Existing work has examined these effects asymptotically, but implementable finite sample results are needed in practice so that users can specify privacy budgets in advance and implement samplers with exact privacy guarantees.

Code is available at this repository: https://github.com/thuml/Koopa.

The proposed template encompasses a wide range of popular first-order methods, including dual averaging, dual extrapolation andoptimistic gradient algorithms -both adaptive and non-adaptive. Our first result isthat thealgorithm achievestheoptimal rates of convergence for cocoercive problems when the profile of the randomness is known to the optimizer: O(1/ T) for absolute noise profiles, andO(1/T) for relative ones.