Improving pulsar timing precision with single pulses

The measurement error of pulse times of arrival (TOAs) in the high signal-to-noise ratio limit is dominated by the quasi-random variation of a pulsar's emission profile from rotation to rotation. Like measurement noise, this noise is only reduced as the square root of observing time, posing a major challenge to future pulsar timing campaigns with large aperture telescopes, e.g. the Five-hundred-metre Aperture Spherical Telescope and the Square Kilometre Array. We propose a new method of pulsar timing that attempts to approximate the pulse-to-pulse variability with a small family of 'basis' pulses. If pulsar data are integrated over many rotations, this basis can be used to measure subpulse structure. Or, if high-time resolution data are available, the basis can be used to 'tag' single pulses and produce an optimal timing template. With realistic simulations, we show that these applications can dramatically reduce the effect of pulse-to-pulse variability on TOAs. Using high-time resolution data taken from the bright PSR J0835-4510 (Vela), we demonstrate a 25-40 per cent improvement in TOA precision. Crucially for pulsar timing applications, we further establish that these techniques produce TOAs with Gaussian residuals. Improvements of this level halve the telescope time required to reach a desired TOA precision. Although some gains can be achieved with existing data, the greatest improvements result from the 'tagging' approach, which in turn requires online or post hoc analysis of single pulses, an important consideration for the design of future instrumentation.