mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Conjectures 'R Us (https://www.mersenneforum.org/forumdisplay.php?f=81)
-   -   Riesel base 3 reservations/statuses/primes (https://www.mersenneforum.org/showthread.php?t=11151)

 KEP 2008-08-08 15:16

[QUOTE=gd_barnes;138831]I'm hoping that KEP will be done with k<500M on the Riesel side within the next month or so and we can attack that instead. I'll follow up with him and see where he is at.[/QUOTE]

Well as stated in old and recent e-mails to Gary, I had approximately 255,000 k's remaining at n<=500. I decided to split them into 26 sets of 10,000 k's or less. As tomorrow, the first of the 26 sets (covering k>=~490M to k<=500M) will be done. After that I'll start tackeling on full throttle the first set (covering k<=~22M or 27M, can't quite remember). After it completes, I'll put full throttle on set 2,3,4 etc... so in 1 month I should be able to hand over some completed sets covering to at least k<=100M. :smile:

KEP!

Ps. I expect (before removal of "redundant" k's) to end up with approximately 2,900 k's remaining at n<=25K for k<=500M. Which again will have us end up with approximately 365,400 k's remaining at n<=25K! How you decide to get to n<=25K, I'll let you decide, but I've mapped a plan to Gary that you can feel free to follow or discard as useless if something else occurs as more obvious and as being faster :smile:

 KEP 2008-09-05 22:13

[QUOTE=gd_barnes;141022]We were planning on running a team drive on that n-range that would run the same time as a team drive on the Riesel k=0-50M range. I guess the Riesel range is out now.

KEP, please reserve a smaller range that you absolutely KNOW that you can complete. See what Michaf is doing where he reserves a range, finishes it in a few days, then reserves another, etc.

Suggestion: Reserve k=50M-60M up to n=100K. Finish that. Reserve k=60M-70M to n=100K, etc. until you're ready to move on to something else. Thanks.

Why is Riesel base 3 more difficult than Sierp base 3? They're basically one and the same as far as searching for and finding primes. Riesel has a lower conjecture so it should be ultimately easier. The chance of finding primes on either side is going to be nearly identical. I don't understand your reasoning.

Why would a local power outage make it to where you can't run srfile to remove k's with primes? If you can't run srfile, then you certainly couldn't run srsieve and LLR on SIERP base 3 for the range you are wanting to reserve. Once again, your logic baffles me.

No need to answer the last two paragraphs as I was just pointing out a logic flaw but please respond to the paragraph asking you to reduce your reservation.

Gary[/QUOTE]

I'm discarding the work on the sierp base 3, and decides to put the Quad on the entire 500M k range for riesel base 3. So do not begin work on Riesel base 3 and do not consider this range lost! I ask for your patience, and ETA is somewhere in the future. Thank you.

I'm not sure if there is any logic, but in my humble experience, it seems to be a lot easier to prime the Sierpinski k's compared to priming the Riesel k's.

KEP

 KEP 2008-09-05 23:30

OK to point out exactly what I do:

1. Redoing the entire k<=500M range for Riesel base 3
2. Doing PRP tests up to n<=25,000
3. Verifies PRP tests
4. Does strict test of the PRP k's that turned out composite to n<=25,000

A request for the future:

-- Someone to develop af .pl script which can use "primes.out"-file from Michafs script as input and write the verified primes to "verprimes.out" and the composites to "vercomposit.out". --

Such a script, will make it easier to complete large ranges and reduce the need for reading through an entire log file on several GB. The wanted script needs to be developed, if any one would like to see me continue doing work on the Riesel base 3 conjecture.

And now I appologize, it appears that for Base 3 the PRP testing is a lot faster (more than just 30 %) so now I just awaits to see how many composites is among those k's remaining, but if it in the future can be done this fast, no doubt I'll continue if such a script as requested previously gets developed.

Regards

KEP

Ps. All k<=92997048 is in progress at the moment, so I hope everyone can accept the further delay, but with the lesson that I learned this time, I really think that a further and more speedy progress can be completed in the near future :smile:

Pps. It appears to be at least 7 times faster to do PRP tests compared to strict prime tests! So sorry Gary and anyone else for not listening or respecting your words of wisdom!

 gd_barnes 2008-09-06 04:54

[quote=KEP;141044]All k<=92997048 is in progress at the moment, so I hope everyone can accept the further delay, but with the lesson that I learned this time, I really think that a further and more speedy progress can be completed in the near future :smile:

Pps. It appears to be at least 7 times faster to do PRP tests compared to strict prime tests! So sorry Gary and anyone else for not listening or respecting your words of wisdom![/quote]

Please complete all k<100M to n=25K and send me the primes for n>=1K as well as the k's remaining. That is what will help us most and what I need to verify it all and start a team sieve and drive for the Riesel side.

In other words, don't concern yourself with anything for k>100M right now until k<=100M is complete.

Like you have concluded, just do PRP tests for now. When you have the entire k<100M range done, then do the proofs. For the few composite PRP's that you find, I'm sure you'll quickly find slightly higher primes for them.

Thanks,
Gary

 KEP 2008-09-06 08:51

[QUOTE=gd_barnes;141081]Please complete all k<100M to n=25K and send me the primes for n>=1K as well as the k's remaining. That is what will help us most and what I need to verify it all and start a team sieve and drive for the Riesel side.

