mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > FermatSearch

Reply
 
Thread Tools
Old 2018-12-18, 13:55   #133
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

72·97 Posts
Default

Quote:
Originally Posted by Dylan14 View Post
Perhaps I am a bit confused here, as on the new range page, it only shows that up to k = 30e9 has been tested when I search for n = 550-559. Whereas on the n factoring page it shows that it has been searched to 3e12.
Here.

Where does your 30e9 comes from?
ET_ is offline   Reply With Quote
Old 2018-12-18, 14:35   #134
Dylan14
 
Dylan14's Avatar
 
"Dylan"
Mar 2017

46810 Posts
Default

Quote:
Originally Posted by ET_ View Post
Here.

Where does your 30e9 comes from?
When I go to the new range search and I type 550 and 559 into the boxes below the table, I get two lines of data after I click on “Input data”. The last line shows gives “30,000,000,000” or 30e9 in the To K column. That is where I get the 30e9.
Dylan14 is offline   Reply With Quote
Old 2018-12-18, 16:36   #135
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

112218 Posts
Default

Quote:
Originally Posted by Dylan14 View Post
When I go to the new range search and I type 550 and 559 into the boxes below the table, I get two lines of data after I click on “Input data”. The last line shows gives “30,000,000,000” or 30e9 in the To K column. That is where I get the 30e9.
Try 540-559...

Unfortunately, the ranges have been requested and done in different chunks, not just in chunks of 10 N, and the range search operates on the database of requests.

Mark told me many times that I could rework the ranges accordingly:
chunks of 10 N for 100 < N < 999
chunks of 100 N for 1000 < N < 9999
chunks of1000 N for 10000 < N < 99999

I objected that it would be a bit tricky, it would not reflect the work of the researchers, and even if I did there would be odd chunks anyway, so he decided to scrape the results and use his own program to get the correct values in correct chunks.

Assuming that the range search page is a continuous source of misunderstanding, I could decide in the future to recode his program in PHP (provided is does not become expensive on SQL queries).

Sorry for the issue.

Luigi

Last fiddled with by ET_ on 2018-12-18 at 16:37
ET_ is offline   Reply With Quote
Old 2018-12-18, 18:00   #136
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

570910 Posts
Default

Luigi and I are on the same page. I wrote my program to help identify gaps, ranges of n and k that were somehow missed by previous searchers. This created two challenges. First, for any given n, what is a reasonable expectation for searching other n above or below that n to the same k. For example, if you have n = 501, should that n be searched to the same k as n = 500? What about n = 502? The answer is "it depends". Second, given a grouping of n, what is a reasonable expectation for searching a range of k for that n? For example, assuming that we grouped 10 n, 500-509 as an example, should a searcher reserve a k range of 1e5, 1e8, 1e10, etc? The answer to this is driven more by the speed of programs used to search the range of n and how long it would take a single core to complete the search of the range.

To help me group the n into ranges, I looked at the data to find the gaps. Those gaps along with the patterns of max k per n lead to the recommendations I gave to Luigi. I think that most participants who read mersenneforum have been following them. Hopefully they have made his life easier when it comes to managing reservations and completed ranges.

To summarize:
Code:
         30 <= n <    100, k multiple of 1e14
        100 <= n <    500, k multiple of 1e12
        500 <= n <   1000, k multiple of 1e11
       1000 <= n <   5000, k multiple of 1e9
       5000 <= n <  10000, k multiple of 1e8
      10000 <= n <  40000, k multiple of 1e6
      40000 <= n < 100000, k multiple of 1e4
     100000 <= n are not included
These are arbitrary, but make the most sense based upon the data. For n >= 100000 all k < 10000 are being searched by PrimeGrid.

I have attached source code and lists (generated by that code) to this post. The "groups.txt" file shows the groups of n and the depth to which they have been searched. The "gaps.txt" file shows n where some n for the range have been searched to different k than other n in that range. The "wanted.txt" file shows n where a higher n has been searched to a higher k.

BTW, there is one difference between the attached code and the output. The output has details for n < 30. The attached code excludes n < 30. The reason is that ECM factoring has not found any new factors for these Fermat numbers. Statistically it is extremely unlikely that there are any unknown factors for these Fermat numbers that are less than 50 digits in length which is well beyond the ranges already tested using GFN testing software.
Attached Files
File Type: 7z flist.7z (7.7 KB, 42 views)

Last fiddled with by rogue on 2018-12-18 at 18:10
rogue is offline   Reply With Quote
Old 2018-12-18, 20:37   #137
Dylan14
 
Dylan14's Avatar
 
"Dylan"
Mar 2017

46810 Posts
Default

Thanks Mark and Luigi for the clarifications. I will instead take n = 250-259, k = 71e12-75e12 to start with.
Dylan14 is offline   Reply With Quote
Old 2018-12-31, 17:41   #138
Dylan14
 
Dylan14's Avatar
 
"Dylan"
Mar 2017

22·32·13 Posts
Default

Quote:
Originally Posted by Dylan14 View Post
Thanks Mark and Luigi for the clarifications. I will instead take n = 250-259, k = 71e12-75e12 to start with.

Range is complete. No factors were found.
Dylan14 is offline   Reply With Quote
Old 2019-01-09, 19:14   #139
Dylan14
 
Dylan14's Avatar
 
"Dylan"
Mar 2017

22·32·13 Posts
Default

Taking n = 92000-92999, k = 15e4-16e4.
Dylan14 is offline   Reply With Quote
Old 2019-01-18, 13:09   #140
Dylan14
 
Dylan14's Avatar
 
"Dylan"
Mar 2017

1110101002 Posts
Default

Quote:
Originally Posted by Dylan14 View Post
Taking n = 92000-92999, k = 15e4-16e4.

This range is completed; I used gfndsieve + pfgw with the -gx flag. 158 PRP's were found, but no factors.
Dylan14 is offline   Reply With Quote
Old 2019-01-21, 15:04   #141
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

3·11·173 Posts
Default

Taking n 46001-46999 k 100e4-120e4.

This range shows up as a "gap" because someone completed a range that ended at 46000 instead of 45999.

I see that Serge has a number of reservations dating back to August. With the horsepower available to him, I'm surprised that those ranges are not done.
rogue is offline   Reply With Quote
Old 2019-01-30, 02:21   #142
Dylan14
 
Dylan14's Avatar
 
"Dylan"
Mar 2017

22·32·13 Posts
Default

Taking n = 490-499, k = 6e12-7e12.
Dylan14 is offline   Reply With Quote
Old 2019-02-05, 01:30   #143
Dylan14
 
Dylan14's Avatar
 
"Dylan"
Mar 2017

22·32·13 Posts
Default

Quote:
Originally Posted by Dylan14 View Post
Taking n = 490-499, k = 6e12-7e12.
Range is complete. No factors were found.
Taking n = 250-259, k = 75e12-100e12.
Dylan14 is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Reservations ET_ Operazione Doppi Mersennes 484 2020-03-23 19:13
Reservations kar_bon Riesel Prime Data Collecting (k*2^n-1) 129 2016-09-05 09:23
Reservations? R.D. Silverman NFS@Home 15 2015-11-29 23:18
Reservations R.D. Silverman Cunningham Tables 15 2011-03-04 21:01
4-5M Reservations paulunderwood 3*2^n-1 Search 15 2008-06-08 03:29

All times are UTC. The time now is 06:26.

Mon Jun 1 06:26:20 UTC 2020 up 68 days, 3:59, 2 users, load averages: 1.62, 1.46, 1.43

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.