Untested Sierp conjectures sorted by conjecture

[rogue]

Here is a list of all Sierpinski conjectures which have not been tested, ordered by ck.

I figured that people who are new to the search can pick bases from this list with low k to get their feet wet.

base          conjectured k
 705          10159692
 726          10923176
 481          11680548
1015          12079606
 400          12492354
1008          12730554
 585          13929512
 456          14836963
 541          15253776
 865          15460266
 586          21262902
 715          21508102
 346          22248598
 963          23276986
 630          24015859
 777          24088826
 358          27478218
 700          36945505
 801          45030296
 546          45119296
 925          59162602
 756          67836285
 891          80364072
 325          82786558
 760         113906481
 141         129697332
 796         137085595
 210         147840103
 906         161291811
 271         168541962
 876         258073912
 195         289401986
  85         346334170
 345         356851424
 120         374876369
 690         395800653
 175         606727296
 966         924230558
 399        2525403854
 415        3088787196
 435        3699251266
  71        5917678826
 921       10596584690
 127       11254645362
 156       18406311208
 826       19730748315
 981       50234474650
 960      197938246042
 661     1185028551412
 946     1524340829428
 799     1864613598304
 511     3876595974396
 910     4989054506113
 855     7954937545864
 280    82035074042274

[Mini-Geek]

Originally Posted by rogue

 839        4
 854        4
 139        6

These have been tested and proven according to http://www.noprimeleftbehind.net/cru...onjectures.htm
I haven't checked many others.
How did you make this list that those wouldn't be excluded? Or was it just a mistake?

[rogue]

Originally Posted by Mini-Geek

These have been tested and proven according to http://www.noprimeleftbehind.net/cru...onjectures.htm
I haven't checked many others.
How did you make this list that those wouldn't be excluded? Or was it just a mistake?

Most likely an edit error on my part. I started with a spreadsheet listing the conjectures for bases <= 1024, then removed rows from the linked page. I wouldn't be surprised if my eyes got a little buggy in the process.

[Mini-Geek]

Originally Posted by rogue

Most likely an edit error on my part. I started with a spreadsheet listing the conjectures for bases <= 1024, then removed rows from the linked page. I wouldn't be surprised if my eyes got a little buggy in the process.

You did it manually? Hmm...that's pretty prone to error, especially over ~2000 lines. I'll do it automated, and attach the results...
Done. (It doesn't have the nice-looking number of spaces like your list did, just a single space between each one. We'd probably want it looking like yours did before replacing it in the posts here.) For the record (so everyone else and I know how to do it in the future), here's how I did it:

get the list of conjectures, sorted by k then b in the format "b k" in one file, and
the list of bases that have been worked (from CRUS site) in another file (same format and sorting),
run "diff sierp-conjectures.csv sierp-worked.csv > sierp-diff.txt" (diff from cygwin/unix), with the two file names plugged in correctly
find/replace to remove all "> " from sierp-diff.txt
import sierp-diff.txt into OOo calc, splitting on 'd' and ',' (this will make almost all extraneous lines have more than one column, allowing us to get just the part we want)
sort by column B then A
remove all lines we don't want (which will be grouped together due to previous sorting)
now split the remaining columns on a space, sort by column B (the conj. k) then A (the base), and save as a .csv using a space to separate the lines
repeat for the other side (Riesel/Sierp)

Tada! Yeah, it's quite a bit, but it practically eliminates user error. And I'm sure it's gotta be WAY faster than going through the list of bases that have been worked manually. I found this many discrepancies between our lists: (all are ones that you included but I didn't, unless otherwise noted; all likely due to human error; someone might want to check these discrepancies from the original files and pages to be 100% sure which of us made the mistake)
Riesel: 5 diffs

 142        12
 137        22
 200        68
 298       116
 241     15918

Sierp: 13 diffs

 839        4
 854        4
 139        6
 146        8
 142       12
 129       14
 110       38
 122       40
962 106 (on my list only)
 961     1000
 138     2781
 143     7628
 211    20238

[rogue]

That's great! I didn't think strongly about an automated way of doing it, but that clearly is an easier way to do what I tried to do. Now if Gary would go and update the first and second posts in these threads...

