Some Somewhat Easier n^i Sequences Available for Termination - Page 77

gd_barnes · 2022-09-14, 20:36

Ed,

I've decided that I want to ECM all of what is remaining of our same-parity exponents / double-square bases. This would only involve exponents at starting size > 185 digits. Previously I only went up to 185 digits.

So I'll be asking about some reservations on high exponents before I get started. Are you done with base 98? It doesn't look like it's had any work in a few months. If you'd still like to keep it, would you be OK with me running ECM to t35 on them? As high as they are assuming that you have not ECM'd everything, most would probably not advance and the ones that do would only likely add a few iterations.

This would also ECM a handful of the ones reserved for Curtis here that are at starting size > 185 digits. I'm thinking this is OK because sequences that have had no ECM run on them would not be good tests for his process. You or he can let me know your thoughts on that.

I'm looking forward to this one. Ever since I've gotten the automated wget and aliqueit processes set up, I can just zoom through these high sequences ECMing everything to t35 while also running aliquiet.exe and adding iterations without having to babysit them. That was the whole point in getting aliquet.exe running without Yafu. That process is what I've been using on the opposite-parity effort for all bases higher exponents that I've had going in the other thread. It's been great!

Gary

VBCurtis · 2022-09-14, 20:53

Quote:

Originally Posted by gd_barnes

This would also ECM a handful of the ones reserved for Curtis here that are at starting size > 185 digits. I'm thinking this is OK because sequences that have had no ECM run on them would not be good tests for his process. You or he can let me know your thoughts on that.
Gary

Which ones are those? I didn't know I had more work reserved in this project; I can finish up ones I have reserved, particularly if they're bigger starting terms in base 13.

gd_barnes · 2022-09-14, 21:01

Quote:

Originally Posted by VBCurtis

Which ones are those? I didn't know I had more work reserved in this project; I can finish up ones I have reserved, particularly if they're bigger starting terms in base 13.

See post #1 in the thread here for current reservations. There are 33 of them.

You had requested a steady supply of factors in the 138-147 digit range in post 683 so Ed said he'd feed you a steady supply of them in post 684. You did a few of them. Let us know if you are done with them. It's not necessary that you finish them. They are low priority due to their size.

EdH · 2022-09-14, 21:42

@Gary: I was thinking of releasing those left in base 98, myself, since they are larger than I'm currently working with - done.

@Curtis: Are you still interested in a stream of a certain range? Should we come up with some new? We can clear things and start fresh.

gd_barnes · 2022-09-14, 22:32

OK I will plan on going ahead with base 98. Most of these will be a quick run-thru similar to the opposite-parities.

This will eventually amount to all exponents on the project having "some" work done on them by someone.

I plan to start on Thursday or Friday.

VBCurtis · 2022-09-15, 03:12

I'm solving a couple of NFS matrices this week, and the aliquot sequences I'm working are currently providing me C140-C150 composites. So, I'm good for a couple weeks.

I did an extensive test-sieve on a C142 with CADO's las, like 40 params settings. I now have yield and speed data for a variety of mfb settings, and need to combine that data with the estimates I've recorded for number of relations. I've learned that mfb and lambda settings are really hard to test because the number of relations required for the job changes too, so sec/rel is an illusion. I've also learned that ncurves doesn't need to be very big, and I=14 is clearly faster than I=13 at c142.

The good news is that I can test 3LP at all the sizes I suspect may be useful. My plan is to find likely fastest 2LP and 3LP settings from C150 to C165, and run full jobs comparing those two options working up from the bottom until I find one where 3LP is a winner.

As I get into the C160s, I expect to have duelling params settings for you to test on full jobs on your farm.

gd_barnes · 2022-09-15, 03:17

So from our perspective, we can release anything that we previously had reserved in the C138-147 range? Is that correct? I just don't want to step on any toes with my ECM effort coming up in the next few days.

VBCurtis · 2022-09-15, 03:41

Gary-
Definitely. I'm not actively working on the list y'all posted for my benefit.
I'm still willing to crack something in the C140s to help along your efforts, but nothing needs to be left for my benefit.

EdH · 2022-09-15, 19:16

Quote:

Originally Posted by VBCurtis

. . .
As I get into the C160s, I expect to have duelling params settings for you to test on full jobs on your farm.

Sounds good. A real rough estimate for total completion of 160-170 digits is 16-85 hours, but I can have the harvest data before LA completes.

gd_barnes · 2022-09-17, 13:05

FactorDB just came back up in the last hour. I have a huge amount of data to enter both for same and opposite-parity sequences but will not be able to do that until early Sunday morning.

I have just enough time to enter my 3 same-parity terminations for sequences starting at > 185 digits here before signing off:

137^87, 139^89, and 191^83 terminate

137^87 was fun. It terminated at index=4 with a P179! This is a personal record. I believe it is the 3rd largest ever found on the project. Rich found a P183 for 197^83 at index=2 and Yoyo found a P182 for 200^79 at index=1. I can still say that I've found the largest one on any sequence for index > 2. :-)

Work is still ongoing on the > 185-digit effort. Right now, there looks like there might be only 1 or 2 more interesting sequences that could possibly terminate but no guarantee. But when everything is entered, all same-parity exponents / double-square bases on the project will have had open sequences ECM'd to at least t35. Ed, give me until later Sunday night to have all of them searched to C>131.

garambois · 2022-09-17, 14:12

Quote:

Originally Posted by gd_barnes

FactorDB just came back up in the last hour. I have a huge amount of data to enter both for same and opposite-parity sequences but will not be able to do that until early Sunday morning.

