Talk:Sieve of Eratosthenes: Difference between revisions

 

::: I agree with you on the last point. Re the digression: on to ''a'' vs. ''b'': the naive way to generate primes from ''a'' to ''b'' is to generate to ''b'' and then ignore all the ones earlier than ''a''. Hence to ''b''. The main point being that we don't make an array for all values from ''0'' to ''b'', but just ''a'' to ''b'', plus intermediate storage (to ''sqrt(b)'' for a full sieve). I suppose the right way to do the task is to let the implementers show multiple ways to do it. You are correct on the distinction between segment and segmented. As my text is written the ''a'' to ''b'' range could be done as a single block, hence it is a segment or offset and is a different concept from a segmented sieve. Certainly a bounded range can be done as a segmented sieve and for large ranges they are much faster. Easily allowing non-zero starting points is a bonus. [[User:Danaj|Danaj]] ([[User talk:Danaj|talk]]) 11:17, 11 September 2014 (UTC)