## Other

drummers-lowrise

Message boards : General discussion : Generalized Fermat Progression Search

 Post to thread Subscribe SortOldest firstNewest firstHighest rated posts first
Author Message
Kellen

Joined: 10 Jan 18
Posts: 457
ID: 967938
Credit: 1,339,357,863
RAC: 0

Message 142428 - Posted: 13 Aug 2020 | 13:15:34 UTC

Greetings All,

Following some discussions on the GFN1-12 Discord server about values of b for which there are likely no primes of the form b2n+1, we turned to the opposite case where a given b value may result in a long progression of prime values with increasing n.

The known minimum values for which all numbers of the form b2n+1 are prime from n=0 to n=a were as follows, with no further terms know:

a(0): b=2
a(1): b=2
a(2): b=2
a(3): b=2
a(4): b=2
a(5): b=7072833120
a(6): b=2072005925466

Our resident GFN wizard, Yves, wrote a series of programs to efficiently sieve and test b values to search for the a(7) term and we are happy to announce that the search was successful! After ~ 75 CPU-days, the following term was found, and confirmed to be the lowest b for which b2n+1 is prime for all n from 0 to 7.

a(7): b=240164550712338756

So, with that, we would love to find a solution for n=8, however we have reached a point where a small search is unlikely to find a solution for n=8, as that is a few orders of magnitude more difficult. Solutions for n=9 or n=10 are each that much more difficult again. However; these are not outside of the range of a distributed project, and the computational effort required for n=10 is comparable to that of the AP27 search, especially if a GPU implementation is created.

This is an exciting revival of an old search, and I hope that we are able to add more to the maximum progression length in the future!

Regards,
Kellen

JeppeSN

Joined: 5 Apr 14
Posts: 1378
ID: 306875
Credit: 21,623,822
RAC: 0

Message 142431 - Posted: 13 Aug 2020 | 15:12:52 UTC - in response to Message 142428.

Really nice, and I see you are adding it to https://oeis.org/A090872.

Well done by Yves!

Here are the eight primes (not that I think the readers are idiots, just to celebrate them):

240164550712338756 + 1
240164550712338756^2 + 1
240164550712338756^4 + 1
240164550712338756^8 + 1
240164550712338756^16 + 1
240164550712338756^32 + 1
240164550712338756^64 + 1
240164550712338756^128 + 1

The last two are so-called titanic primes (at least 1000 decimal digits).

Would be crazy if a longer chain of this sort could be found.

/JeppeSN

dannyridel
Volunteer tester

Joined: 3 Feb 19
Posts: 791
ID: 1097922
Credit: 3,783,734
RAC: 0

Message 142432 - Posted: 13 Aug 2020 | 15:28:14 UTC - in response to Message 142431.
Last modified: 13 Aug 2020 | 15:28:56 UTC

Really nice, and I see you are adding it to https://oeis.org/A090872.

Well done by Yves!

Here are the eight primes (not that I think the readers are idiots, just to celebrate them):

240164550712338756 + 1
240164550712338756^2 + 1
240164550712338756^4 + 1
240164550712338756^8 + 1
240164550712338756^16 + 1
240164550712338756^32 + 1
240164550712338756^64 + 1
240164550712338756^128 + 1

The last two are so-called titanic primes (at least 1000 decimal digits).

Would be crazy if a longer chain of this sort could be found.

/JeppeSN

"Fermat-ic" Progression of n primes :D lol

Also:
This is an exciting revival of an old search, and I hope that we are able to add more to the maximum progression length in the future!

Revival?
____________
SHSID Electronics Group
SHSIDElectronicsGroup@outlook.com

GFN-14: 50103906^16384+1
Proth "SoB": 44243*2^440969+1

Kellen

Joined: 10 Jan 18
Posts: 457
ID: 967938
Credit: 1,339,357,863
RAC: 0

Message 142434 - Posted: 13 Aug 2020 | 15:37:45 UTC - in response to Message 142432.

This is an exciting revival of an old search, and I hope that we are able to add more to the maximum progression length in the future!

Revival?

Yep! https://www.primepuzzles.net/puzzles/puzz_137.htm. Most recent term added May 10, 2007 after Yves discovered the first 6 :)

Michael Goetz
Volunteer moderator

Joined: 21 Jan 10
Posts: 13513
ID: 53948
Credit: 236,922,854
RAC: 0

Message 142435 - Posted: 13 Aug 2020 | 16:05:10 UTC - in response to Message 142432.

dannyridel wrote:
"Fermat-ic" Progression of n primes :D lol

I was thinking they could be called "Gallot Numbers" (or "Kellen Numbers"? I don't know the history of these), but Fermatic Progression is actually kind of good.

____________
My lucky number is 75898524288+1

Kellen

Joined: 10 Jan 18
Posts: 457
ID: 967938
Credit: 1,339,357,863
RAC: 0

Message 142436 - Posted: 13 Aug 2020 | 16:08:55 UTC - in response to Message 142435.

dannyridel wrote:
"Fermat-ic" Progression of n primes :D lol

I was thinking they could be called "Gallot Numbers" (or "Kellen Numbers"? I don't know the history of these), but Fermatic Progression is actually kind of good.

