[QUOTE=rogue;631118]I was comparing apples to oranges. The results were okay. It was an issue with restarting srbsieve from a checkpoint.[/QUOTE]

OK. Are you wanting to re-reserve S586? If so, what test depth will you be running to?

[QUOTE=gd_barnes;631142]OK. Are you wanting to re-reserve S586? If so, what test depth will you be running to?[/QUOTE]

Probably. I would take to n=2500 initially.

R622 tested to n=450k (400-450k)

1 prime found, base proven (send only the 50 results, another lucky hit)

Results emailed - Base released

[QUOTE=rebirther;631216]R622 tested to n=450k (400-450k)

1 prime found, base proven (send only the 50 results, another lucky hit)

Results emailed - Base released[/QUOTE]

We are certainly due for some lucky hits this year! :smile:

S575 is complete to n=25K; 527 primes were found for n=10K-25K; 1823 k's remain; base released.

R652 is completed to n=10k

4144 primes found, 4946 remain

Primes attached, base released

I've just noticed that all my primes files for my assignments have a strange sorting order

Idk if its an issue since no one's ever complained about it, but it annoys me so I'm going to add the proper ORDER BY to my SQL now haha.

[QUOTE=pokemonlover123;631281]I've just noticed that all my primes files for my assignments have a strange sorting order

Idk if its an issue since no one's ever complained about it, but it annoys me so I'm going to add the proper ORDER BY to my SQL now haha.[/QUOTE]

It is preferred if they are sorted numerically either by k-value or n-value. Either one works. I've often gotten files sorted alphanumerically so I have a spreadsheet that is able to parse out the k and n-values and sort them properly if the files aren't too big. It is an extra step that I go through so if you can post them sorted numerically that is greatly appreciated. :-)

[QUOTE=gd_barnes;631314]It is preferred if they are sorted numerically either by k-value or n-value. Either one works. I've often gotten files sorted alphanumerically so I have a spreadsheet that is able to parse out the k and n-values and sort them properly if the files aren't too big. It is an extra step that I go through so if you can post them sorted numerically that is greatly appreciated. :-)[/QUOTE]

Yep they'll be sorted in the future! I did an Inner Join on the candidate table since candidatetestresult doesn't have the k value seperate, then i just sorted on candidate.k :)

Progress update
 

K64 S920 from 300K to 500K
No new primes
base released-res attached

Reserving
S767 up to 550K

Reserving S510 to n=500k (300-500k) for SRBase

R1010 tested to n=300k (100-300k)

4 primes found, 16 remain

Results emailed - Base released

R871 tested to n=300k (100-300k)

7 primes found, 13 remain

Results emailed - Base released

Progress update
 

S767 at 555K
Extending reservation to 700K

Reserving S773 up to 500K

S510 tested to n=500k (300-500k)

nothing found, 2 remain

Results emailed - Base released

Reserving S533 to n=500k (400-500k) for SRBase

S897 tested to n=300k (100-300k)

4 primes found, 15 remain

Results emailed - Base released

S586 is complete to n=2500. The results were sent to Gary via e-mail.

Taking S865 to n=2500

[QUOTE=rogue;632574]Taking S865 to n=2500[/QUOTE]

You are using [I]n[/I] really loosely here. n=6 and n=2500. To [U]exactly[/U] what do these apply? I hate guessing! :no:

[QUOTE=storm5510;632577]You are using [I]n[/I] really loosely here. n=6 and n=2500. To [U]exactly[/U] what do these apply? I hate guessing! :no:[/QUOTE]

In the post I estimated the number remaining at n=6. I am taking all k for this conjecture to n=2500.

[QUOTE=storm5510;632577]You are using [I]n[/I] really loosely here. n=6 and n=2500. To [U]exactly[/U] what do these apply? I hate guessing! :no:[/QUOTE]

We use n really precisely here. n is the exponent, k is the coefficient of the base.
k * b ^ n +-1

There is no other meaning for n throughout this subforum.