In other words, don't concern yourself with anything for k>100M right now until k<=100M is complete.

Like you have concluded, just do PRP tests for now. When you have the entire k<100M range done, then do the proofs. For the few composite PRP's that you find, I'm sure you'll quickly find slightly higher primes for them.

Thanks,
Gary[/QUOTE]

I'll do PRP for all k<=100M to n<=25K. Current status at approximately 11 hours is n<=7,300 to n<=7800. So hopefully next weekend you can start up the mini-drive for base 3 Riesel. After I complete k<=100M, I'll send you the k's remaining and only the primes that you asked for unless you can accept that the primes is n>=501 in stead of n>=1K. Anyway the ranges is overall 33% complete. Now I'm just waiting for the result to see how many PRP is actually composite (if any) :smile:

KEP!

 gd_barnes 2008-09-06 08:56

[quote=KEP;141103]I'll do PRP for all k<=100M to n<=25K. Current status at approximately 11 hours is n<=7,300 to n<=7800. So hopefully next weekend you can start up the mini-drive for base 3 Riesel. After I complete k<=100M, I'll send you the k's remaining and only the primes that you asked for unless you can accept that the primes is n>=501 in stead of n>=1K. Anyway the ranges is overall 33% complete. Now I'm just waiting for the result to see how many PRP is actually composite (if any) :smile:

KEP![/quote]

Sure, go ahead and send me all primes n>=501. No problem.

 gd_barnes 2008-09-22 06:09

Riesel base 3 at k=93M and n=25K

KEP reported in an Email on 9/12 that he had completed Riesel base 3 to k=93M and n=25K. He also sent primes > 500 and all 407 k's that were remaining.

KEP, I've checked the k's remaining and your logic for removing k's that were multiples of the base. I could find no errors! Very nice work! :smile:

I did find one k remaining that had a previous top-5000 prime. It is:
75030224*3^133779-1

So there are officially 406 k's remaining for the range. The web pages now show the k's remaining and top 10 primes for the base.

I will reserve k=93M-100M up to n=25K to complete all k<100M. ETA is ~3-4 days on 2 slow cores. Sometime within a few weeks after that, we will start a mini-drive to take the entire k-range up to n=100K.

Gary

 KEP 2008-09-22 08:41

[QUOTE=gd_barnes;143387]KEP reported in an Email on 9/12 that he had completed Riesel base 3 to k=93M and n=25K. He also sent primes > 500 and all 407 k's that were remaining.

KEP, I've checked the k's remaining and your logic for removing k's that were multiples of the base. I could find no errors! Very nice work! :smile:

I did find one k remaining that had a previous top-5000 prime. It is:
75030224*3^133779-1

So there are officially 406 k's remaining for the range. The web pages now show the k's remaining and top 10 primes for the base.

I will reserve k=93M-100M up to n=25K to complete all k<100M. ETA is ~3-4 days on 2 slow cores. Sometime within a few weeks after that, we will start a mini-drive to take the entire k-range up to n=100K.

Gary[/QUOTE]

Hope you saw my e-mail before you got to far. Thanks for your timings in the other thread. I know some of the k's removed as result of q^n, is deffinently going to be doubles, since I copied a bit more of the list of k's removed than was actually nescessary :smile:

Good work on finding this extra prime, I had only checked for removal with q^n-1, and not for previously top5000 primes. Looking forward to be taking new work on the Riesel base 3 again in the future, as soon as I complete what I've for now reserved :smile:

Also in the future, you will recieve only entire 100M ranges. Just hope you didn't get to far on the Rb3 range, since it is still in work on 2 of my 6 cores, so progress will be steady and slow but I'll get there before long :smile:

KEP

 gd_barnes 2008-09-22 09:24

[quote=KEP;143399]Hope you saw my e-mail before you got to far. Thanks for your timings in the other thread. I know some of the k's removed as result of q^n, is deffinently going to be doubles, since I copied a bit more of the list of k's removed than was actually nescessary :smile:

Good work on finding this extra prime, I had only checked for removal with q^n-1, and not for previously top5000 primes. Looking forward to be taking new work on the Riesel base 3 again in the future, as soon as I complete what I've for now reserved :smile:

Also in the future, you will recieve only entire 100M ranges. Just hope you didn't get to far on the Rb3 range, since it is still in work on 2 of my 6 cores, so progress will be steady and slow but I'll get there before long :smile:

KEP[/quote]

Sounds good. Yes, I got your Email. I've stopped my effort now. Per my Email, please send the primes n>500 for your newest completed k-range.

Edit: Got the primes. Later today, I'll show you as having completed Riesel base 3 to k=134M and n=25K. Very nice work! :smile:

Gary

 gd_barnes 2008-09-22 23:51

KEP has now reported completion to k=134M and n=25K for Riesel base 3.

621 k's are remaining in the range. He is continuing on to k=500M and n=25K.

The web pages now show all appropriate info.

For all of you who like to find larger primes for the smallest k's remaining on a base, here's a great chance! :smile: Per the pages, the k's < 5M that are still remaining for this base are: 2794756, 3591446, 3677878, 4223272, and 4245112. They have only been searched to n=25K so far.

I'm going to start sieving k<100M for n=25K-100K for another mini-drive.

Gary

 henryzz 2008-09-23 16:02

All times are UTC. The time now is 19:55.