Pulsar glitch activity as a state-dependent Poisson process: parameter estimation and epoch prediction. (arXiv:1910.05503v1 [astro-ph.HE])
<a href="http://arxiv.org/find/astro-ph/1/au:+Melatos_A/0/1/0/all/0/1">A. Melatos</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Drummond_L/0/1/0/all/0/1">L. V. Drummond</a>
Rotational glitches in some rotation-powered pulsars display power-law size
and exponential waiting time distributions. These statistics are consistent
with a state-dependent Poisson process, where the glitch rate is an increasing
function of a global stress variable (e.g. crust-superfluid angular velocity
lag), diverges at a threshold stress, increases smoothly while the star spins
down, and decreases step-wise at each glitch. A minimal, seven-parameter,
maximum likelihood model is calculated for PSR J1740-3015, PSR J0534+2200, and
PSR J0631+1036, the three objects with the largest samples whose glitch
activity is Poisson-like. The estimated parameters have theoretically
reasonable values and contain useful information about the glitch microphysics.
It is shown that the maximum likelihood, state-dependent Poisson model is a
marginally (23-27 per cent) better post factum “predictor” of historical glitch
epochs than a homogeneous Poisson process for PSR J1740-3015 and PSR J0631+1036
and a comparable predictor for PSR J0534+2200. Monte Carlo simulations imply
that > 50 glitches are needed to test reliably whether one model outperforms
the other. It is predicted that the next glitch will occur at Modified Julian
Date (MJD) 57784 +/- 256.8, 60713 +/- 1935, and 57406 +/- 1444 for the above
three objects respectively. The analysis does not apply to quasiperiodic
glitchers like PSR J0537-6910 and PSR J0835-4510, which are not described
accurately by the state-dependent Poisson model in its original form.
Rotational glitches in some rotation-powered pulsars display power-law size
and exponential waiting time distributions. These statistics are consistent
with a state-dependent Poisson process, where the glitch rate is an increasing
function of a global stress variable (e.g. crust-superfluid angular velocity
lag), diverges at a threshold stress, increases smoothly while the star spins
down, and decreases step-wise at each glitch. A minimal, seven-parameter,
maximum likelihood model is calculated for PSR J1740-3015, PSR J0534+2200, and
PSR J0631+1036, the three objects with the largest samples whose glitch
activity is Poisson-like. The estimated parameters have theoretically
reasonable values and contain useful information about the glitch microphysics.
It is shown that the maximum likelihood, state-dependent Poisson model is a
marginally (23-27 per cent) better post factum “predictor” of historical glitch
epochs than a homogeneous Poisson process for PSR J1740-3015 and PSR J0631+1036
and a comparable predictor for PSR J0534+2200. Monte Carlo simulations imply
that > 50 glitches are needed to test reliably whether one model outperforms
the other. It is predicted that the next glitch will occur at Modified Julian
Date (MJD) 57784 +/- 256.8, 60713 +/- 1935, and 57406 +/- 1444 for the above
three objects respectively. The analysis does not apply to quasiperiodic
glitchers like PSR J0537-6910 and PSR J0835-4510, which are not described
accurately by the state-dependent Poisson model in its original form.
http://arxiv.org/icons/sfx.gif