[QUOTE=rogue;632555]S586 is complete to n=2500. The results were sent to Gary via e-mail.[/QUOTE]

[QUOTE=rogue;632574]Taking S865 to n=2500[/QUOTE]

Mark,

You are jumping the gun here. Since you never responded to it, perhaps you missed this post:

[url]https://mersenneforum.org/showpost.php?p=631848&postcount=106[/url]

I am not ready to accept your results from S586. As best as I can tell, many different versions of various programs were used in running the base.

There has been a tendency on this project in the past to run very lightly-tested programs, which have often created a lot of problems. In the past few years, I've done my best to avoid that and I'd like to avoid it in the future.

I'm still deciding in my mind whether I want to do a full double-check myself or ask you to re-run it.

Although I feel that srbsieve and its associated programs are now good enough, I'm fairly convinced that they were not during part of your run.

For this reason, I'd like to ask that you release S865.

Best for the project is if you start from scratch and double-check yourself on S586 with the latest versions of srbsieve and everything else on the entire run.

Gary

[QUOTE=storm5510;632577]You are using [I]n[/I] really loosely here. n=6 and n=2500. To [U]exactly[/U] what do these apply? I hate guessing! :no:[/QUOTE]

As Curtis stated, we never use n-value loosely here. The n-value is the one thing that has a very exact definition in all cases for this project and for many others on the Mersenneforum. It is always the exponent. Perhaps an example would help:

When you see this on the project pages for Riesel base 30:
659 (500K)

It means that k=659 has been tested to n=500000 for R30. More specifically:

659*30^500000-1 is the highest test that has been run for that form and no prime has been found. In other words, for all n<=500000 either factors were found during sieving or primality tests showed that they were composite.

[QUOTE=gd_barnes;632595]As Curtis stated, we never use n-value loosely here. The n-value is the one thing that has a very exact definition in all cases for this project and for many others on the Mersenneforum. It is always the exponent..[/QUOTE]

If it [U]strictly[/U] applies to k * b ^ [B]n[/B] +/- c, then I get it clearly. I would not have been able to run sieves without it. I would take Mark's "n=6" as being [I]srbsieve's[/I] "maxNfbncsieve" value. It is familiar because I have used it in the past.

Please forgive me. Sometimes, my comprehension abilities drift into the gray areas. Once something becomes absolute then I can remember it, like the above.

There will come a time in the not-so-distant-future when I will no longer be able to do any of this. :paul:

[QUOTE=gd_barnes;632592]Mark,

You are jumping the gun here. Since you never responded to it, perhaps you missed this post:

[url]https://mersenneforum.org/showpost.php?p=631848&postcount=106[/url]

I am not ready to accept your results from S586. As best as I can tell, many different versions of various programs were used in running the base.

There has been a tendency on this project in the past to run very lightly-tested programs, which have often created a lot of problems. In the past few years, I've done my best to avoid that and I'd like to avoid it in the future.

I'm still deciding in my mind whether I want to do a full double-check myself or ask you to re-run it.

Although I feel that srbsieve and its associated programs are now good enough, I'm fairly convinced that they were not during part of your run.

For this reason, I'd like to ask that you release S865.

Best for the project is if you start from scratch and double-check yourself on S586 with the latest versions of srbsieve and everything else on the entire run.

Gary[/QUOTE]

I provided you files that account for every k in the conjecture. All of the k for n <= 6 are from the latest code. I used a version of srsieve2 that did not remove sequences incorrectly due to algebraic factorizations. In the worst case you would need to run the remaining k from n=7 to n=2500.

If you choose to throw away all of the work, then that is your choice. Just let me know what you choose to do.

R634 tested to n=600k (500-600k)

nothing found, 1 remain

Results emailed - Base released

[QUOTE=rogue;632609]I provided you files that account for every k in the conjecture. All of the k for n <= 6 are from the latest code. I used a version of srsieve2 that did not remove sequences incorrectly due to algebraic factorizations. In the worst case you would need to run the remaining k from n=7 to n=2500.

