2022 Project Goal discussion

[rogue]

Here are some potential 2022 goals:

1. Prove 5 bases
2. Find 100 top 5000 primes
3. Find 10 top 1000 primes
4. Find 5 top 500 primes
5. Test all base 2 (not powers of 2) conjectures to n=10M
6. Test all bases with 1 k remaining to n>=500K
7. Test all bases with 2 k remaining to n>=400K
8. Test all bases with <= 20 k remaining to n>= 200K
9. Test all bases with <= 500 k remaining to n>= 100K
10. Test all bases with <= 1000 k remaining to n>= 50K
11. Test all bases with <= 2000 k remaining to n>= 25K
12. Test all bases with conjectured k < 5M to n>= 10K

Some of these goals are massive, but size of tests run a fairly large range from small tests (a few seconds), medium tests (up to a minute), to long tests (a few minutes, but less than an hour), to very long tests (more than a hour).

I agree that it should be pared down, but at least it starts the conversation.

[dannyridel]

Originally Posted by rogue

Here are some potential 2022 goals:

1. Prove 5 bases
2. Find 100 top 5000 primes
3. Find 10 top 1000 primes
4. Find 5 top 500 primes
5. Test all base 2 (not powers of 2) conjectures to n=10M
6. Test all bases with 1 k remaining to n>=500K
7. Test all bases with 2 k remaining to n>=400K
8. Test all bases with <= 20 k remaining to n>= 200K
9. Test all bases with <= 500 k remaining to n>= 100K
10. Test all bases with <= 1000 k remaining to n>= 50K
11. Test all bases with <= 2000 k remaining to n>= 25K
12. Test all bases with conjectured k < 5M to n>= 10K

Some of these goals are massive, but size of tests run a fairly large range from small tests (a few seconds), medium tests (up to a minute), to long tests (a few minutes, but less than an hour), to very long tests (more than a hour).

I agree that it should be pared down, but at least it starts the conversation.

I think we should focus more on the limit of k remaining? the last ones look very ambitious to me.

[gd_barnes]

Here is likely to be my thinking, which we've already gotten a start on in the recommended bases thread:

1. Prove 4 bases.
2. Find 75 top 3000 primes.
3. Find 10 top 1000 primes.
4. Find 5 top 500 primes.
5. Test all bases with <= 12 k's remaining to n=300K. (Could be increased to <= 15 k's remaining.)
6. Test all 2k bases < 500 to n=500K.
7. Test all 1k bases < 600 to n=600K.
8. Test R2 2nd conjecture to n=5M.
9. Test all R2/S2 even-odd k-n to n=10M. (Likely the toughest for BOINC...need to ask Reb about it.)

Optional:
10. Test all bases with <= 1500 k's remaining (at n=10K) to n=25K.
11. Test all bases with <= 7000 k's remaining (at n=2500) to n=10K.

In general these are somewhat more aggressive than 2021 since many of the goals were accomplished a little early.

[rogue]

Originally Posted by gd_barnes

Here is likely to be my thinking, which we've already gotten a start on in the recommended bases thread:

1. Prove 4 bases.
2. Find 75 top 3000 primes.
3. Find 10 top 1000 primes.
4. Find 5 top 500 primes.
5. Test all bases with <= 12 k's remaining to n=300K. (Could be increased to <= 15 k's remaining.)
6. Test all 2k bases < 500 to n=500K.
7. Test all 1k bases < 600 to n=600K.
8. Test R2 2nd conjecture to n=5M.
9. Test all R2/S2 even-odd k-n to n=10M. (Likely the toughest for BOINC...need to ask Reb about it.)

Optional:
10. Test all bases with <= 1500 k's remaining (at n=10K) to n=25K.
11. Test all bases with <= 7000 k's remaining (at n=2500) to n=10K.

In general these are somewhat more aggressive than 2021 since many of the goals were accomplished a little early.

I see that we agree on a few of these or are at least close.

I think that goal 5 seems easy since most bases with <= 10 k appear to be tested to that limit already.

For 10 I would change that to bases with <= 2000 k remaining. It only adds 5 bases.

How many bases fall into category 11? I don't see those on the unproven page.

There are some conjectures that will require a collective effort to test further (S3, S63, S66, R63, R66 to name a few). Since it is unlikely that one person can take the conjecture to a higher n, maybe we should have goals that do part of them. For example, S3 for k < 50G to n=50000 or R63 for k < 50M to n=50000.

In any case I think that the key is to give participants options based upon how long it takes to do a PRP test. Outside of goal 5, all of the other goals (thru 9) will require PRP testing mega-digit numbers and goal 5 is almost guaranteed to be testing numbers that are at least 500000 digits.

[gd_barnes]

OK we can make goals 10 and 11 as definite goals to add more variety.

There are 15 bases with <= 7000 k's remaining at n=2500. 11 bases with <= 6000 k's remaining at n=2500. (A manual count was required.) I feel that 15 of these is too many. We can lower that to bases with <= 6000 k's remaining.

There are 16 bases with <= 2000 k's remaining at n=10K. 11 bases with <= 1500 k's remaining at n=10K. Once again I think that 15 of these is too many. Let's stick with <= 1500 k's.

With 11 remaining on each, having as a goal slightly less than one of these per month seems like plenty.

Keep in mind that only individual searchers will be working on these and they are not very popular. n=10K-25K is a pretty significant effort for > 1000 k's. You and/or Reb may be the only ones working on one or both of these goals for extended periods. (I might do 1-2 of them each for the year.) Of course we can take them further than the goal if we wish and they become more popular.

