For large data, should we create an option to calculate ACEs etc. on a subset of the data to speed up the calculation? I would in fact suggest to set the default for ACEs to n=5000 or so and only calculate with the full data if the user explicitly requests it.