Whereas many earthquakes on Earth occur at depths less than 60 km, around 25 % happen at larger depths of up 700 km. Despite their distance from the surface, large deep earthquakes represent a considerable seismic hazard. These earthquakes are particularly intriguing, as unattainably large forces are required to break rocks at those depths, a problem that was already recognized by K. Wadati in 1928. In the 90 years that passed since, different hypotheses have surfaced to explain these events, but despite the wealth of observational data and laboratory experiments, the underlying nucleation and rupture mechanisms of deep earthquakes remain enigmatic. To solve this long-standing problem, QuakeID will employ an interdisciplinary approach with numerical modelling at its core. Numerical simulations will be combined with novel laboratory experiments and seismological observations, to upscale results from the laboratory to the Earth. For this, cutting edge numerical models will be developed that can be applied at both seismic and geological time scales to model deep earthquake rupture. Together with specifically designed high pressure/high temperature experiments, they will be used to

1. determine the conditions which favour deep earthquake generation,
2. determine the rupture mechanism(s) responsible for earthquake nucleation at the respective depth,
3. identify feedbacks between different rupture mechanism,
4. test the hypothesis that a cascade of rupture mechanisms is responsible for generating large deep earthquakes and
5. provide synthetic results that can be compared to seismological and geological observations. QuakeID will thus solve the crucial research problem of how deep earthquakes are generated, which will improve seismic hazard assessment and, more generally, contribute to a better understanding of the stability of deep geothermal reservoirs as well as general non-brittle failure processes in material science.