mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > Prime Gap Searches

Reply
 
Thread Tools
Old 2020-03-09, 20:24   #12
MJansen
 
Jan 2018

37 Posts
Default

Hi Seth, (it is Seth right?)

I tested the gap enrty page with another batch, and it ran fine (although a bit long ... ;-)

Some remarks:
1. if I have more than one day of gaps to enter, I will have to add them in daily batches. Since there are a lot of gaps (1000+ since januari 1st 2020) that would not be very convenient.
2. I do like your set up of day selection though. is there a solution both ways can be supported? So adding a batch with a data line including date_of_gap_found and the second option is selecting a date?
3. Bigger gaps take some time to verify, so around 1000 gaps will take some serious time to verify, would that be a problem (are the resources and especially the server capacity guaranteed?)
4. The script noticed one gap was not an improvement (77986 17.028162 11413*4657#/30030 - 52636), a bigger merit had been found for that gap already and the script did correctly skip that gap from the verification (just so you know!)

All in all, pretty nice steps forward!
Kind regards
Michiel Jansen


(36160, '2167345*1559#/210-20134', 'Verified! (4.1s)')
(36444, '2153219*1567#/210-24126', 'Verified! (3.9s)')
(39724, '2120297*1571#/210-16396', 'Verified! (4.5s)')
(52688, '1098161*2557#/210-30638', 'Verified! (19.4s)')
(59010, '1108747*2557#/210-30724', 'Verified! (22.4s)')
(59688, '1115003*2557#/210-19542', 'Verified! (23.9s)')
(57092, '1106605*2591#/210-26538', 'Verified! (23.5s)')
(79420, '111895*4157#/2310-39778', 'Verified! (106.4s)')
(80664, '111025*4159#/2310-62446', 'Verified! (112.1s)')
(97324, '11225*6529#/30030-42358', 'Verified! (323.0s)')
(92022, '11429*4801#/30030-48674', 'Verified! (150.5s)')
(114294, '11381*4987#/30030-61496', 'Verified! (219.0s)')
(78752, '11377*5003#/30030-21974', 'Verified! (143.1s)')
(83152, '11381*4517#/30030-46446', 'Verified! (118.0s)')
(86508, '110905*4153#/2310-26366', 'Verified! (111.5s)')
(94954, '109471*4139#/2310-50124', 'Verified! (107.3s)')
MJansen is offline   Reply With Quote
Old 2020-03-09, 20:36   #13
SethTro
 
SethTro's Avatar
 
"Seth"
Apr 2019

101111112 Posts
Default

Quote:
Originally Posted by MJansen View Post
Hi Seth, (it is Seth right?)

I tested the gap enrty page with another batch, and it ran fine (although a bit long ... ;-)

Some remarks:
1. if I have more than one day of gaps to enter, I will have to add them in daily batches. Since there are a lot of gaps (1000+ since januari 1st 2020) that would not be very convenient.
2. I do like your set up of day selection though. is there a solution both ways can be supported? So adding a batch with a data line including date_of_gap_found and the second option is selecting a date?
3. Bigger gaps take some time to verify, so around 1000 gaps will take some serious time to verify, would that be a problem (are the resources and especially the server capacity guaranteed?)
4. The script noticed one gap was not an improvement (77986 17.028162 11413*4657#/30030 - 52636), a bigger merit had been found for that gap already and the script did correctly skip that gap from the verification (just so you know!)

All in all, pretty nice steps forward!
Kind regards
Michiel Jansen
It is Seth,

1. I think I have a limit somewhere of 30, I'm happy to bulk process all of your entries.
2. I'm not sure what "both ways" is, I'm happy to parse out the date from a csv line (just give me an example format).
3. This is running on my home PC but I'll upload it to Google Cloud Platform at some point. I'm not sure what to do about slowness, but let's deal with that at that time. (maybe the solution is to only test a random fraction of the inner numbers for some trusted uploaders).
4. There's currently an issue where the database's merit is rounded so I add 0.005 to it before comparing with input merit, so it's possible your record was just a small improvement (and hit this bug). I'll add a todo to handle this better (maybe parsing the string value for all records?)

