Dear Dr. Broadhurst, may I join the AP30 club?
30 terms; delta = 210*236810 = 210*(2*5*7*17*199)
All either have tau=4, or have cofactors that fool GMP-ECM's PRP test. David's recommended inclusion of 11 pays off. My tests with inclusion of 7 also seem to be _less_ successful. i.e. 210 seems the best value for money.
And I thought that my sieve-like mind would make this task easy... It's had ~30 hours of AMD GHz so far, which seems to make me not noticably more productive than anyone else. However, I'm able to _sieve_ a range in about 0.1s, it's the AP finding that's the killer.
I can't quite work out my big-Oh, but it seems to be O(r^2/l&2), where r=range, l=length. For n<<r, n=number of elements, it appears the more traditional O(n^2/l) may be superior.
Tally so far:
AP29: 2 (from *30) + 4+2 (from *210) + 1 (from 2310) = 9
AP30: 1 (from *210)
I realise I've just mailed Anna the wrong command-line, and have probably just got her to overwrite the entirity of last-night's logging :-(, so I can't support some of those claimed AP29's with adducements.
Don't be fooled, CRC Press are _not_ the good guys.
They've taken Wolfram's money - _don't_ give them yours.
Find the best deals on the web at AltaVista Shopping!