I won't be updating this weekend anyway.
Let's wait until the following week for the update.

2022-09-17, 13:05	#846
gd_barnes "Gary" May 2007 Overland Park, KS 2·3·11·179 Posts	FactorDB just came back up in the last hour. I have a huge amount of data to enter both for same and opposite-parity sequences but will not be able to do that until early Sunday morning. I have just enough time to enter my 3 same-parity terminations for sequences starting at > 185 digits here before signing off: 137^87, 139^89, and 191^83 terminate 137^87 was fun. It terminated at index=4 with a P179! This is a personal record. I believe it is the 3rd largest ever found on the project. Rich found a P183 for 197^83 at index=2 and Yoyo found a P182 for 200^79 at index=1. I can still say that I've found the largest one on any sequence for index > 2. :-) Work is still ongoing on the > 185-digit effort. Right now, there looks like there might be only 1 or 2 more interesting sequences that could possibly terminate but no guarantee. But when everything is entered, all same-parity exponents / double-square bases on the project will have had open sequences ECM'd to at least t35. Ed, give me until later Sunday night to have all of them searched to C>131. Last fiddled with by gd_barnes on 2022-09-17 at 13:08

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Unexpected termination of PM-1	Miszka	Software	22	2021-11-19 21:36
Easier pi(x) approximation	mathPuzzles	Math	8	2017-05-04 10:58
Would finding a definate Pi value easier if...	xtreme2k	Math	34	2013-09-09 23:54
Aliquot Termination Question - Largest Prime?	EdH	Aliquot Sequences	6	2010-04-06 00:12
A new termination below 100k	10metreh	Aliquot Sequences	0	2010-03-11 18:24

2022-09-14, 20:36	#837
gd_barnes "Gary" May 2007 Overland Park, KS 2·3·11·179 Posts	Ed, I've decided that I want to ECM all of what is remaining of our same-parity exponents / double-square bases. This would only involve exponents at starting size > 185 digits. Previously I only went up to 185 digits. So I'll be asking about some reservations on high exponents before I get started. Are you done with base 98? It doesn't look like it's had any work in a few months. If you'd still like to keep it, would you be OK with me running ECM to t35 on them? As high as they are assuming that you have not ECM'd everything, most would probably not advance and the ones that do would only likely add a few iterations. This would also ECM a handful of the ones reserved for Curtis here that are at starting size > 185 digits. I'm thinking this is OK because sequences that have had no ECM run on them would not be good tests for his process. You or he can let me know your thoughts on that. I'm looking forward to this one. Ever since I've gotten the automated wget and aliqueit processes set up, I can just zoom through these high sequences ECMing everything to t35 while also running aliquiet.exe and adding iterations without having to babysit them. That was the whole point in getting aliquet.exe running without Yafu. That process is what I've been using on the opposite-parity effort for all bases higher exponents that I've had going in the other thread. It's been great! Gary

2022-09-14, 21:42	#840
EdH "Ed Hall" Dec 2009 Adirondack Mtns 146F₁₆ Posts	@Gary: I was thinking of releasing those left in base 98, myself, since they are larger than I'm currently working with - done. @Curtis: Are you still interested in a stream of a certain range? Should we come up with some new? We can clear things and start fresh.

2022-09-14, 22:32	#841
gd_barnes "Gary" May 2007 Overland Park, KS 2×3×11×179 Posts	OK I will plan on going ahead with base 98. Most of these will be a quick run-thru similar to the opposite-parities. This will eventually amount to all exponents on the project having "some" work done on them by someone. I plan to start on Thursday or Friday.

2022-09-15, 03:12	#842
VBCurtis "Curtis" Feb 2005 Riverside, CA 1602₁₆ Posts	I'm solving a couple of NFS matrices this week, and the aliquot sequences I'm working are currently providing me C140-C150 composites. So, I'm good for a couple weeks. I did an extensive test-sieve on a C142 with CADO's las, like 40 params settings. I now have yield and speed data for a variety of mfb settings, and need to combine that data with the estimates I've recorded for number of relations. I've learned that mfb and lambda settings are really hard to test because the number of relations required for the job changes too, so sec/rel is an illusion. I've also learned that ncurves doesn't need to be very big, and I=14 is clearly faster than I=13 at c142. The good news is that I can test 3LP at all the sizes I suspect may be useful. My plan is to find likely fastest 2LP and 3LP settings from C150 to C165, and run full jobs comparing those two options working up from the bottom until I find one where 3LP is a winner. As I get into the C160s, I expect to have duelling params settings for you to test on full jobs on your farm.

2022-09-15, 03:17	#843
gd_barnes "Gary" May 2007 Overland Park, KS 27046₈ Posts	So from our perspective, we can release anything that we previously had reserved in the C138-147 range? Is that correct? I just don't want to step on any toes with my ECM effort coming up in the next few days.

2022-09-15, 03:41	#844
VBCurtis "Curtis" Feb 2005 Riverside, CA 2·3²·313 Posts	Gary- Definitely. I'm not actively working on the list y'all posted for my benefit. I'm still willing to crack something in the C140s to help along your efforts, but nothing needs to be left for my benefit.