If you choose to throw away all of the work, then that is your choice. Just let me know what you choose to do.[/QUOTE]

I'm aware of all of that. I ran the starting bases script on the whole base to n=20 to get the counts on the GFN, MOB, and trivial files. All matched. Also after a little finagling of the primes files, I was able to compare your primes on the base for n<=20. All matched.

The problem is that a version of srbsieve that was unstable was used when the base was begun. I also asked in several posts if you would please slow down until we had a stable version.

After further consideration, I'd rather not do a double-check. Nearly all of my machines are tied up running the bad sieve file effort and I would like to keep them there almost exclusively for ~2 more months.

The work is not being thrown away. It was an extensive test, which was needed on new software. I will keep the results on hand and when someone runs it again, we will have a nice big comparison to make.

In the future, it would be better if we run new software on bases that have already been tested instead of on new bases where we have nothing to compare it to. Use the already existing bases as a comparison to get to a stable version, then run it on new bases.

Unless I hear otherwise, I will release S586. If you want to continue with S865, that is fine since srbsieve's only remaining issues are on smaller-conjectured and/or faster testing bases. Let me know what you want to do.

FYI, I did re-run srbsieve for everything thru the first phase on S586 before I submitted. It didn't have the latest changes that you have requested in e-mail, but none of those impact S586. I used the output files from that run with srbverify to ensure that I didn't report any k twice and to verify that no k were missing.

[QUOTE=rogue;632634]FYI, I did re-run srbsieve for everything thru the first phase on S586 before I submitted. It didn't have the latest changes that you have requested in e-mail, but none of those impact S586. I used the output files from that run with srbverify to ensure that I didn't report any k twice and to verify that no k were missing.[/QUOTE]

OK, thanks for the info. I'll reconsider. I can do a double-check on 2-3 cores to n=1000 and be satisfied if everything matches. I'm unsure how long that will take.

I'll go ahead and leave S586 reserved for you until I'm done. If all checks well, I'll update it with your results.

Are you continuing with S865?

[QUOTE=gd_barnes;632653]OK, thanks for the info. I'll reconsider. I can do a double-check on 2-3 cores to n=1000 and be satisfied if everything matches. I'm unsure how long that will take.

I'll go ahead and leave S586 reserved for you until I'm done. If all checks well, I'll update it with your results.

Are you continuing with S865?[/QUOTE]

Yes.

Prime #2 from the bad sieve file effort. Not big enough for top-5000 but still noteworthy.

94*814^140039+1 is prime

7 k's now remain on S814 at n=300K.

[QUOTE=rogue;632555]S586 is complete to n=2500. The results were sent to Gary via e-mail.[/QUOTE]

S586 has been double-checked to n=1000. Everything checks out. The results to n=2500 have been posted on the pages.

Reserving R643 to n=500k (300-500k) for SRBase

Reserving R634 to n=1M (600k-1M) for SRBase

S938 tested to n=300k (100-300k)

2 primes found, 16 remain

Results emailed - Base released

s1017 update
 

s1017 to 267k and 275k respectively on two workers, no primes
second worker started at 270k

S533 tested to n=500k (400-500k)

nothing found, 2 remain

Results emailed - Base released

Starting R865

Reserving S612 from n=10K to n=25K (first reservation so let's hope I didn't bite off more than I can chew!)

S865 completed to n=2500. Results sent to Gary

[QUOTE=mrlol945;633877]Reserving S612 from n=10K to n=25K (first reservation so let's hope I didn't bite off more than I can chew!)[/QUOTE]

Welcome to CRUS! These efforts often take more time than what people think when they reserve them for the first time.

This is on our 2023 goals list so you have plenty of time to finish it.

You can get an idea of how long it will take by running a few k's for n=10K-25K and then using that time times the total number of k's remaining divided by the k's that you searched to compute a total time that it will take.

[QUOTE=rogue;633824]Starting R865[/QUOTE]

Looks like you'll have 4 k's eliminated by partial algebraic factors on this one. Now that I've updated them, you can see them on the CRUS pages.

