1)

CSIRO. Gravitational wave detection Difficulties when categorising timing noise B B

CSIRO. Gravitational wave detection Difficulties when categorising timing noise: depends on data span PSR B Any simple classification scheme would change with data span. Most large-scale analyses of timing noise used ~3 yr of data.

CSIRO. Gravitational wave detection What timing noise is not! Not observatory dependent - many pulsars also observed at other observatories - see same timing noise Not off-line processing (use ‘tempo2’ and ‘psrtime’) Not terrestrial time scales/planetary ephemeris errors - too large Not ISM effect - not frequency dependent

CSIRO. Gravitational wave detection Previous models of timing noise Random walks in the pulse frequency or its derivatives Free-precession of the neutron star Unmodelled planetary companions Asteroid belts Magnetospheric effects Interstellar/interplanetary medium effects Unmodelled Post-Keplerian orbital parameters Accretion onto the pulsar’s surface Large numbers of small glitch events These models were based on short data sets Mainly model random, “noise-like” timing residuals

CSIRO. Gravitational wave detection Significant F2 values (= cubics in timing residuals) Glitch events => F2 > 0 (Lyne, Shemar & Graham-Smith 2000) All pulsars with  c 0 For older pulsars 52% have F2 > 0. Timing noise in young pulsars caused by glitch recovery. Timing noise in older pulsars caused by something else! Globular cluster pulsar

CSIRO. Gravitational wave detection Periodicities: B Significant 4.38yr periodicity If planet then Earth-mass. However, significant residuals remain in the timing after fitting for a planet

CSIRO. Gravitational wave detection Periodicities: B Time between successive peaks range from 3.4yr to 6.6yr Radius of curvature smaller at local maxima than at minima

CSIRO. Gravitational wave detection Periodicities: B Time between peaks ranges between 7 and 10 years. No significant individual periodicities.

CSIRO. Gravitational wave detection Periodicities: B Significant periodicity at 2.9yr (however time between peaks varies by ~10%). Local maxima have smaller curvature than minima

CSIRO. Gravitational wave detection Periodicities: B Significant periodicities - main periodicity at 500d. 3 components to the slow-down Modelled by Stairs et al. as free- precession

CSIRO. Gravitational wave detection Periodicities: B Significant periodicity at 3.2yr, 7.1yr and 2.1yr. Larger radius of curvature at maxima than at minima

CSIRO. Gravitational wave detection PSR B PSR B has recently been reported to undergo “extreme nulling” events (Kramer et al. 2006) Normal pulsar for 5 to 10 days Switches off for up to 35 days The pulsar spin-down rate changes by ~50% between the on and off states (pulsar spinning down faster when “on”)

CSIRO. Gravitational wave detection Directly looking at F1 values PSR J First pulsar we looked at: Has 2 F1 values Has correlated pulse shape changes

CSIRO. Gravitational wave detection Modelling B Implication: B is not undergoing free-precession! Undergoes mode 1, 2, 3, 2 ….

CSIRO. Gravitational wave detection More simulations

CSIRO. Gravitational wave detection PSR J Recently discovered pulsar with three pulse profiles: 1) a very strong profile (brightness rivals that of Vela) 2) weak profile 3) completely undetectable => some pulsars exhibit 3 “magnetospheric modes” - have not yet checked to look for correlated slow-down rates.

CSIRO. Gravitational wave detection The implications of the model Have a link between various time-dependent phenomena in pulsars: long-term moding/extreme nulling/intermittency/free-precession/timing noise Timing noise linked to magnetospheric changes Quasi-random nature of the mode switches => a random walk in F1 => large scale cubics can exist in the timing residuals Note: have no understanding of the process creating multiple spin-down rates, but it seems that large changes in spin-down rate => large pulse shape changes. 50% change in spin-down rate B (large shape changes) ~% change in spin-down rate B (moderate shape changes) Fraction of a % change in spin-down rate B (small shape changes?)

CSIRO. Gravitational wave detection Glitches

CSIRO. Gravitational wave detection An aside: slow glitches Zou et al. (2004) reported a new phenomenon known as “slow glitches” No difference between “slow glitches” and timing noise! B (vertical lines are slow- glitches according to Shabanova (2007)

CSIRO. Gravitational wave detection Glitches Sudden speedup in rotation period, relaxing back in days to years The pulse structure is not notably affected by a glitch => phenomena internal to the neutron star Current model is that superfluid vortices in the neutron star ‘pin’ to the surface/crust. Catastrophic unpinning leads to a glitch event. 285 glitches published in 101 objects. 65% of the glitching pulsars have only glitched once PSR J has glitched 33 times

CSIRO. Gravitational wave detection Glitches Melatos, Peralta & Wyithe (2008, ApJ) suggest that glitch events follow an avalanche model. Waiting time between glitches is consistent with a Poissionian process. … we’re writing a new paper containing more pulsar glitches … What would you like us to present? Clearly, the glitches are telling us something about the interior of the pulsar … but how do we extract the information?

CSIRO. Gravitational wave detection Conclusion You can do lots of physics/astronomy with radio pulsar timing observations Most millisecond pulsars are very stable rotators The spin-down of the youngest pulsars is dominated by glitch recovery The spin-down of most pulsars is dominated by a quasi- periodic phenomenon. This is probably telling us something about the interior of the neutron star!