Some questions
Message boards :
Questions/Problems/Bugs :
Some questions
Message board moderation
Author | Message |
---|---|
Send message Joined: 12 Mar 12 Posts: 12 Credit: 11,488 RAC: 0 |
I have some questions regarding this project: 1. I have received a workunit that begins "S5p433". I have received another workunit that begins "S2p1000b". I can understand the second one (it's for 2,1000+), but not the first - that seems to be 5,433+, but it's not on the Status of Numbers page. What's going on here? And what do the names of the workunits mean in the first place? 2. What's the status of the sieving for 2,1061- and 7,355+? 3. When are the numbers not yet started (ie. 7,365- / 10,305+ / 11,290+ / 2,1000+) going to start? 4. Is there a visual app (like those for WCG and Rosetta@home) that will display the progress of sieving for a workunit? 5. Are there any plans to factor more non-Cunningham numbers (like EM48)? Freywa on FIMFiction |
Send message Joined: 2 Oct 09 Posts: 50 Credit: 111,128,218 RAC: 0 |
I have some questions regarding this project: Parts of these two questions are related. The Cunningham number 5, 433+ is the project that is intended to follow 2, 1061-, once 2, 1061- is completed. If you're receiving tasks for 5p433, that's a good indication that 2, 1061- has completed sieving, or will soon; provided that the next step is up and running (the matrix step, a non-bonic calculation, done under the terragrid grant, on a national supercomputing site). In particular, 5p433 is the next number that uses the 16e siever.
The other numbers listed on the status page are all 15e projects (that's why 5p433 is being added), and one expects that they'll be done in order, 2,1000+ first then the next, 11,290+, then ... I can't speak to plans for visual apps or non-Cunningham numbers; I would expect that a fair part of Greg's attention is going to the next step for 2,1061- (a record- setting number; and the most wanted Mersenne number --- the last one from George Woltman's list of 2, n- with n <1200 having no known factor). -bdodson* |
Send message Joined: 12 Mar 12 Posts: 12 Credit: 11,488 RAC: 0 |
Your previous post answers all my questions, except part of number 2. I would expect that a fair part of Greg's attention is going to the next step for 2,1061- (a record-setting number, and the most wanted Mersenne number - the last one from George Woltman's list of 2,n- with n < 1200 having no known factor). 2^1061-1. What's the next Mersenne without known factors? Just for reference, here's the full number in base 10: 2470730631 1927565716 8573421287 7408533319 7833223161 8796822389 3530608280 5123046306 9936475077 7605433648 6228891340 8589858290 2707626188 7914242781 6178466724 5343138690 3982455635 5421587484 0182398598 8322905245 0779385675 1325219817 9128990807 9367801947 8139154740 4884040101 6062951113 6882502627 3254703636 0263072077 6443643892 9167613951 But you forgot 7,355+! What's the status of that? Freywa on FIMFiction |
Send message Joined: 2 Oct 09 Posts: 50 Credit: 111,128,218 RAC: 0 |
Your previous post answers all my questions, except part of number 2. Until recently, this was M1237, but there was a spectacular ECM factor with 70-digits found using a network of PS3's. The cofactor has 303-digits, and is composite, so it could be a very long time before we know the rest of the factors. There's currently an effort on mersenneforum to factor M929, after which M947 will be the smallest not completely factored. Note that M1007 is currently fourth, in about the same range as 2p1000, which is first on the 2+ list, and 10th on the Cunningham Most Wanted list. So anyway, the GIMPS report currently lists M1277 as the next smallest after M1061, then M1619. The exponent range there is up to 10000 looks more than 50 less than 100 http://www.mersenne.org/report_factoring_effort
I thought you knew that. So long as tasks for 2p1000 are being distributed, that means that Greg doesn't need more reports to start the matrix step. These 15e projects are smaller numbers, and the matrix is most likely running on the cluster at Cal State -bdodson* (still NFS@Home's number 1 contributor, hoping that both Greg and I will get bumped by someone else!) |
Send message Joined: 12 Mar 12 Posts: 12 Credit: 11,488 RAC: 0 |
I figure, then, that 7,355+ and 2,1061- are both in the matrix step. On the PrimeNet page (http://www.mersenne.org/report_exponent/?exp_lo=1061&exp_hi=1061&B1=Get+status) it says there are no factors below 2^63. That looks OK. Now I want to know where I can track the progress of factoring 2,929- on Mersenne Forums, how long it takes to do the postprocessing, and what SNFS difficulty is. Freywa on FIMFiction |
Send message Joined: 12 Mar 12 Posts: 12 Credit: 11,488 RAC: 0 |
