Following what Ryan did for HLL sketches, I would also add an aggregate expression for unions as the aggregate version of the binary union expression.
The expressions that Ryan added are: hll_sketch_agg hll_union hll_union_agg hll_sketch_estimate Following the same naming convention I would probably go for: theta_sketch_agg(sketch_col) theta_union(sketch1, sketch2) theta_union_agg(sketch_col) theta_difference(sketch1, sketch2) theta_intersection(sketch1, sketch2) theta_intersection_agg(sketch_col) theta_sketch_estimate(sketch) I do not think having an aggregate expression for differences would make sense (just not obvious to me). What do you think? - Menelaos > On Jun 3, 2025, at 4:27 PM, Boumalhab, Chris <cboum...@amazon.com> wrote: > > I think something like this could work: > theta_sketch_agg(col) to build the sketch > theta_sketch_union(sketch1, sketch2) to union the sketches > theta_sketch_estimate(sketch) or theta_sketch_estimate_count(sketch) to > estimate count > … > > Something similar can be done for tuple support. > > Let me know what you think. > > Chris
