M E Lower

A systematic, semi-automated search for pulsar glitches in the first UTMOST public data release is presented. The search is carried out using a hidden Markov model which incorporates both glitches and timing noise into the model of the assumed phase evolution of the pulsar. Glitches are detected through Bayesian model selection between models with and without glitches present with minimal human intervention. Nine glitches are detected among seven objects, all of which have been previously reported. No new glitches were detected. Injection studies are used to place 90 per cent frequentist upper limits on the size of undetected glitches in each of the 282 objects searched. The mean upper limit obtained is $\Delta f^{90{{\%}}}/f = 1.9 \times 10^{-8}$, with a range of $4.1 \times 10^{-11} \le \Delta f^{90{{\%}}}/f \le 2.7 \times 10^{-7}$, assuming step events with no post-glitch recoveries. It is demonstrated that including glitch recovery has a mild effect, in most cases increasing the upper limit by a factor of ≲5 conservatively assuming complete recovery on a time-scale of $100\ \mathrm{d}$.