Primenet web design

[S485122]

Originally Posted by Uncwilly

...
Many sites have a desktop and also a mobile version.

The problem is those sites that (for economic reasons ?) have only one version for all clients.

[S485122]

Originally Posted by S485122

The problem is those sites that (for economic reasons ?) have only one version for all clients.

What I wanted to say :

The problem is those sites that (for economic reasons ?) have only the mobile version for all clients.

(I could have added : for instance having to scroll while there are less than 10 lines to display :-(

[James Heinrich]

Originally Posted by cxc

might it be possible to allow the Fermat exponents to appear under Reports > Detailed Reports > Exponent Status rather than being disallowed (as they are not primes)?

If you enter a 2^x exponent there you'll get a redirect link pointing you to the ECM Progress table. As far as I know, that's all the data that's tracked for Fermat numbers, I don't think there are any more detailed logs kept of who-what-when-how.

[cxc]

Hi James! Sorry to derail, feel free to put this comment elsewhere if it’s unwanted here. Primenet does record a little bit more info on the Fermat 2ⁿ exponents, but they are only viewable when you have an active assignment, by using the checkbox for “Show exponents currently assigned to me”. Once you finish your assignment you lose access to being able to see the previous results.
Primenet also records PP1 work in addition to ECM, but not PM1; I’m not sure why one but not the other.

[Andrew Usher]

That was also shown in my thread (https://www.mersenneforum.org/showth...t=28281&page=3). And that strengthened my conviction that, as cxc here reiterated, Fermat numbers should be granted full status in the database, as it should be very little additional work to do so. P-1 is in fact not accepted by the server, which is why I must maintain a thread for it.

As she has taken an assignment on F29, I'll add one thing that that would make obvious: the amount of TF done (~2^91) makes running any curves at the t25 level useless, and any ECM on F29 (or F30) should be at t30 regardless. However, it's probably unwise compared to further TF (with an adequate GPU) or P-1. (TF limits from Luigi's poorly-organised page: http://www.fermatsearch.org/stat/done.php). The reason F29 but not F30 is included is presumably the prime95 limit at the time it was started.

In this connection, I just noticed the behavior again that clicking on a thread title from the main forum menu gets the second-last page: should it not be either the first or the last?

[James Heinrich]

Originally Posted by cxc

Primenet does record a little bit more info on the Fermat 2ⁿ exponents, but they are only viewable when you have an active assignment, by using the checkbox for “Show exponents currently assigned to me”. Once you finish your assignment you lose access to being able to see the previous results.

It took a little trickery but the exponent detail page should now show details for Fermat exponents (e.g. F28).

Originally Posted by cxc

Primenet also records PP1 work in addition to ECM, but not PM1; I’m not sure why one but not the other.

I'm not sure if that's intentional or oversight. Do you have an example P-1 result line that isn't accepted that I can test with?

[cxc]

Many thanks, James! Of course at the moment we are in the maintenance window, so I’ll look forward to admiring your trickery in the near future. There’s around 50 people doing ECM-F work presently, but given the Fermats are so few in number (18 in the range from 4k to 512M) it’s hard to know whether that is because they’re somewhat overlooked.

I don’t have a P–1 result to hand, however the thread which Andrew linked in his comment above has several contributors who have done P–1 work: Serge Batalov gives 10 NF results for F21 up to F30 at comment 17, Jan S described finishing 3 tests on F20, F24, and F27 which were also NF and rejected by the server. Elsewhere I saw a thread somewhere mentioning that Bob Silverman did some P–1 testing on Fermats back in the day! So there has been some activity there over a long period.

[Andrew Usher]

Yes, that's now the issue - I think with the pages visible there can no longer be any reason to refuse worktypes (any more than there is on Mersennes), and people seeing them should assume that they would not be. I also now see that known factors are not listed, which is odd.

I would say that it would be best if you (or Aaron) ran a test P-1 and submitted it - you'd see better what went wrong. I strongly suspect it's code that intentionally refuses 'wrong' worktypes, and P+1 gets by as it didn't exist here when the code was written - if so that code could easily be removed.