[MyDogBuster]

Tim, Now if there was a way to automate that, I'm sure Gary or Max could run that stuff at least once a week or so. Nice job by the way. Those lists are nice, but Gary will want automation or they probably won't be maintained.

[Mini-Geek]

Originally Posted by MyDogBuster

Tim, Now if there was a way to automate that, I'm sure Gary or Max could run that stuff at least once a week or so. Nice job by the way. Those lists are nice, but Gary will want automation or they probably won't be maintained.

Yeah...reminds me of the top 5000 change list over at NPLB. Automating it could be possible, but is probably too much trouble to be worth it. It's not that there's anything that couldn't be automated fairly easily. It's just that putting it all together could be pretty tiresome.
If anyone's interested, it would probably be a good starting point to make a Perl script that takes the two sorted lists and outputs the difference directly in the format we want. I could see this being far easier than my method (and done entirely within a simple perl script, instead of the convoluted back-and-forth going on now involving multiple programs). One of the lists is fixed (i.e. the list of conjectures will almost never change) after it's prepared once, so we wouldn't need to consider that. Then the only other things you need are an automated way to get the sorted list of worked k's (this might be the hard part; in my method I copy/paste it from Firefox into OOo calc, a script would likely have to look at the HTML itself) and to do something with the output automatically (should be pretty easy, assuming it's being run as a cron job on the server where the result's hosted).

[gd_barnes]

Thanks again for your tireless efforts guys. I'll correct the postings later tonight.

[gd_barnes]

After thinking about this for a bit and after adding and updating about 10-15 new and already-tested bases on the web pages, I realized that it's going to add yet another step to the updates to manually update the 1st two posts in these two threads. It would be very difficult to keep them constantly in sync, especially as multiple small bases are done.

If anyone can automate this and post an updated list every few days based on testing shown on the web pages, I'll update the posts.

If we don't automate it, I might suggest something: List ALL bases sorted by conjecture and show if each base is proven or not. That would be a fairly quick one-time thing and it would be fairly easy to manually keep such a post updated whenver a conjecture becomes proven.

"Untested Sierpinski conjectures with sorted by k" seemed a little confusing as to what this is so I changed the 2 thread names slightly.


Gary

[gd_barnes]

Attached is something that I did a long time back. All Sierp conjectures sorted by size, that is conjectured-k.

I just put Robert's Sierp conjectures text file into a spreadsheet and quickly used formulas to eliminate the period, cover, and blank lines, and split out the base and conjecture into separate cells. I then sorted it. So it is a virtual guarantee that the list is all there and sorted correctly.

Perhaps someone can do something a little more with this.

[Mini-Geek]

Originally Posted by Mini-Geek

If anyone's interested, it would probably be a good starting point to make a Perl script that takes the two sorted lists and outputs the difference directly in the format we want. I could see this being far easier than my method (and done entirely within a simple perl script, instead of the convoluted back-and-forth going on now involving multiple programs).

I've made this as a Java program, (attached is a .jar of it, along with the source code, uncommented, and a readme) so the whole process is now pretty fast, but still requires the semi-manual step of getting the info from the web page into the format we need. Fortunately it's really not too hard (or long, or involved) to do that. I've tested it and its output matches up exactly with my other method.
I think it's easy enough now to not be too hard to regularly keep up to date (be it you doing it yourself or you having me, or someone else, do it after you update the pages).
Just in case you didn't know: you can double click a .jar to run it easily.

I'll attach some updated lists in the next post. (since I can only have one attachment per post and don't want it in the same as the Java script) There are a few updates for Riesel and none for Sierp.