5*2^n-1 undersieved in Psieve files, n>3M

[shanecruise]

plus i tried sr1seive on one of the 3M range in 5*2^n-1 drive, and could remove 10 factors in 5 minutes

shouldn't we consider updating them.

Also is this sr1sieve great for mersenne prime? k=1

[Kosmaj]

Regarding 5*n^2-1, can you tell us which file have you tried to sieve, and what factors have you found? Those test files are not the latest version, but they are fairly new, from a sieving file released a few months ago.

[shanecruise]

@kosmaj here is an example

[shashank@iota sr1sieve-1.4.1-bin]$ ./sr1sieve-x86_64-linux -i 3170to3172.txt -o 3170-72new.txt -P 10000000000
sr1sieve 1.4.1 -- A sieve for one sequence k*b^n+/-1.
Read 96 terms for 5*2^n-1 from NewPGen file `3170to3172.txt'.
sr1sieve 1.4.1 started: 3170014 <= n <= 3171990, 1000000000 <= p <= 10000000000
1043838949 | 5*2^3170044-1
1560478571 | 5*2^3170682-1
2041977359 | 5*2^3170070-1
4315418641 | 5*2^3171180-1
8725797991 | 5*2^3170714-1
8788835771 | 5*2^3171554-1
p=8788953629, 45064789 p/sec, 6 factors, 86.5% done, 28 sec/factor
sr1sieve 1.4.1 stopped: at p=10000000000 because range is complete.
Wrote 90 terms for 5*2^n-1 to NewPGen file `3170-72new.txt'.

Note "28sec/factor"
I could sieve more
_________________________
more

sr1sieve 1.4.1 started: 3170014 <= n <= 3171990, 10000000000 <= p <= 30000000000
11785244819 | 5*2^3171588-1
15222573571 | 5*2^3171614-1
16227917989 | 5*2^3171194-1
21866270731 | 5*2^3170634-1
22354234291 | 5*2^3170344-1
28403730361 | 5*2^3171214-1
29964540439 | 5*2^3170980-1
p=29964625319, 45712439 p/sec, 7 factors, 99.8% done, ETA 14 Sep 15:40

[Thomas11]

Originally Posted by shanecruise

@kosmaj here is an example

[shashank@iota sr1sieve-1.4.1-bin]$ ./sr1sieve-x86_64-linux -i 3170to3172.txt -o 3170-72new.txt -P 10000000000
sr1sieve 1.4.1 -- A sieve for one sequence k*b^n+/-1.
Read 96 terms for 5*2^n-1 from NewPGen file `3170to3172.txt'.
sr1sieve 1.4.1 started: 3170014 <= n <= 3171990, 1000000000 <= p <= 10000000000
1043838949 | 5*2^3170044-1
1560478571 | 5*2^3170682-1
2041977359 | 5*2^3170070-1
...

That's indeed interesting. Thanks for pointing this out, Shanecruise!

Verification with Pari shows that these are indeed proper factors.
So, most probably those factors were somehow lost in the initial stage of the sieve file preparation.
Is anyone in contact with the PrimeSearch team?

[shanecruise]

seive files are only till 1 Billion , we can easily do 100 Billion in less than 30 minutes. I have only extended it.

same goes for 8th drive.

I guess we need to seive deeper and there's no need to give credit to pSeive anymore, its outdated

[Thomas11]

I just putted all k<300 into one file and ran sr2sieve on it.
It seems that only k=5 is affected:

./sr2sieve -i sr_2.abcd -p 1000000000 -P 2000000000
sr2sieve 1.8.9 -- A sieve for multiple sequences k*b^n+/-1 or b^n+/-k.
Read 4180903 terms for 148 sequences from ABCD format file `sr_2.abcd'.
Split 148 base 2 sequences into 3117 base 2^60 subsequences.
Expecting to find factors for about 135315.90 terms in this range.
sr2sieve 1.8.9 started: 3000000 <= n <= 4000000, 1000000000 <= p <= 2000000000
1000682041 | 5*2^3740864-1
1000968181 | 5*2^3237128-1                                   
1000982629 | 5*2^3139532-1
1001250689 | 5*2^3529918-1                                        
1002273931 | 5*2^3277698-1                                    
1002291839 | 5*2^3453994-1
1002751229 | 5*2^3602168-1                                        
1002788879 | 5*2^3580938-1                                    
1003213439 | 5*2^3700772-1                                        
...

The missing factors are starting from p=1,000,000,000. I haven't found any below this.

[Thomas11]

Originally Posted by shanecruise

seive files are only till 1 Billion , we can easily do 100 Billion in less than 30 minutes. I have only extended it.

same goes for 8th drive.

Those indicators, e.g. "1000000000:M:1:2:258", are just placeholders. This wasn't updated by the PrimeSearch team.

The 8th drive shouldn't be affected. At least I hope so...

[shanecruise]

ok yes you are right only k=5 is affected. others look ok

so we can update that one.

[Kosmaj]

Hi guys,
can you try to sieve the enclosed file, it's the latest Psieve version of July 27.

Thanks!

[Kosmaj]

Sorry, no need to sieve, it's exactly the same file like the one on the server with all those composites.

How about k=7, I tried a little but found no factors from 1-4E9. Thanks!

[Thomas11]

Originally Posted by Kosmaj

Sorry, no need to sieve, it's exactly the same file like the one on the server with all those composites.

How about k=7, I tried a little but found no factors from 1-4E9. Thanks!

I tried all k<300 and only k=5 seems to be affected. See my earlier posting.
Seems that the composites/factors file starting from p=1B was lost.

Meanwhile I tested the n=4-5M sieve file and there are also missing factors:

./sr1sieve -i t17_b2_k5.npg -o test.txt -p 1000000000 -P 2000000000
sr1sieve 1.4.1 -- A sieve for one sequence k*b^n+/-1.
Read 42582 terms for 5*2^n-1 from NewPGen file `t17_b2_k5.npg'.
sr1sieve 1.4.1 started: 4000004 <= n <= 4999954, 1000000000 <= p <= 2000000000
1000077409 | 5*2^4085350-1
1000636969 | 5*2^4123310-1                                    
1000706821 | 5*2^4056718-1                                          
1000721539 | 5*2^4950400-1
1000920451 | 5*2^4037194-1                                      
1001099629 | 5*2^4141240-1

Thus, I suggest that we put k=5 on hold until we've investigated this a little further.
I will put the whole n=3-6M range for k=5 into one file and start sieving from p=1B.
Perhaps someone else could start at higher levels and try to locate the upper bound of the missing factors file. Maybe only p=1B-1T is lost...