mersenneforum.org Aliquot cycles search, the state of the art.
 Register FAQ Search Today's Posts Mark Forums Read

 2016-03-07, 19:59 #89 Sergei Chernykh   Jun 2015 Stockholm, Sweden 83 Posts Exciting times! Today I've found the first type (8, 1) amicable pair: Code: 81 Chernykh 2016 14438796350376953678174322791271636288778421212339464839728409858893107308551670=2*5*13*59*71*73*1061*3019*4211*65521001603*4841941814273029742117557*84877110421196263915758493 14865176913743789894174101450879657763077918214583046351774081022511427255902730=2*5*13*59*1938093469849255527271721180036461246815895464743552327480323470992363397119 http://sech.me/ap/news.html#20160307
2016-03-08, 15:25   #90
R. Gerbicz

"Robert Gerbicz"
Oct 2005
Hungary

34×17 Posts

Quote:
 Originally Posted by Sergei Chernykh Exciting times! Today I've found the first type (8, 1) amicable pair: Code: 81 Chernykh 2016 14438796350376953678174322791271636288778421212339464839728409858893107308551670=2*5*13*59*71*73*1061*3019*4211*65521001603*4841941814273029742117557*84877110421196263915758493 14865176913743789894174101450879657763077918214583046351774081022511427255902730=2*5*13*59*1938093469849255527271721180036461246815895464743552327480323470992363397119 http://sech.me/ap/news.html#20160307
Congrats! That's really nice.

Other: any possibility to use (at least at upload) my stripped format (see post number 73), another improvement:
allow hexadecimal base, or even base=62 (see: https://gmplib.org/manual/I_002fO-of...fO-of-Integers), obviously there are more than 62 printable characters, but that now it would be enough. The only ambiguity with this that there is NO prefix for base=62, we could use say 0s as a prefix. Or no prefix at all (what gmp uses for base=62), but in that case we should use base=62 everywhere (excluding at year?). Of course we would not need these base tricks for a zip file (as in the c2 file there is almost no other than 0-9 character).

Any other suggestion?

2016-03-08, 15:30   #91
Sergei Chernykh

Jun 2015
Stockholm, Sweden

83 Posts

Quote:
 Originally Posted by R. Gerbicz Congrats! That's really nice. Other: any possibility to use (at least at upload) my stripped format (see post number 73), another improvement: allow hexadecimal base, or even base=62 (see: https://gmplib.org/manual/I_002fO-of...fO-of-Integers), obviously there are more than 62 printable characters, but that now it would be enough. The only ambiguity with this that there is NO prefix for base=62, we could use say 0s as a prefix. Or no prefix at all (what gmp uses for base=62), but in that case we should use base=62 everywhere (excluding at year?). Of course we would not need these base tricks for a zip file (as in the c2 file there is almost no other than 0-9 character). Any other suggestion?
I already use stripped format for backups. Minified + compressed database is only ~5.3 GB in size. If there is a need in this, I can publish these backups (all c2*.min.txt files in one archive). There is no need in base=62 because these text files are compressed very effectively using PPMd compression available in 7-zip.

2016-03-08, 15:38   #92
R. Gerbicz

"Robert Gerbicz"
Oct 2005
Hungary

25418 Posts

Quote:
 Originally Posted by Sergei Chernykh I already use stripped format for backups. Minified + compressed database is only ~5.3 GB in size. If there is a need in this, I can publish these backups (all c2*.min.txt files in one archive). There is no need in base=62 because these text files are compressed very effectively using PPMd compression available in 7-zip.
I would download that (I have an old database, containing roughly 40m amicable pairs).

But then why not allow the stripped format at upload (without the base=16,62 trick), and stil allowing the old antique format? It is easy to recognise these in the same file as in the old format there is an '=' sign what is missing in the stripped format.

2016-03-08, 16:05   #93
Sergei Chernykh

Jun 2015
Stockholm, Sweden

10100112 Posts

Quote:
 Originally Posted by R. Gerbicz I would download that (I have an old database, containing roughly 40m amicable pairs). But then why not allow the stripped format at upload (without the base=16,62 trick), and stil allowing the old antique format? It is easy to recognise these in the same file as in the old format there is an '=' sign what is missing in the stripped format.
Submission form removes everything before '=' sign on upload + gzip compression is used if supported by client, I think this is enough for reducing bandwith. The point is to check that whoever submitted pairs has found all prime factors correctly, so they're all needed.

