Anonymous user
Talk:Sequence of non-squares: Difference between revisions
An integer problem instead?
(An integer problem instead?) |
|||
Line 4:
:I suggest they choose/use the precision that gives correct results over the stated range, or reduce the range to suit the precision they have and add a comment. I think its true that most languages on Rosetta supporting floating point also support more than 32 bit FP values. --[[User:Paddy3118|Paddy3118]] 11:28, 24 August 2008 (UTC)
::Maybe it would be better to require an implementation not to use floating-point at all. I mean that m=floor(1/2 + sqrt(n)) should be solved iteratively using, say Newton method, in integers. That would guaranty that rounding error do not mangle the result. The case arises when sqrt(n) lands into the interval m.49999..m.50001. This is instable if the precision of floating-point numbers is 0.0001, because one cannot decide whether the floor of + 0.5 were m or m+1. Mathematically, for any given precision of floating-point numbers, there exist an infinite number of n, such that sqrt(n) is in the interval like above. There exist algorithms for integer sqrt(n) defined as the integer lower bound of real sqrt(n). One could modify them for 0.5+sqrt(n), that would be the cleanest way.
::I have just noticed that there is a typo error in the task name: "sequance" instead of "sequence". --[[User:Dmitry-kazakov|Dmitry-kazakov]] 13:30, 24 August 2008 (UTC)
== Zero ==
| |||