k=96880162584451340130465
 

I am worried that I cannot keep up with the TPS 333333 search (as far as finding primes for the top 5000) in my search for the most prime series k*2^n+1.

The most found is 166 primes for k=96880162584451340130465 and the last one found was at n=243709. I am currently at n= 277833, just enough to keep my head above the 5000 waterline.

Maybe this would form the basis for a new drive, but please appreciate that I don't want to take resources from other drives.

What do people think?

Regards

Robert Smith

People like to search for small primes on the side, and that's not a problem. But we (RPS) declared that we are searching for Riesel k*2^n-1 primes, note "-" not "+" and a number of people around Top-5000 are strict about project definitions. Therefore, we cannot report Proth ("+") primes as found by RPS.

Of course, if somebody wants to help you he is welcome.

BTW, do you have an attractive k for the k*2^n-1 form?

Ahhh, I forgot - fair comment Kosmaj

Phil Carmody would have the details of the best k in k*2^n-1, he was working with someone whose name escapes me, who found a very prime series.

Encouraged by Geoffrey's srsieve/sr1sieve, which provide a fast sieve also for k>2^31, I already considered a new run for heavy weight Ks. I'm currently generating a new weights list (looking only at "15Ks" at the moment).

Here are some preliminary results (for k up to about 2^34):
[CODE] k Nash weight
------------------------
7924279935 7420
2889081195 7261
16260063105 7231
12528535305 7227
15065044995 7225
10956100155 7220
16189312425 7208
7548763365 7175
6823405875 7133
6669596505 7126
11314525365 7120
3428677395 7102
11941744815 7101
6291634635 7093
4142541975 7066
1650292215 7057
6714590025 7057
16733391675 7049
14279404425 7042
2861741025 7037
6356924145 7029
14583220575 7027
17587779495 7024
6060309255 7013
13836444765 7008
1748348745 7006
13141001445 7005[/CODE]

For comparison: The "best" (Riesel-)k I know about is Robert's
k=5326992948092709915443565 (p106) having a Nash weight of 8223!

I already started sieving k=7924279935 to get some "feeling" about running srsieve/sr1sieve on the heavy weights. Please feel free to take any of the other k-values (but check Larry's list somewhere here in the forum, since some of the k<2^32 may already being tested...)

It will be fun to try a huge k, like the one Robert suggested above. LLR slows down a little (10-20% ?) in the vicinity of k=2^50 (or was that 2^48 ?) but we can expect a high weight. Additionally if we can come up with
k = k0 * k1^2,
where k0 < 16, then srsieve will be very fast.

<snip>
[I]wrong reservation[/I]

[QUOTE=CedricVonck;99743]I will try to attack 2889081195[/QUOTE]

This k seems already been tested up to n=27000 resp. n=140000.
Have a look at [URL="http://www.15k.org"]http://www.15k.org[/URL] for more information...

Thomas,

Then I choose k=16260063105 for a round of sieving.

BTW, can you give instructions on how to sieve this?
I donwloaded sr1sieve from Geoff but following code don't seem to work:

[quote]
sr1sieve-pentium4.exe -i test.txt -o test.txt --pmax 100000000000 -f factors.txt --save 60 --verbose k="16260063105*2^n-1"
[/quote]

I tried also
[quote]
-n 0 -N 1e6
[/quote]

It did not work either.
I will try with sr2sieve => no dice either

Thank you

[QUOTE=CedricVonck;99749]
BTW, can you give instructions on how to sieve this?
Thank you[/QUOTE]

You have to start with [B]srsieve[/B] first, because for (the faster) [B]sr1sieve[/B] pmin needs to be larger than k.

Start with:
[CODE]srsieve -N 100000 "16260063105*2^n-1"[/CODE]
(-N specifies nmax)

You may stop it immediately and restart it as follows:
[CODE]srsieve -q srsieve.out[/CODE]
This causes it to run in "quiet" mode, otherwise the sceen output would slow down the process...

You need to run it up to at least pmax=k by:
[CODE]srsieve -q -P 16300000000 srsieve.out[/CODE]
You may replace the "-q" switch by "-v" once the elimination rate is below 1 n/sec to watch the progress...

To turn over to [B]sr1sieve[/B] you need to convert the [B]srsieve.out[/B] file into a NewPGen file by [B]srfile[/B] (part of the srsieve package):
[CODE]srfile -g srsieve.out[/CODE]
You'll get a file called "t17_b2_k16260063105.npg", which can be sieved further by sr1sieve by:
[CODE]sr1sieve -i t17_b2_k16260063105.npg -o t17_b2_k16260063105.npg -f factors.txt -P 100000000000[/CODE]
(setting pmax to 100 billion, for example)
You may skip the "-f factors.txt" switch if you don't want to keep track of the factors...

There might be some additional information in this thread:
[URL="http://www.mersenneforum.org/showthread.php?t=6968"]http://www.mersenneforum.org/showthread.php?t=6968[/URL]

Thank you for the info.
I will get started as soon as I am home!

srsieve
 

If someone could please direct me to a win32 or windows based version of srsieve, it would be very much appreciated!

Roger