--- In primeform@yahoogroups.com, "Jens Kruse Andersen"
<jens.k.a@...> wrote:
>
> Mike Oakes wrote:
>
> > Here is a new AP6 record at 1503 digits:-
> >
> > (27477652+n*3603821)*3529#+1 is prime for n=0..5
>
> Congratulations!
>
> I have set two far easier records.
> 100-digit AP12: (53660216 + 27243281*n)*229#+1, n=0..11
> 64-digit AP13: (208988611 + 32224644*n)*139#+1, n=0..12
>
> Prp'ing by the GMP library and proofs by PARI/GP.
> http://hjem.get2net.dk/jka/math/aprecords.htm is updated.
>
> --
> Jens Kruse Andersen
>
... Congratulations! I have set two far easier records. 100-digit AP12: (53660216 + 27243281*n)*229#+1, n=0..11 64-digit AP13: (208988611 + 32224644*n)*139#+1,...
Small mistake on my part too, I'm afraid (one really shouldn't initialize an email using CTRL-C+CTRL-V...):- For "APSieve" read "NewPGen". Thanks for updating...
... Congrats on those! I would be really interested in some details and run-times of your algorithm for finding APn's. Your methods are obviously very...
... Smallest known difference and end for an AP23. ... Not a record after the former AP23 is known. ... This is the largest known AP22 from March 15: ...
... Thanks. ... All my largest known AP records are on the form a*p#+1 for an AP of a's. I haven't searched AP lengths above 19 with this program. For AP sizes...
... Thanks for that description, Jens. Apart from the first and last of these features, that is precisely my detector too (it's exhaustive, but doesn't try to...
http://primes.utm.edu/primes/page.php?id=77458 ... Perhaps Mike needs to submit the other 5, even though those supporting primes are unlistable? David...
... If it's exhaustive then it may spend time testing different parts of the same potential AP of target length. No cache optimization is bad in a program...
... The beauty of working in terms of APs which straddle a particular point (the half-way index) is that this can be avoided. I start only from an AP2 which...
... Jens: if you don't mind, can I probe a bit here? The question is important, if we are to home in on the optimal coding. I have stored all the candidate k...
... Yes, table=array. ... I have limited knowledge of how cache actually works. If MB's of other data has been read from ram since the next value of q was last...
My algorithm is several orders of magnitude faster then what you guys have/propose, and the more primes you throw at it the faster it gets. In the simplest...
Jens, The Athlon XP's, to my experience have two flavors: 256kb and 512kb of L2 cache. I suspect the later (final run) of XP's were pushing 1mb of L2. The...
... I have 256kB. ... I don't even know what such tables do (but would know how to Google it). ... That's not necessary. I think my AP detector mostly avoids...
... Maybe in an AP25 search but I would use another algorithm there - and did so in a short AP23 search. The AP detection for my 48-digit AP15 record with...
... Marcus [I agree with the points Jens has already made, and wish to make just one more here.] For APn's with n >= 20 your algorithm obviously has...
... Which is why I concentrate on that end of the spectrum It doesn't need efficient code to find an AP5 amongst 200000 prps (but it takes a bit of time to...
Mike, I think your points are valid for sub AP11 but not above AP12 or so. If using 101#+1 for example you can gain a nice speed improvement if the ranges...
... If you sieve a*101#+1 for a = 1..10^9, with NewPGen, you can reduce these to about 2*10^8 sieved candidates without too much trouble. And of these, about...
... these to about 2*10^8 sieved candidates without too much trouble. And of these, about 8.7*10^7 will be prime, i.e. 1 in 2.3. ... the 15 numbers has only a...
... these to about 2*10^8 sieved candidates without too much trouble. And of these, about 8.7*10^7 will be prime, i.e. 1 in 2.3. ... the 15 numbers has only a...
... The density is far from double. My 64-digit AP13 record is at a*139#+1. The prp density was 1 in 16.30. 16.30/5.79 = 2.815, and 2.815^13 = 697000 =...
... PLEASE can we compare like with like! When I read this, I took it you were claiming to be 4* faster. GHz-hrs is the unit to use, not hours. Mike...
Here is a new AP6 record at 1606 digits:- (32649185+n*3884057)*3739#+1 is prime for n=0..5 All confirmed prime with PFGW -t Input/output statistics:- Numbers...
Here is a new AP6 record at 1606 digits:- (32649185+n*3884057)*3739#+1 is prime for n=0..5 All confirmed prime with PFGW -t Input/output statistics:- Numbers...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
Offline 
Send Email