Yeah, that was one of the reasons we chose it as one of the ones to implement, seemed like that was a really interesting tradeoff, we also used uddsketch[1] which provides relative error guarantees, which is pretty nifty. We thought they provided different enough tradeoffs that we wanted to implement both.
Hi, in an unrelated nitpick: The relative error should be calculated by dividing the error by the true value, not by it's approximation. Still, very nice writeup!
Yeah, that was one of the reasons we chose it as one of the ones to implement, seemed like that was a really interesting tradeoff, we also used uddsketch[1] which provides relative error guarantees, which is pretty nifty. We thought they provided different enough tradeoffs that we wanted to implement both.
[1]: https://arxiv.org/abs/1908.10693