Hi, I thought averaging strings was the non-trivial case. I knew (x+y)/2 for averaging integers would have overflow problems and not work for all inputs. It turns out there is a short but non-trivial expression for averaging signed/unsigned integers for 'all' inputs. I've also added Mean*() for 32/64 bit signed/unsigned integers with comprehensive tests to sixb <https://github.com/jfcg/sixb> v1.3.0.
Let me know if you identify an input that breaks any of these functions. Cheers.. On Monday, November 1, 2021 at 8:47:03 PM UTC+3 jfcg...@gmail.com wrote: > Hi, > > I wrote MeanStr(s1,s2) <https://pkg.go.dev/github.com/jfcg/sixb#MeanStr> > to calculate lexicographic average of two strings. It works fine with > ascii strings but I want to get feedback on general (utf-8) inputs. It > should satisfy: > > - a good average of two inputs (in some context): For most/all s (in > that context) with s1 < s < s2, s is supposed to be < MeanStr(s1,s2) > about half the time > > > - MeanStr(s1,s2) = MeanStr(s2,s1) > > > - For s1 < s2, s1 <= MeanStr(s1,s2) < s2 > > > - MeanStr(s,s) = s for all s > > Can you take a look at the function and try to find (utf-8 or not) inputs > that break these requirements? > > Thanks.. > > Note: Ordering is regular <, not any fancy unicode order. The function > considers the first 31 bytes. > Note: I am planning to switch from median-of-2n+1 pivot calculation to > (arithmetic) median-of-2n in sorty <https://github.com/jfcg/sorty>, which > is cheaper to calculate and better pivot (smaller variance than > median-of-2n+1) > -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/34ab60ad-20ea-406d-8f33-01f8892b981dn%40googlegroups.com.
