20080506, 01:56  #34 
May 2007
Kansas; USA
10101101111111_{2} Posts 
I've sent an email to michaf asking for his status on his Sierp base 3 reservation and his base 24 reservations on both sides.
Gary 
20080506, 03:03  #35 
A Sunny Moo
Aug 2007
USA (GMT5)
3·2,083 Posts 

20080506, 03:45  #36 
May 2007
Kansas; USA
5·17·131 Posts 

20080506, 04:00  #37 
Mar 2003
New Zealand
13·89 Posts 
Here is a rough diagram showing where I think the different methods for sieving k*2^n+/1 are efficient.
Code:
many k values ^  NewPGen same number of (fixedn) k and n values  /  /  /  /  /  /  /  gcwsieve  /  /  /  /  / srsieve  / sr2sieve proth_sieve/JJsieve  trial (SPH algorithm)  factoring / NewPGen (fixedk) sr1sieve +> many n values For an extreme case like the initial stages of Base 3 Sierpinski/Riesel, the ideal sieve would probably be found in the area above the diagonal. This would be a sort of fixedn sieve that works with multiple nvalues. 
20080506, 04:41  #38 
A Sunny Moo
Aug 2007
USA (GMT5)
3×2,083 Posts 

20080506, 05:43  #39 
Quasi Admin Thing
May 2005
977 Posts 
@Gary, you can sure change it, maybe I should just bring it up to n=100,000, so you can do that. No problem at all Good luck on getting your pages done. I'll check back later tonight, if no one has taken the sierpinski gap and reserved it, I'll throw in my last free ressoruce and start working on that range, but of course I'll tell in the base 3 forum weather or not I take it

20080506, 11:36  #40 
Quasi Admin Thing
May 2005
3D1_{16} Posts 
Reenforcement has arrived. 1 quad Q6600 has been thrown into the war. So for now work seems to be done in this way:
1. No reservation of sierpinski base 3 2. Test all sierpinski base 12 candidates up to n=250,000 or prime found 3. Test all riesel base 27 candidates up to n=1,000,000 or prime found 4. Test all sierpinski base 19 candidates up to n=100,000 or primes found 5. Continue on the base 3 riesel conjecture Regards KEP 
20080506, 19:44  #41  
May 2007
Kansas; USA
5·17·131 Posts 
Quote:
With the exception of Siemelink's recent primes on Riesel base 19, the CRUS pages should be fully up to date. Gary 

20080506, 19:49  #42  
May 2007
Kansas; USA
5·17·131 Posts 
Quote:
That's an interesting graph, Geoff. Can you clarify your last statement? Do you mean a fixedn sieve that works with multiple kvalues or a fixedk sieve that works with multiple nvalues? So based on this, what software do you feel would work best for sieving base 3? It almost looks like srsieve sieving many k's across a somewhat narrower nrange because there are SO many k's for base 3 and it's such a prime base. But I know sr2sieve has been improved a lot to handle many k's. Also, you said gcwsieve could be generallized. Does that mean that the program can be changed so that k does not have to equal n? Thanks for the info. Gary Last fiddled with by gd_barnes on 20080506 at 19:51 

20080509, 03:00  #43  
Mar 2003
New Zealand
13×89 Posts 
Quote:
Quote:
Quote:
Having said that I should make it clear that I am not planning to write any such programs in the near future, I don't have the time. 

20080509, 04:02  #44  
May 2007
Kansas; USA
5·17·131 Posts 
Quote:
Now I get it. Speaking of the fixedn sieve with multiple nvalues, it makes me wonder if using NewPGen with the increment counter set on would work well for this. That's what I use for my 'all twin prime search' where I sieve all k<1M for 1000 nvalues at a time. It's not extremely efficient but it is the most effective way I know of for twin prime searching. I think I'll play around with NewPGen here. The fact that it's such a prime base with so many k's may make it a possibility. I never thought I'd say that about NewPGen again except for special multipleprime forms such as twins, SG's, etc. But it may make sense, as KEP has alluded to, to use PFGW to a very small nvalue (perhaps n=2500 or less) and leave thousands or even millioins of k's remaining and then use NewPgen with the increment counter set on to sieve a large range of those k across one nvalue at a time. With base 3 being such a prime base, that might be a possibility. For that matter, perhaps we shouldn't use PFGW trial factoring at all! Maybe we could sieve billions of k's at once starting at n=1 with NewPgen's increment counter and go up from there. Of course, then you get into the issue of huge file sizes requiring more manual intervention, the fact that NewPGen (or any sieving software for that matter) doesn't actually do prime searching so it continues sieving k's after primes would have been found, etc. Thanks for your input. Gary Last fiddled with by gd_barnes on 20080509 at 04:04 

Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Bases 251500 reservations/statuses/primes  gd_barnes  Conjectures 'R Us  2462  20220812 15:32 
Bases 33100 reservations/statuses/primes  Siemelink  Conjectures 'R Us  1729  20220805 15:35 
Bases 101250 reservations/statuses/primes  gd_barnes  Conjectures 'R Us  989  20220626 08:35 
Riesel base 3 reservations/statuses/primes  KEP  Conjectures 'R Us  1134  20220624 20:28 
Bases 632 reservations/statuses/primes  gd_barnes  Conjectures 'R Us  1417  20220525 19:33 