Last fiddled with by SethTro on 2020-03-09 at 20:37
SethTro is online now   Reply With Quote
Old 2020-03-09, 21:51   #14
mart_r
 
mart_r's Avatar
 
Dec 2008
you know...around...

23×3×52 Posts
Default

Nice work!

Though I'm not currently looking for the "simple" prime gaps, I'm looking forward to do so again in the near future.
mart_r is offline   Reply With Quote
Old 2020-03-09, 22:32   #15
MJansen
 
Jan 2018

37 Posts
Default

I can PM the improved gaps after august 12th 2019. Will be tomorrow evening (off for bed now)
With both ways, I mean your current batch processing option and another OR on the web page that has a date in the data-line (e.g. Gap DateFound Merit6 GapStart like: 36160 8-3-2020 23.746468 2167345*1559#/210 - 20134)
I don't think the verifying is too slow, just when processing many gaps it might be a strain on resources and they have to be available otherwise it will be frustrating. Choices are to test all gaps and accept slowness (I personally will), or test a few and give out C?? ratings for the rest, or trust the people that have proven to deliver correct results in the past (Tom did not trust anyone, he liked to verify every result if possible ;-)
regarding the rounding, why not use 6 digits and round from there? Chance of incorrect merit values decreases significantly!

Kind regards
Michiel Jansen


Quote:
Originally Posted by SethTro View Post
It is Seth,

1. I think I have a limit somewhere of 30, I'm happy to bulk process all of your entries.
2. I'm not sure what "both ways" is, I'm happy to parse out the date from a csv line (just give me an example format).
3. This is running on my home PC but I'll upload it to Google Cloud Platform at some point. I'm not sure what to do about slowness, but let's deal with that at that time. (maybe the solution is to only test a random fraction of the inner numbers for some trusted uploaders).
4. There's currently an issue where the database's merit is rounded so I add 0.005 to it before comparing with input merit, so it's possible your record was just a small improvement (and hit this bug). I'll add a todo to handle this better (maybe parsing the string value for all records?)
MJansen is offline   Reply With Quote
Old 2020-03-14, 07:53   #16
SethTro
 
SethTro's Avatar
 
"Seth"
Apr 2019

191 Posts
Default

I proposed some submitted records to the main github.
https://github.com/primegap-list-pro...ap-list/pull/5

If it gets accepted I'll try to automate that step too.
SethTro is online now   Reply With Quote
Old 2020-03-14, 08:28   #17
MJansen
 
Jan 2018

37 Posts
Default

That would be great. As I said earlier, I am new to GitHub, can you show a view / webpage on github where the record gaps are shown?

Kind regards
Michiel

Quote:
Originally Posted by SethTro View Post
I proposed some submitted records to the main github.
https://github.com/primegap-list-pro...ap-list/pull/5

If it gets accepted I'll try to automate that step too.
MJansen is offline   Reply With Quote
Old 2020-03-14, 11:11   #18
SethTro
 
SethTro's Avatar
 
"Seth"
Apr 2019

19110 Posts
Default

Quote:
Originally Posted by MJansen View Post
That would be great. As I said earlier, I am new to GitHub, can you show a view / webpage on github where the record gaps are shown?

Kind regards
Michiel
This is the current master list
https://raw.githubusercontent.com/pr...er/allgaps.sql

I've proposed some changes (a pull request) which will make these changes to this, but it's waiting for another person (Graham in this case) to review

Graham also wrote a nice frontend to view the data in some interesting ways.
https://primegap-list-project.github.io/lists/
For example
Top merit: https://primegap-list-project.github...gaps-by-merit/
Top discoverers: https://primegap-list-project.github...p-discoverers/
SethTro is online now   Reply With Quote
Old 2020-03-16, 16:31   #19
MJansen
 
Jan 2018

37 Posts
Default

Thanks, appreciated!

PS how are the 2500 updates going? I noticed you were adding the gaps in batches to your SQL site, but verifying seems a lengthy process? I am adding the daily improvements to the website as you probably have noticed.