[James Heinrich]

Originally Posted by Andrew Usher

... if so that code could easily be removed.

Please don't make any assumptions about what is actually "easy" in modifying the code. It might be, it might not be. A lot of the underlying code and database is 25+ years old and may be predicated on assumptions that may not make sense now but did back then. The fundamental "bones" of GIMPS was basically: Mersenne numbers are either Mersenne Primes, or Mersenne Prime candidates, or discarded. If it's LL-composite, it's no longer interesting; if it has a factor, it's no longer interesting. That of course has changed over ~30 years, along with all the additions of P-1, ECM, PRP, P+1, Fermat, etc. But a lot of that stuff is still fundamentally "hacked in", so please don't assume that making apparently-small changes is necessarily "easy".

Originally Posted by Andrew Usher

I also now see that known factors are not listed, which is odd.

If they're not currently listed on the exponent page, I'm not even sure if PrimeNet knows about Fermat factors. They might possibly be stored in another table, they might equally not be stored at all. Since apparently you're up on these things, perhaps you could provide a complete list of known Fermat factors and I'll see if I can find where/if they're stored.

[cxc]

Hi again James, I’d like to take the opportunity to thank you, Aaron, George, and everyone else maintaining Primenet – your work is really, really appreciated, and I’m especially grateful and impressed for the quick turn-around on my request. I am still noticing little tweaks by links that become active on certain pages, and as Aaron said in the other thread there’s a whole lot of stuff that most users are not going to see.

Originally Posted by James Heinrich

Since apparently you're up on these things, perhaps you could provide a complete list of known Fermat factors and I'll see if I can find where/if they're stored.

Herewith:

F12 known-factors = "114689,26017793,63766529,190274191361,1256132134125569,568630647535356955169033410940867804839360742060818433"
F13 known-factors = "2710954639361,2663848877152141313,3603109844542291969,319546020820551643220672513"
F14 known-factor  = "116928085873074369829035993834596371340386703423373313"
F15 known-factors = "1214251009,2327042503868417,168768817029516972383024127016961"
F16 known-factors = "825753601,188981757975021318420037633"
F17 known-factors = "31065037602817,7751061099802522589358967058392886922693580423169"
F18 known-factors = "13631489,81274690703860512587777"
F19 known-factors = "70525124609,646730219521,37590055514133754286524446080499713"
F21 known-factor  = "4485296422913"
F22 known-factor  = "64658705994591851009055774868504577"
F23 known-factor  = "167772161"
F25 known-factors = "25991531462657,204393464266227713,2170072644496392193"
F26 known-factor  = "76861124116481"
F27 known-factors = "151413703311361,231292694251438081"
F28 known-factor  = "1766730974551267606529"
F29 known-factor  = "2405286912458753"

F20 and F24 are in the GIMPS numeric range, and composite, but unfactorised. Some factors are already there by virtue having been recovered by GIMPS ECM work, and there are four unique discoveries, Tapio Rajala’s F14 factor, and the three discovered by David Bessell for F17, F19, and F22. (If you look in the data for F12 there’s also a truly prodigious 75-digit composite factor found by user Yxinity in 2017, which turns out to be the product of the first, second, third, and sixth prime factors by size.)

It may not even really be necessary to list these against each of those Fermat exponents, except for the sake of thoroughness and completeness? The number of Fermat factors is so small (compared with the amount of factor data for the ~30 million factored Mersennes with exponent below 10⁹) that it’s not as if they’re likely to be missed!

[Prime95]

Originally Posted by James Heinrich

If they're not currently listed on the exponent page, I'm not even sure if PrimeNet knows about Fermat factors. They might possibly be stored in another table, they might equally not be stored at all. Since apparently you're up on these things, perhaps you could provide a complete list of known Fermat factors and I'll see if I can find where/if they're stored.

Fermat factors are in t_gimps_ecm_data -- at least for factors that were known in 2000. Prime95 also has the known factors hardwired into the source code.