Some CRUS stats - Page 2

rogue · 2010-11-12, 16:23

It doesn't appear that this table:

Top-20 conjectures with 1k remaining at lowest n by highest weight

is sorted correctly. It appears to be sorted by the test limit.

Could you add remaining k to these tables:

Top-20 conjectures tested to at least n=25K by highest CK
Top-20 conjectures tested to at least n=100K by highest CK

vmod · 2010-11-12, 19:14

Quote:

Originally Posted by rogue

It doesn't appear that this table:
Top-20 conjectures with 1k remaining at lowest n by highest weight is sorted correctly. It appears to be sorted by the test limit.

It was designed to be like that. When adding that table, my thinking was rather than a milestone table, to be more of a helper table for those deciding to take single-k work. So it came down to which criterion would be more helpful for someone choosing a 1k base reservation. Is it the weight or is one looking for bases at the lower limits?

I went for the second one. Looking first for the lower searched bases and then among them finding which one has the highest weight. So there is dual sorting. First by lowest search limit and at the same limit by highest weight (I realise top-20 isn't the most fitting title for something like that). When that table was first designed there was an abundance of 1k bases at n=25K. Since then, 1k'ers were heavily attacked and very soon all will be at n=100K or higher. So now there's much less scope to cherry pick low search ones and I guess weight is becoming the primary criterion.

I hope I explained my rationale behind the sorting criteria. If the preference is for a straightforward sort by weight (regardless of search limit) then I can change it. It's a much easier implementation than the current one.

Quote:

Could you add remaining k to these tables:
Top-20 conjectures tested to at least n=25K by highest CK
Top-20 conjectures tested to at least n=100K by highest CK

If by k remaining you mean the number of k's, not the actual k values, then yes, I can add an extra column.

mdettweiler · 2010-11-12, 19:47

The page is now online!

There is a link to it on the CRUS index page next to the one for Karsten's script, or you can access it directly at http://www.noprimeleftbehind.net/cru...crus-stats.htm. (I put the page in /crus/vstats/ instead of the more obvious /crus/stats/ because I wanted to leave the latter open in case we ever want to set up an NPLB-like stats database for CRUS in the future.) The script is run at 15-minute intervals, as with Karsten's.

Thanks vmod for writing a very cool and handy script!

rogue · 2010-11-12, 19:59

Quote:

Originally Posted by vmod

I hope I explained my rationale behind the sorting criteria. If the preference is for a straightforward sort by weight (regardless of search limit) then I can change it. It's a much easier implementation than the current one.

My issue was that it says by highest weight. I interpret that to mean that it is ordered by highest weight. Your list is ordered by test limit. A higher weight implies easier to prove. A lower test limit implies that less work has been done.

Quote:

Originally Posted by vmod

If by k remaining you mean the number of k's, not the actual k values, then yes, I can add an extra column.

Yes, the number or remaining k's, but not necessarily what those k's are.

rogue · 2011-01-26, 17:56

I have some ideas to add to the summarized CRUS stats page.

I would like to see a list of conjectures with the most k remaining. This would presume that all k have been tested to some limit, thus would eliminate b = 3 since it hasn't been fully tested. This list might provide some easy targets for removing k's.

I would also like to see a list of conjectures not tested to n=25000 ordered by n. Like the other list, it would presume that all k have been been tested to that n. This list would help get all started conjectures to n=25000.

I would also like to see a list of conjectures where different k have been tested to different n or where n is not divisible by 5000. This would be a list of "clean-up" work due to someone abandoning a reservation partway through.

Finally, could you change some of the column headings to add some color?

mdettweiler · 2011-01-27, 05:05

Another idea for the next version of the script: a summary of all bases with k=2 remaining.

Since these are generally considered "interesting" bases, it would be nice to have a list of them for people to take a whack at.

rogue · 2011-01-27, 16:10

Quote:

Originally Posted by mdettweiler

Another idea for the next version of the script: a summary of all bases with k=2 remaining.

Since these are generally considered "interesting" bases, it would be nice to have a list of them for people to take a whack at.

How about any k that is a power of 2?

mdettweiler · 2011-01-27, 17:30

Quote:

Originally Posted by rogue

How about any k that is a power of 2?

Hmm...why not?

