Title: Long-term statistics of pulsar glitches due to history-dependent avalanches

Abstract: Stress accumulation-relaxation meta-models of pulsar glitches make precise, microphysics-agnostic predictions of long-term glitch statistics, which can be falsified by existing and future timing data. Previous meta-models assume that glitches are triggered by an avalanche process, e.g. involving superfluid vortices, and that the probability density function (PDF) of the avalanche sizes is history-independent and specified exogenously. Here a recipe is proposed to generate the avalanche sizes endogenously in a history-dependent manner, by tracking the thresholds of occupied vortex pinning sites as a function of time. Vortices unpin spasmodically from sites with thresholds below a global, time-dependent stress and repin at sites with thresholds above the global stress, imbuing the system with long-term memory. The meta-model predicts PDFs, auto- and cross-correlations for glitch sizes and waiting times, which are provisionally inconsistent with current observations, unlike some previous meta-models (e.g. state dependent Poisson process), whose predictions are consistent. The theoretical implications are intriguing, albeit uncertain, because history-dependent avalanches embody faithfully the popular, idealized understanding in the literature of how vortex unpinning operates as a driven, stochastic process. The meta-model predicts aftershocks, which occur with larger than average sizes and longer than average waiting times after the largest, system-resetting glitches. This prediction will be tested, once more data are generated by the next generation of pulsar timing campaigns.