We have been calling them "GFP" for short, but if they were going to get a named series I vote for Gallot Numbers :)

Yves = Interested in GFP + Big Brain + All the hard work
Kellen = Interested in GFP + Lots of CPU cores ;)

Yves Gallot
Volunteer developer
Project scientist

Joined: 19 Aug 12
Posts: 644
ID: 164101
Credit: 305,010,093
RAC: 0

Message 142438 - Posted: 13 Aug 2020 | 16:57:01 UTC - in response to Message 142436.

GFP is great, because it can be Generalized or Gallot ;-)

If we read Fermat's letter, this progression is in the same vein.
He built the sequence 21, 22, 23, ... and added one. He proved that they are composite if the exponent is not in the sequence (i.e. a power of two) and he thought that the other ones were primes because he didn't find their divisors.
Here the sequence is b1, b2, b3, ... Similarly bn + 1 is composite is n is not a power of two. We try to replace 3, 5, 17, 257 with a longer sequence of primes.

robish
Volunteer moderator
Volunteer tester

Joined: 7 Jan 12
Posts: 1774
ID: 126266
Credit: 5,066,569,935
RAC: 0

Message 142439 - Posted: 13 Aug 2020 | 17:03:37 UTC

I agree, Gallot numbers, really interesting and incredibly cool. ðŸ˜Ž
____________
My lucky numbers 10590941048576+1 and 224584605939537911+81292139*23#*n for n=0..26

Kellen

Joined: 10 Jan 18
Posts: 457
ID: 967938
Credit: 1,339,357,863
RAC: 0

Message 142591 - Posted: 19 Aug 2020 | 0:39:08 UTC

Greetings Folks,

Lots of stuff going on with this search still, so it is probably a good time for an update.

The nomenclature has been sorted out a little more officially, as it can be a little confusing with the progressions starting from b20. We have settled into "GFP-" followed by a number indicating the length of the progression, rather than the maximum value of n in the final term of the progression.

To use the original find as an example:

24016455071233875620+1 (240164550712338756 + 1)
24016455071233875621+1 (240164550712338756^2 + 1)
24016455071233875622+1 (240164550712338756^4 + 1)
24016455071233875623+1 (240164550712338756^8 + 1)
24016455071233875624+1 (240164550712338756^16 + 1)
24016455071233875625+1 (240164550712338756^32 + 1)
24016455071233875626+1 (240164550712338756^64 + 1)
24016455071233875627+1 (240164550712338756^128 + 1)

This is a GFP-8, indicating a progression of length 8, despite the final term being b27.

With that sorted out; on to some real news.

Yves made specialized versions of his GFP programs to search for GFP of different lengths, and computed the first 1000 b values which result in a GFP-5, the first 148 b values which result in a GFP-6 and the first set of GFP-7 b values. Rob computed the remaining GFP-7 b values so that we have the first 101 known, and I finished off GFP-6 to 1000 values.

These sequences, or updates to existing sequences, have now been submitted to OEIS by JeppeSN and are in various stages of approval. They can be found at the following links:

GFP-5 (i = 0...4): https://oeis.org/A070694
GFP-6 (i = 0...5): https://oeis.org/A235390
GFP-7 (i = 0...6): https://oeis.org/A335805

The comments have not been approved yet, so they are not showing up, but the astute will notice that GFP-5 term #173 corresponds to the first GFP-6 and GFP-6 term #148 corresponds to the first GFP-7. We are working to determine the total number of GFP-7 to the first GFP-8, however the computation involved is significant and this will take some time.

The GFP-8 search also continues and should be completed to software limits (b=264) in the next few weeks. There should be ~5 more GFP-8 in this range, so there will be a new OEIS sequence shortly for those as well!

From here the search gets significantly more involved as sieve efficiency and search complexity, which have increased relatively uniformly, decouple. The search for the first GFP-9 term is a substantial undertaking (estimate of 275,000 core-days, compared to the 75 it took to find the first GFP-8).

We will see where this takes us and report back with news and progress updates when things happen :)

Regards,
Kellen

robish
Volunteer moderator
Volunteer tester

Joined: 7 Jan 12
Posts: 1774
ID: 126266
Credit: 5,066,569,935
RAC: 0

Message 142766 - Posted: 29 Aug 2020 | 0:16:29 UTC
Last modified: 29 Aug 2020 | 0:17:23 UTC

For those who are interested, the search continues :)

Progress so far .....

GFP-8 progressions:
1
240164550712338756
3686834112771042790
6470860179642426900
7529068955648085700
10300630358100537120
16776829808789151280
17622040391833711780
19344979062504927000

More coming soon ;)

GFP-9 May require assistance however, on a Boinc level. :)
____________
My lucky numbers 10590941048576+1 and 224584605939537911+81292139*23#*n for n=0..26

Message boards : General discussion : Generalized Fermat Progression Search