20130906, 20:59  #2  
"Åke Tilander"
Apr 2011
Sandviken, Sweden
2·283 Posts 
Quote:
Congratulations to the prime! Which is also first MMp factor prime on the list of top 5000! :) Last fiddled with by aketilander on 20130906 at 21:01 

20130906, 21:08  #3 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2
9,629 Posts 
It certainly is not an MMp factor, though. (This has been checked.)

20130906, 21:54  #4 
Account Deleted
"Tim Sorbera"
Aug 2006
San Antonio, TX USA
10256_{8} Posts 

20130907, 05:07  #5  
"Åke Tilander"
Apr 2011
Sandviken, Sweden
2·283 Posts 
Quote:
MMp factorish prime MMp factor class prime MMp possible factor prime etc. but none was really good so I simply choose the simplest. It would be good though if we all could agree on a common way of labeling these primes. Maybe we could first collect a number of possibilities and then conduct a vote? Many kind of primes seem to be labeled after the one who "discovered" them or made the significant mathematical analysis ot them, Riesel, Mersenne etc., others seem to have their name from a significant mathematical property like factorial, home etc Last fiddled with by aketilander on 20130907 at 05:20 

20130907, 07:06  #6 
"Åke Tilander"
Apr 2011
Sandviken, Sweden
2·283 Posts 
One other thing Batalov: Have you tested all other possible smaller K:s (< 507568) for M1398269 or was this just a storke of luck?

20130907, 07:45  #7 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2
9,629 Posts 
I haven't checked all possible k's.
I did however check all k values that could produce MMpdivisors, i.e. k 0, 5, 8, 9 (mod 12). That is roughly half of them. I did not check k 2, 3, 6, 11 (mod 12). 
20130907, 09:22  #8 
"Åke Tilander"
Apr 2011
Sandviken, Sweden
1000110110_{2} Posts 

20130907, 09:23  #9  
Banned
"Luigi"
Aug 2002
Team Italia
11342_{8} Posts 
News from subproject Deep Sieving
Quote:
I have recorded all possible k for MM34 and MM35 up to k=500,000 thanks to Tony: I may clear some of them and extend the table limit if you still have your logs... Luigi Last fiddled with by ET_ on 20130907 at 09:26 

20130907, 18:47  #10  
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2
9,629 Posts 
Quote:
I sieved to >20G for k<2,000,000. In just a few hours, on one CPU. It is straightforward with Pari: for each prime q, k_{0} = Mod((q1)/2,q) / (Mod(2,q)^p1), then eliminate all k = k_{0} + mq from an array of k's. For sufficiently large q, this is not even a loop, but only k_{0} or nothing at all. In other words, I don't quite understand why you are so stuck on sieving: 1. Sieving alone is pointless without tests. 2. One should sieve only as long as sieving is faster than actually testing. I ran LLR for MM35 "divisor pool" (still running, approaching k~1,400,000). This was the only prime so far. This approximately matches the expectation: probability of being prime is O(1/p), and on top of that the probability of dividing MMp would be O(1/k) (unsure)? So, for large p, even finding a prime was relatively lucky, let alone a divisor of MMp. I also have sieve files for MM34, 36 and 37, but the 36/37 tests are five times slower, and a divisor of MM34 would soon be flushed out of Top5000, so I concentrated on attacking MM35. 

20130907, 19:08  #11 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2
9,629 Posts 
Here's the Pari interactive sieving script. It can be stopped periodically, the sieve array saved, and then the sieving can be continued.
Code:
# gp p 2000000 #sieve limit: M=2000000; #the P in MMp: P=1398269; allocatemem(800000000) K=vector(M); forprime(p=5,M, k=Mod((p1)/2,p)/(Mod(2,p)^P1);forstep(i=lift(k)+1,M,p,K[i]=1)); p=M; while(p=nextprime(p+1), if((i=lift(Mod((p1)/2,p)/(Mod(2,p)^P1)))<M,K[i+1]=1);K[1]=p); # can stop with CtrlC #write the sieve file write("si13y",P); write("si13y","#"K[1]); forstep(i=5,M1,[3,1,3,5],if(K[i+1]==0,write("si13y",i))) #go on p=K[1]; while(p=nextprime(p+1), if((i=lift(Mod((p1)/2,p)/(Mod(2,p)^P1)))<M,K[i+1]=1);K[1]=p); #etc, after a while stop with CtrlC, write, continue 
Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Deep Sieving MM49 in parallel  ET_  Operazione Doppi Mersennes  22  20160728 11:23 
Deep Hash  diep  Math  5  20121005 17:44 
The news giveth, the news taketh away...  NBtarheel_33  Hardware  17  20090504 15:52 
Question on going deep and using cores  MercPrime  Software  22  20090113 20:10 
Deep Sieving 10m Digit Candidates  lavalamp  Open Projects  53  20081201 03:59 