Lately we've been very lucky on the new large-conjectured Riesel bases. Quite a few in a row had no algebraic factors to eliminate any k's and this one isn't bad. I'm waiting for the ball to drop on one of these where there are 100s of k's eliminated by algebraic factors. R400 was one example although as a base that was a perfect square, it meant that many k's were eliminated fully by algebraic factors. Those are much easier to pick out.

[QUOTE=gd_barnes;633918]Welcome to CRUS! These efforts often take more time than what people think when they reserve them for the first time.

This is on our 2023 goals list so you have plenty of time to finish it.

You can get an idea of how long it will take by running a few k's for n=10K-25K and then using that time times the total number of k's remaining divided by the k's that you searched to compute a total time that it will take.[/QUOTE]

Thanks a lot! Did the calculations and it looks like it will definitely be finished before the end of the year :smile:

Progress update
 

S767 at 605K
S773 at 430K

R643 tested to n=500k (300-500k)

nothing found, 2 remain

Results emailed - Base released

Reserving R667 to n=500k (300-500k) for SRBase
Reserving R686 to n=500k (300-500k) for SRBase
Reserving S530 to n=500k (300-500k) for SRBase
Reserving R665 to n=600k (400-600k) for SRBase

Progress update
 

S773 at 500K -base released
RES file attached

Reserving R975 to n=25K to test while sieving.

[QUOTE=k0r3;634536]Reserving R975 to n=25K to test while sieving.[/QUOTE]

Thanks for info. R975 is fully sieved. If you are sieving something else for CRUS, let us know what base so we don't duplicate anything.

[QUOTE=gd_barnes;634538]Thanks for info. R975 is fully sieved. If you are sieving something else for CRUS, let us know what base so we don't duplicate anything.[/QUOTE]

Sorry for the confusion, I meant while sieving S420 the other base I have reserved.

unreserving 731

[QUOTE=birtwistlecaleb;634564]unreserving 731[/QUOTE]

Thank you for letting us know. I have released R732.

Reserving R925 as new base using srbsieve up to 2.5k

R865 completed to n=2500. Files will be sent to Gary shortly. Here are the results (output from srbsieve):

[code]
Woohoo! All 21155947 ks are accounted for.
Final tally for R865 with k range from 1 to 21155947:
Trivials: 14103965
GFNs: 0
MOBs: 6154
Primes: 6809181
Remaining: 236647
[/code]

[QUOTE=rogue;634780]R865 completed to n=2500. Files will be sent to Gary shortly. Here are the results (output from srbsieve):

[code]
Woohoo! All 21155947 ks are accounted for.
Final tally for R865 with k range from 1 to 21155947:
Trivials: 14103965
GFNs: 0
MOBs: 6154
Primes: 6809181
Remaining: 236647
[/code][/QUOTE]

Thanks.

4 k's were proven composite by partial algebraic factors. Actual 236643 k's remaining.

R975 tested to n=25K. 662 found, 1583 remain.
Base released.

R925 tested to n=2.5k

104068 remain

Results emailed - Base released

Progress update
 

S767 at 660K
Extending only K4 S767 up to 750K

S530 tested to n=500k (300-500k)

nothing found, 2 remain

Results emailed - Base released

Starting S963.

R686 tested to n=500k (300-500k)

nothing found, 2 remain

Results emailed - Base released

Progress update
 

S767 at 700K -base released - res attached
K4 S767 at 715K

Reserving R760 as new base using srbsieve up to 2.5k

R665 tested to n=600k (400-600k)

nothing found, 1 remain

Results emailed - Base released

Reserving S620 to n=500k (300-500k) for SRBase
Reserving R668 to n=600k (400-600k) for SRBase
Reserving R679 to n=600k (400-600k) for SRBase

Reserving R684 to n=600k (400-600k) for SRBase

R760 up to n=2.5k done

[CODE]
Woohoo! All 22676276 ks are accounted for.
Final tally for R760 with k range from 1 to 22676276:
Trivials: 9530610
GFNs: 0
MOBs: 12325
Primes: 12933477
Remaining: 199864
All results have been sorted by ascending k in their respective files[/CODE]

