On Sat, Aug 5, 2017 at 3:51 AM, Steve D'Aprano <steve+pyt...@pearwood.info> wrote: > On Sat, 5 Aug 2017 02:31 am, MRAB wrote: > >> Why would isqrt_float not give the correct answer? Probably because of >> truncation (or rounding up?) of the floating point. I'd expect it to >> fail first near a square. > > Assuming your math.sqrt() is an IEEE-754 correctly-rounded square root
To clarify: What you're assuming here is that the floating-point value that math.sqrt returns is exactly the value that you get by rounding the true square root (even if transcendental) to the available precision. I *believe* that that's guaranteed for a floating-point input value, but is it for integer input? Is there any way that (say) 2**53+1 could sqrt to a different value depending on whether you convert the input to float first? ChrisA -- https://mail.python.org/mailman/listinfo/python-list