Let's leave it open on whether goal 5 should be bases with <= 12 or <= 15 k's remaining to n=300K. There are quite a few in progress right now by BOINC and we can see more where they are at closer to the end of December. It is a significant effort, even for BOINC, to test so many k's for n=100K-300K.

Further adding to the last para., goal 6 will likely be the second largest effort of the group and a big one. I'd rather not have goal 5 get too much in its way.

I'm not a big fan of the collective effort on huge-conjectured bases, especially for yearly goals, as they always require a lot of admin effort. An exception has been R3 because that base is so prime and I believe it will ultimately compete with R2 on when it will be proven (centuries from now). :-)

[rogue]

Originally Posted by gd_barnes

OK we can make goals 10 and 11 as definite goals to add more variety.

There are 15 bases with <= 7000 k's remaining at n=2500. 11 bases with <= 6000 k's remaining at n=2500. (A manual count was required.) I feel that 15 of these is too many. We can lower that to bases with <= 6000 k's remaining.

There are 16 bases with <= 2000 k's remaining at n=10K. 11 bases with <= 1500 k's remaining at n=10K. Once again I think that 15 of these is too many. Let's stick with <= 1500 k's.

With 11 remaining on each, having as a goal slightly less than one of these per month seems like plenty.

Keep in mind that only individual searchers will be working on these and they are not very popular. n=10K-25K is a pretty significant effort for > 1000 k's. You and/or Reb may be the only ones working on one or both of these goals for extended periods. (I might do 1-2 of them each for the year.) Of course we can take them further than the goal if we wish and they become more popular.

Let's leave it open on whether goal 5 should be bases with <= 12 or <= 15 k's remaining to n=300K. There are quite a few in progress right now by BOINC and we can see more where they are at closer to the end of December. It is a significant effort, even for BOINC, to test so many k's for n=100K-300K.

Further adding to the last para., goal 6 will likely be the second largest effort of the group and a big one. I'd rather not have goal 5 get too much in its way.

I'm not a big fan of the collective effort on huge-conjectured bases, especially for yearly goals, as they always require a lot of admin effort. An exception has been R3 because that base is so prime and I believe it will ultimately compete with R2 on when it will be proven (centuries from now). :-)

I expect goal #7 to be a massive effort as well. As primes are found by goal #6, those k will automatically fall into goal #7.

Does SRBase have a page listing all of their reserved k and bases along with statuses of each?

[rudy235]

Originally Posted by rogue

I have no idea why you are making any suggestions when you do so little work for the project. I am actually uncertain if you have done any work for the project.

Then I won't suggest anything. Sorry!

[gd_barnes]

Originally Posted by rudy235

Then I won't suggest anything. Sorry!

Mark (rogue) was responding to our resident troll, Sweety. Rudy, you are welcome to offer suggestions.

I always delete Sweety's posts and their direct responses from CRUS because they are mostly meaningless and irritating. He also never does any work here.

[gd_barnes]

Originally Posted by rogue

I expect goal #7 to be a massive effort as well. As primes are found by goal #6, those k will automatically fall into goal #7.

Does SRBase have a page listing all of their reserved k and bases along with statuses of each?

Very few from #6 will trickle down into #7; maybe 10-15% of them (3-4 bases). I will pick up some of the trickle-downers myself if necessary when the time comes.

#7 is only one k each and ALL of the Sierp bases are only the smaller n-range of n=500K-600K. BOINC (and maybe some individuals) will knock these out quickly.

See the 1st post of the recommended bases thread at CRUS for all bases that BOINC has reserved and when they were reserved. Also you can get an idea of the number of bases remaining for goals #6 and 7 at this point.

You can also see the number of tests remaining and those handed out on the SRBase server status page here:
https://srbase.my-firewall.org/sr5/server_status.php

[rudy235]

Hello: more than goals as such, this would be my wish list.

A) Prime #1220 over 1,000,000 digits
B) Prime #210 over 2,000,000 digits
C) Prime #100 over 3,000,000 digits
D) At least 1 prime between 8,000,000 and 8,999,999 digits
E) At least 1 prime between 10,000,000 and 10,999,999 digits
F) Prime #5000 over 500,000 digits

The most difficult one I believe would be E) and the simplest F)

[gd_barnes]

Originally Posted by rudy235

Hello: more than goals as such, this would be my wish list.

A) Prime #1220 over 1,000,000 digits
B) Prime #210 over 2,000,000 digits
C) Prime #100 over 3,000,000 digits
D) At least 1 prime between 8,000,000 and 8,999,999 digits
E) At least 1 prime between 10,000,000 and 10,999,999 digits
F) Prime #5000 over 500,000 digits

The most difficult one I believe would be E) and the simplest F)

I believe you are talking about CRUS project goals, not personal goals.

A and F are covered by goals 2 and 3 in my first post of this thread. B and C are possible and effectively included in goal 4. B is maybe a 50-50 chance to happen in a year's time frame for us. (CRUS has found four 2M+ digit primes in its history.) C is unlikely to happen within a year but is possible. (CRUS has only found one 3M+ digit prime in its history.)

D and E are unrealistic at this point in time for all projects except PrimeGrid. (GIMPS has already done all of its tests of these sizes.) Even PrimeGrid is probably less than 50-50 to find one of those primes in a single year's time frame and they have far more resources than us.

The difficult part of setting goals with a specific time frame is setting ones that are realistic but challenging at the same time.