2016-03-09, 10:09   #94
AndrewWalker

Mar 2015
Australia

1228 Posts

Quote:
 Originally Posted by Sergei Chernykh Today I started up my automatic BDE software on one PC to process some 6-primes breeders I found, but I forgot how much automatic it was: it also picked up this pair and processed it, so it counted 11868 type (5,2) pairs (103-128 digits) on me. I'll see what I can do to fix it.
Congrats on the new (8,1) pair! I'm just wondering how your software found so many more extra pairs than I found. I'm guessing you get extra breeders as mentioned on your pages? What I've done with my new 2 and some others is feed them into the Wiethaus Rule (in the survey paper). Regular (n,1) pairs can be used as inputs into this. For k=0 I *think* it gives the same pairs as te Riele's rule, for higher k the numbers drop off rapidly but sometimes may give a big pair. I'm mainly using these (n,1)s to search for a new record. No luck yet but did get a 28k pair today!

Andrew

2016-03-09, 10:15   #95
Sergei Chernykh

Jun 2015
Stockholm, Sweden

83 Posts

Quote:
 Originally Posted by AndrewWalker Congrats on the new (8,1) pair! I'm just wondering how your software found so many more extra pairs than I found. I'm guessing you get extra breeders as mentioned on your pages? What I've done with my new 2 and some others is feed them into the Wiethaus Rule (in the survey paper). Regular (n,1) pairs can be used as inputs into this. For k=0 I *think* it gives the same pairs as te Riele's rule, for higher k the numbers drop off rapidly but sometimes may give a big pair. I'm mainly using these (n,1)s to search for a new record. No luck yet but did get a 28k pair today! Andrew
I used the standard "Borho's Rule with breeders" from survey paper (page 13) which is the same as te Riele's rule in your case (amicable pair of type (i, 1)).

 2016-03-09, 11:03 #96 AndrewWalker     Mar 2015 Australia 1228 Posts Ok that makes sense. I mixed up the pairs sorry, the 106 pairs were from the (3,1) pair which I searched with Wiethaus to k=45. The (4,1) pair I only searched to k=6 as it was taking a lot longer! A request for a future feature on your pages would be to have downloads for (N,1) and X(N,1) pairs. Jan did have these and other types on his pages, but for most of the other types the files would now be too big! Actually I just saw you added it so thanks anyway! Andrew I submitted the 28635 digit pair, it comes from the above (3,1) pair using the Wiethaus rule with parameters # a=25827742215 # S=172906877439567041 # p=126910347247487795993 # q=177417032444500799 # k=711 # q1=127087764279932296792*126910347247487795993^711-1 # q2=126737440370048228952*126910347247487795993^711-1 for anyone interested.
2016-03-10, 21:36   #97
R. Gerbicz

"Robert Gerbicz"
Oct 2005
Hungary

34·17 Posts

Quote:
 Originally Posted by Sergei Chernykh I already use stripped format for backups. Minified + compressed database is only ~5.3 GB in size. If there is a need in this, I can publish these backups (all c2*.min.txt files in one archive).

2016-04-05, 09:31   #98
R. Gerbicz

"Robert Gerbicz"
Oct 2005
Hungary

34·17 Posts

Quote:
 Originally Posted by Sergei Chernykh Submission form removes everything before '=' sign on upload + gzip compression is used if supported by client, I think this is enough for reducing bandwith.
And what should I have for that compression? I have Google Chrome 49.0.2623.110 (64-bits)
My recent submission of a 190.4Mb (this is 2*10^8 bytes) file (sorted_aps000368.txt),
from the middle of the file a typical message:
Code:
sorted_aps000368.txt: lines 906284 - 910403 of 1712537
Tue, 05 Apr 2016 07:32:51 UTC
Syntax check done in 11.322 ms
Verified 1030 pairs in 295.231 ms
Added 1027 new pairs in 27.460 ms
New pairs from GERBICZ were added to database
Updated stats and logs in 6.590 ms
Total execution time: 341.827 ms
The total time needed to upload and check took roughly 1211 seconds.
since the file has 1712537 lines the checking took approx. 0.341*1712536/(910403-906284+1)=141 seconds, so in upload spent more than 1000 seconds. Clearly it hasn't done compression (or other thing?). Currently my upload speed is 0.85 Mbit.

The size of the above file if I remove in every line everything before '=' sign is 104.7 Mbytes, after zipping its size is only 28.5 Mbytes.

 2016-04-05, 14:45 #99 Sergei Chernykh   Jun 2015 Stockholm, Sweden 83 Posts I was under impression that it's done by the browser automatically, since it works for downloading files... Apparently, it must be done explicitly on client side when uploading: http://stackoverflow.com/questions/8...-to-the-server I'll try to do something tomorrow morning.

 Similar Threads Thread Thread Starter Forum Replies Last Post Drdmitry Aliquot Sequences 25 2016-12-16 15:26 Drdmitry Aliquot Sequences 302 2016-05-11 02:17 schickel Aliquot Sequences 7 2013-02-08 01:33 Drdmitry Aliquot Sequences 0 2011-12-14 13:50 R. Gerbicz Math 0 2010-07-01 12:30

All times are UTC. The time now is 00:23.

Fri Aug 7 00:23:27 UTC 2020 up 20 days, 20:10, 1 user, load averages: 1.43, 1.44, 1.49