Results emailed - Base released

Reserving R777 as new base using srbsieve up to 2.5k

R667 tested to n=500k (300-500k)

nothing found, 2 remain

Results emailed - Base released

[QUOTE=rebirther;636755]R760 up to n=2.5k done

[CODE]
Woohoo! All 22676276 ks are accounted for.
Final tally for R760 with k range from 1 to 22676276:
Trivials: 9530610
GFNs: 0
MOBs: 12325
Primes: 12933477
Remaining: 199864
All results have been sorted by ascending k in their respective files[/CODE]

Results emailed - Base released[/QUOTE]
4 k's with algebraic factors leaves 199860 k's remaining.

R634 tested to n=1M (600k-1M)

nothing found, 1 remain

Results emailed - Base released

S612 tested to n=25K, emailed the primes

720 found, 1432 remain

Time for the next sub-goal!

Reserving R732 from n=2.5K - 10K

Reserving R958 to 600K

R777 up to n=2.5k done

[CODE]Woohoo! All 23485095 k's are accounted for.
Final tally for R777 with k range from 1 to 23485095:
Odd k and base: 11742548
Trivials: 121057
Algebraic: 12
GFNs: 0
MOBs: 11513
Primes: 11225024
Remaining: 384941
All results have been sorted by ascending k in their respective files[/CODE]

Results emailed - Base released

Reserving R556 as new base using srbsieve up to 2.5k

R679 tested to n=600k (400-600k)

nothing found, 1 remain

Results emailed - Base released

Reserving S634 to n=500k (300-500k) for SRBase
Reserving S684 to n=500k (300-500k) for SRBase
Reserving R688 to n=600k (400-600k) for SRBase
Reserving R695 to n=600k (400-600k) for SRBase

R668 tested to n=600k (400-600k)

nothing found, 1 remain

Results emailed - Base released

S620 tested to n=500k (300-500k)

nothing found, 2 remain

Results emailed - Base released

R684 tested to n=600k (400-600k)

1 prime found, base proven (already reported)

Results emailed - Base released

Reserving R702 to n=600k (400-600k) for SRBase

Progress update
 

K4 S767 at 750K
extending reservation to 800K

R958 at 725K
extending reservation to 800K

S963 done to n=2500. Results (minus pl_trivial.txt) sent to Gary.

[code]
Woohoo! All 23276985 ks are accounted for.
Final tally for S963 with k range from 1 to 23276985:
Trivials: 12824120
GFNs: 0
MOBs: 8532
Primes: 9928921
Remaining: 515412
All results have been sorted by ascending k in their respective files
[/code]

Reserving R963. S963 was done with srbsieve 1.7. This will be with srbsieve 1.8.1.

R556 up to n=2.5k done

[CODE]Woohoo! All 26019698 k's are accounted for.
Final tally for R556 with k range from 1 to 26019698:
Trivials: 12517585
Algebraic: 5
GFNs: 0
MOBs: 18380
Primes: 13215097
Remaining: 268631
All results have been sorted by ascending k in their respective files[/CODE]

Results emailed - Base released

#7 & #8 from the bad sieve file effort:
3267*672^102919+1 is prime
1804*672^115286+1 is prime

S672 now has 12 k's remaining.

Sweety Emailed me with an unusual partial algebraic factorization of 13689*936^n-1 that makes a full covering set. k=13689 has now been removed as remaining on R936 and added to the algebraically factored k's list for the base. R936 now has 90 k's remaining at n=100K.

He also earlier pointed out that k=59904 has a similar factorization but since 59904=64*936, the k would have fallen into the MOB category in our search and so it didn't affect the remaining list. It should properly be shown as algebraic so that change was also made.

Originally only k=64 was listed as algebraic for the base. Now it has 3 k's listed.

Good job Sweety! (and you too!)

For 13689*936^n-1 with n being 3, 4, or 5 (mod 6), that is always divisible by (respectively) 37, 7, 109. You can add to your "condition 3" (the part about 7 is missing, albeit n=4|6 is covered by former 2 cases, I see you usually record the redundant cases too, so 7 being a factor is worth mentioning).