Kind regards
Michiel

Quote:
Originally Posted by SethTro View Post
This is the current master list
https://raw.githubusercontent.com/pr...er/allgaps.sql

I've proposed some changes (a pull request) which will make these changes to this, but it's waiting for another person (Graham in this case) to review

Graham also wrote a nice frontend to view the data in some interesting ways.
https://primegap-list-project.github.io/lists/
For example
Top merit: https://primegap-list-project.github...gaps-by-merit/
Top discoverers: https://primegap-list-project.github...p-discoverers/
MJansen is offline   Reply With Quote
Old 2020-03-18, 07:08   #20
SethTro
 
SethTro's Avatar
 
"Seth"
Apr 2019

191 Posts
Default

Quote:
Originally Posted by MJansen View Post
Thanks, appreciated!

PS how are the 2500 updates going? I noticed you were adding the gaps in batches to your SQL site, but verifying seems a lengthy process? I am adding the daily improvements to the website as you probably have noticed.

Kind regards
Michiel
It's been slow work. I should really throw more than one core at it but I've been loathed to do that. I'm still waiting on a gmp commit that should speed it up ~50%.

Hopefully after these are in the daily updates only take an hour or so.
SethTro is online now   Reply With Quote
Old 2020-03-19, 23:24   #21
SethTro
 
SethTro's Avatar
 
"Seth"
Apr 2019

191 Posts
Default

@robert44444uk @MJansen

Graham gave me the power to commit records to the official GitHub repository (i have currently been submitting them to my personal copy), I asked a week ago about frequency of submitting records and hey said I should hash it out with Rob.

I proposals the primegaps website should start pushes batches of records directly to the repository with no review (AKA trust me and my code)
Pros:
  • I'm a reasonable person
  • Code is open source and simple to verify.
  • Even if something went wrong there's a nice audit trail of what went wrong

An alternative would be that the primegaps website creates Pull Requests (like this one) and I/Rob/Graham hand review them before they are made official. This adds human eye in the process to make sure nothing looks weird, but I feel it's going to be overly burdensome and substantially slow down the time for reporting till they are part of the official record (we can imagine fully automating reporting form Pgsurround.pl in the future if we go with proposal 1)

What I'm asking for.
1. A yes or no if both of you are okay if I start submitting records I hand review to the repository
2. A yes or no if both of you are okay with me changing the website to automatically submit records after verification.
SethTro is online now   Reply With Quote
Old 2020-03-20, 08:45   #22
robert44444uk
 
robert44444uk's Avatar
 
Jun 2003
Oxford, UK

1,901 Posts
Default

Quote:
Originally Posted by SethTro View Post
@robert44444uk @MJansen

Graham gave me the power to commit records to the official GitHub repository (i have currently been submitting them to my personal copy), I asked a week ago about frequency of submitting records and hey said I should hash it out with Rob.

I proposals the primegaps website should start pushes batches of records directly to the repository with no review (AKA trust me and my code)
Pros:
  • I'm a reasonable person
  • Code is open source and simple to verify.
  • Even if something went wrong there's a nice audit trail of what went wrong

An alternative would be that the primegaps website creates Pull Requests (like this one) and I/Rob/Graham hand review them before they are made official. This adds human eye in the process to make sure nothing looks weird, but I feel it's going to be overly burdensome and substantially slow down the time for reporting till they are part of the official record (we can imagine fully automating reporting form Pgsurround.pl in the future if we go with proposal 1)

What I'm asking for.
1. A yes or no if both of you are okay if I start submitting records I hand review to the repository
2. A yes or no if both of you are okay with me changing the website to automatically submit records after verification.
Sorry for my absence from the scene. Up to my eyes with a variety of things.

I'm really happy to say yes to both. I have time to find records, but no real time (or skills) to get involved with the GitHub end of things.
robert44444uk is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Prime counting function records D. B. Staple Computer Science & Computational Number Theory 48 2020-10-28 07:54
records for primes 3.14159 Information & Answers 8 2018-12-09 00:08
Records for complete factorisation Brian-E Math 25 2009-12-16 21:40
gmp-ecm records page question yqiang GMP-ECM 6 2007-05-18 12:23
Records in January wblipp ElevenSmooth 10 2004-03-22 01:26

All times are UTC. The time now is 19:42.

Fri Oct 30 19:42:16 UTC 2020 up 50 days, 16:53, 1 user, load averages: 1.79, 1.61, 1.70

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.