undefined | Better HN

0 pointsced5y ago0 comments

Yeah, but this version is not necessarily optimal. If you have 4 positive samples, then you will test 2+2+1+1+1+1 = 8, instead of 4 in the naive case.

It feels like a problem for information theory.

0 comments

bequanna5y ago

I didn’t say it was optimal, but it is a huge improvement on doing 64 tests for 64 samples.

cedOP5y ago

Oh, sure!

j / k navigate · click thread line to collapse