(Probably won't be too much more work to implement.)

rogue · 2011-01-27, 19:34

Quote:

Originally Posted by mdettweiler

Hmm...why not?

(Probably won't be too much more work to implement.)

I would be willing to spend some time on the script to build such tables. I recommend that the scripts for the "condensed" and "summarized" pages be made publicly available (via a link from the CRUS main page) so that users will be able to add new features to them.

mdettweiler · 2011-01-27, 19:44

Quote:

Originally Posted by rogue

I would be willing to spend some time on the script to build such tables. I recommend that the scripts for the "condensed" and "summarized" pages be made publicly available (via a link from the CRUS main page) so that users will be able to add new features to them.

No problem.

All the files for the scripts are in the same directories as the HTML files they output; you can access their respective sources at:

http://www.noprimeleftbehind.net/crus/tab/ for Karsten's script
http://www.noprimeleftbehind.net/crus/vstats/ for vmod's script

vmod · 2011-01-27, 22:40

Quote:

Originally Posted by rogue

I have some ideas to add to the summarized CRUS stats page....

Most should be doable. I believe Gary includes new bases when they reach n=10K, so that seems like a reasonable cut-off for a "most k's remaining table".

Unreserved conjectures below n=25K are very few at the moment although such table could become useful in time when the growing CK of untested ones will be so big that some base starters will not continue to n=25K themselves.

Some cases of conjectures with k's at different test levels (because of big variation in formatting) are difficult to read correctly from the webpages. Showing bases that were released at some interim n level would be useful.

The color choice was made to resemble the rest of the crus pages. I have no objections adding more colors.

Presenting single k conjectures with only remaining k=2 (or powers of) should also be doable.

I'll have time for those additions over the weekend but Rogue if you did get the script and working on those, PM me or post so that I won't.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Sieving for CRUS	rebirther	Conjectures 'R Us	638	2021-06-15 07:55
BOINC effort for CRUS	gd_barnes	Conjectures 'R Us	75	2015-06-17 14:25
What are your CRUS plans?	rogue	Conjectures 'R Us	35	2013-11-09 09:03
how high will CRUS go	Mini-Geek	Conjectures 'R Us	1	2010-11-08 20:50
CSVs for stats available + New combined stats	opyrt	Prime Sierpinski Project	3	2010-05-31 08:13

2010-11-12, 16:23	#12
rogue "Mark" Apr 2003 Between here and the 2⁴·397 Posts	It doesn't appear that this table: Top-20 conjectures with 1k remaining at lowest n by highest weight is sorted correctly. It appears to be sorted by the test limit. Could you add remaining k to these tables: Top-20 conjectures tested to at least n=25K by highest CK Top-20 conjectures tested to at least n=100K by highest CK

2010-11-12, 19:47	#14
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	The page is now online! There is a link to it on the CRUS index page next to the one for Karsten's script, or you can access it directly at http://www.noprimeleftbehind.net/cru...crus-stats.htm. (I put the page in /crus/vstats/ instead of the more obvious /crus/stats/ because I wanted to leave the latter open in case we ever want to set up an NPLB-like stats database for CRUS in the future.) The script is run at 15-minute intervals, as with Karsten's. Thanks vmod for writing a very cool and handy script!

2011-01-26, 17:56	#16
rogue "Mark" Apr 2003 Between here and the 18D0₁₆ Posts	I have some ideas to add to the summarized CRUS stats page. I would like to see a list of conjectures with the most k remaining. This would presume that all k have been tested to some limit, thus would eliminate b = 3 since it hasn't been fully tested. This list might provide some easy targets for removing k's. I would also like to see a list of conjectures not tested to n=25000 ordered by n. Like the other list, it would presume that all k have been been tested to that n. This list would help get all started conjectures to n=25000. I would also like to see a list of conjectures where different k have been tested to different n or where n is not divisible by 5000. This would be a list of "clean-up" work due to someone abandoning a reservation partway through. Finally, could you change some of the column headings to add some color?

2011-01-27, 05:05	#17
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 1100001101001₂ Posts	Another idea for the next version of the script: a summary of all bases with k=2 remaining. Since these are generally considered "interesting" bases, it would be nice to have a list of them for people to take a whack at.