It seems like 7,355+ has been factored: 2515330297 6907793161 8220867745 7273528948 4404522729 6353222660 8775719348 0305097071 * 4916874214 3759415895 4305310966 7460912512 3792342558 8744451375 1476522168 2782157811 6713493332 1338000377 3870659795 5655149900 757971 = (7^355 + 1) / 8270277554 4295268462 1699584764 9086085761 1172272053 2025972445 6265943017 5387827352 3124077926 68984 http://homes.cerias.purdue.edu/~ssw/cun/page123 http://factordb.com/index.php?id=1100000000020350399 ...M1061 is "on hold". The current siever has run out What is special-q and what is the spring break? I live in Singapore, so all this doesn't make sense to me. Freywa on FIMFiction |
Send message Joined: 2 Oct 09 Posts: 50 Credit: 111,128,218 RAC: 0 |
Among the google hits from "special-q factoring" you may find links to mersenneforum, as well as several technical reports on which special-q have been used in various "number field" and SNFS ("special" number field) factorizations. They are used in the "lattice sieving" step in the "number field sieve". This is a message board for a boinc project that uses special-q for lattice sieving. There are many prerequisites necessary for even a general description, which are outside of this project's focus on people intending to request and complete sieving tasks. You might try wikipedia for "spring break". -bdodson* ----- @Bob --- We discussed the selection of 5p433, and here's Greg's view
M1007 was a candidate from the list of targets of the EPFL effort using ECM on their PS3 cluster. We had just recently completed M1031, and even then were hoping that M1061 would finish in the next few weeks. The other numbers weren't in consideration on our list. |
Send message Joined: 12 Mar 12 Posts: 12 Credit: 11,488 RAC: 0 |
This is a message board for a BOINC project that uses special-q for Oh LOL. Assume special-q is a parameter in lattice sieving Aside from that, I've received the first tasks for 11,290+. Much of page 123 on the Cunningham project website is about the recently-extended base 3 tables. Because of this, a lot of holes and wanted numbers have been left untouched. When is NFS@Home going to factor these numbers? Particularly, I would like to see 10,281+ and 10,281-. Freywa on FIMFiction |
Send message Joined: 26 Jun 08 Posts: 645 Credit: 474,630,308 RAC: 211,930 |
Most of your questions have been answered, but here's a bit more info. And what do the names of the workunits mean in the first place? Taking a recent workunit, S2m1061d_1074881, 'S' indicates that it's using SNFS, '2m1061' refers to the number being factored, 2,1061-, 'd' refers to the fourth series of runs through this number (this can get incremented for any number of reasons, including sieving on a different side, using different parameters, a mistake in the first assignments, etc.), and '1074881' refers to the starting q for this workunit. For this workunit, special-q's appearing in the range 1074881000 - 1074882000 are being sieved. The special-q's are always prime numbers, so there's no worry that the end of a workunit and the start of the next workunit, which are always multiples of 1000, overlap. 4. Is there a visual app (like those for WCG and Rosetta@home) that will display the progress of sieving for a workunit? There are no plans to develop one. 5. Are there any plans to factor more non-Cunningham numbers (like EM48)? Yes. Following the currently reserved Cunningham numbers, the 15e sievers will be doing a couple of numbers for the Odd Perfect project. |
Send message Joined: 26 Jun 08 Posts: 645 Credit: 474,630,308 RAC: 211,930 |
Particularly, I would like to see 10,281+ and 10,281-. Not for a while. These numbers are too large for 15e, but too easy for 16e to be of current interest. Eventually, though... |
Send message Joined: 12 Mar 12 Posts: 12 Credit: 11,488 RAC: 0 |
There are no plans to develop one [visual app for NFS@Home]. :/ Yes. Following the currently reserved Cunningham numbers, the 15e sievers will be doing a couple of numbers for the Odd Perfect project. What are the two numbers (or if you can't provide a concise description, provide a link)? Freywa on FIMFiction |
Send message Joined: 26 Jun 08 Posts: 645 Credit: 474,630,308 RAC: 211,930 |
What are the two numbers (or if you can't provide a concise description, provide a link)? 59^149-1 and 197^113-1. |
Send message Joined: 12 Mar 12 Posts: 12 Credit: 11,488 RAC: 0 |
In light of recent completed factorisations, I hereby ask questions 6 and 7. Why does linear algebra take place on TeraGrid and not BOINC? How long does it take? In addition, what new numbers are on the consideration list for NFS@Home factorisation now? Freywa on FIMFiction |
Send message Joined: 12 Mar 12 Posts: 12 Credit: 11,488 RAC: 0 |
Recent news: I have just received some workunits for some GNFS number. One of them reads "GB200_25036_0". What is B200? Freywa on FIMFiction |
Send message Joined: 26 Jun 08 Posts: 645 Credit: 474,630,308 RAC: 211,930 |
The unfactored part of the numerator of the 200th Bernoulli number. See http://www.mersenneforum.org/showthread.php?t=16737 for details. |
Send message Joined: 26 Jun 08 Posts: 645 Credit: 474,630,308 RAC: 211,930 |
The lasieve5 siever was deployed and used to complete 2,1061-. This has been done. |