Improving pulsar-timing solutions through dynamic pulse fitting

Precision pulsar timing is integral to the detection of the nanohertz stochastic gravitational-wave background as well as understanding the physics of neutron stars. Conventional pulsar timing often uses fixed time and frequency-averaged templates to determine the pulse times of arrival, which can lead to reduced accuracy when the pulse profile evolves over time. We illustrate a dynamic timing method that fits each observing epoch using basis functions. By fitting each epoch separately, we allow for the evolution of the pulse shape epoch to epoch. We apply our method to PSR J1103−5403 and find evidence that it undergoes mode changing, making it the fourth millisecond pulsar to exhibit such behaviour. Our method, which is able to identify and time a single mode, yields a timing solution with a root-mean-square error of $1.343$ µs, a factor of 1.78 improvement over template fitting on both modes. In addition, the white-noise amplitude is reduced 4.3 times, suggesting that fitting the full data set causes the mode changing to be incorrectly classified as white noise. This reduction in white noise boosts the signal-to-noise ratio of a gravitational-wave background signal for this particular pulsar by 32 per cent. We discuss the possible applications for this method of timing to study pulsar magnetospheres and further improve the sensitivity of searches for nanohertz gravitational waves.