Also, as a general note, in all those conditions you say "factor of x", but the right words are "multiple of x", or "divisible by x". The 13689*936^3-1 is not a factor of 37. It is the other way around.

[QUOTE=LaurV;638489]Good job Sweety! (and you too!)

For 13689*936^n-1 with n being 3, 4, or 5 (mod 6), that is always divisible by (respectively) 37, 7, 109. You can add to your "condition 3" (the part about 7 is missing, albeit n=4|6 is covered by former 2 cases, I see you usually record the redundant cases too, so 7 being a factor is worth mentioning).

Also, as a general note, in all those conditions you say "factor of x", but the right words are "multiple of x", or "divisible by x". The 13689*936^3-1 is not a factor of 37. It is the other way around.[/QUOTE]

Sweety is fond of pointing out inconsistencies in the way that I show algebraic factorizations on the pages. I've made a number of changes to them to make them more consistent (in my mind) as a result of his Emails.

Here and in all cases, I only show the bare minimum of factors that are needed to make a full covering set. Although 7 covers 4 mod 6, since the k is a square, all even n are covered by the algebraic...and the algebraic is definitely needed so showing 7 would be redundant.

If you can point out some cases where I am showing more than the minimum of factors when partial algebraics make a full covering set, let me know. I will correct them.

I will show the "base case" of algebraics only for bases with conjectures >= 1000, even if no CRUS k's result in forms that are proven composite by partial algebraics by such a case. That is the situation with R936 for condition 1. I show those cases to queue myself or others about potential algebraics if they ever decide to search for k that are higher than the conjecture since the so-called base case here is by far the most common on the Riesel side. A further note: I do not show any other types of algebraic cases like the ones where the base contains a square (see R28) unless they actually affect a CRUS k. Such forms are far less common.

On showing multiple vs. factor, I think it's semantics. Here, k=13689 where n==(3 mod 6) contains a factor of 37. More specifically, it is stating that 13689*936^3-1, 13689*936^9-1, etc. have a factor of 37. I see what you are saying. 13689*936^3-1, 13689*936^9-1 are multiples of 37 but they also have factor of 37. So I think they are one and the same here.

The nice thing is that srsieve2 now detects the algebraic factors and eliminates all candidates. It should be harder to miss this sort of thing in the future.:smile:

[QUOTE=henryzz;638518]The nice thing is that srsieve2 now detects the algebraic factors and eliminates all candidates. It should be harder to miss this sort of thing in the future.:smile:[/QUOTE]

Sometimes it will not detect squares that are n!=(0 mod 2) or cubes that are n!=(0 mod 3), although it seems to in many cases. Regardless, those are much tougher to code for.

13689*936^n-1 has squares when n==(0 mod 2) and cubes when n==(1 mod 3). Impressively it finds them all!

Srsieve2 was changed within the last few months to do a much better job detecting algebraics and not remove n that it shouldn't. Had it been in its current state when R936 was tested, this k would have been caught as no longer remaining since it removes all of its terms.

Bases that contain squares such as R28 can have k's with algebraics where n==(1 mod 2). See 4032*28^n-1. Sometimes it misses those. It misses the algebraics on odd n's here.

I would not expect a sieving program to catch every single type of algebraic such as these. To do so could likely invite bugs that would remove n's that it should not. Srsieve2 does a very good job of catching > 99% of them now.

The best way to manually catch k's that can be removed now is to run all k's remaining through srsieve2 to some nominal sieve depth and look for k's with an abnormally low number of candidates remaining. 4032*28^n-1 has only 4 terms remaining for n<=1000 on a sieve to P=10e6. Counts like that almost always mean some algebraics are coming into play somewhere.

R688 tested to n=600k (400-600k)

nothing found, 1 remain

Results emailed - Base released

#9 from the bad sieve file effort, found by Max:

4612*991^166162+1 is prime

6 k's remain on S991.

Reserving S925 to n=2500.