Index 1 Sequence Work for the "n^i" Aliquot Project - Page 9

EdH · 2022-10-07, 23:55

All sounds good and interesting. I'm not surprised that your numbers are better than my GPU numbers. I haven't really been impressed with my GPU ECM efforts, but that machine isn't doing anything else. The stage 2 work does take away some sievers, but it's only a few.

I've got one more c147 in work, but I don't know if it will finish before bed. If it does, I'll queue the c148s. I might be somewhat busy this weekend. Not sure yet.

gd_barnes · 2022-10-08, 00:09

The 2nd pass work included a 2nd pass of my own ECM effort too. There was only got a slightly higher percentage of 2nd pass splits on yours than mine. Breakdown:

Bases < 100: 10/145 6.9%
Bases > 100: 4/70 5.7%

The big area was the bases in the 90s with 5 splits.

Keep in mind that a large majority of these were lucky finds for factor sizes > 40. (11 out of 14) Some were completely outlandish factors for a t40. 2 of them in the 90s were a P45 and a P46! I always find it amusing when it finds such a large factor with so little effort.

Here is a breakdown of factor sizes found for bases < 100:

38, 39, 39*, 41*, 41, 42*, 42, 45*, 46*, 47(!)
* - bases in 90s

So you only missed 3 factors < 40 digits. I'd say that's pretty good.

Breakdown of factor sizes for bases > 100:
41, 42, 43, 46

The disadvantage of doing this is that it will reduce our t45 percentage of splits by a comparable amount. We may only get ~8-10% instead of ~14-16%.

EdH · 2022-10-08, 02:37

I think my ecmpi cluster is much better than the GPU setup, even though I have nowhere near the nodes to equal the GPU nodes. But the ecmpi cluster is the same cores that do sieving, so it's one or the other or split them for less effectiveness.

I'm going to queue up the c148s and see how far they get.

EdH · 2022-10-11, 02:20

I've queued the first three c149s. They'll probably be finished tomorrow.

gd_barnes · 2022-10-11, 06:43

I'll ECM to t45 the 9 remaining C149s starting with 84^83.

EdH · 2022-10-11, 13:52

We've taken a good chunk out of the original list, but those left are going to be a little more work than I may be interested in on a steady basis. It looks like I'm taking just about 5.5 hours per c149. Maybe I'll work them in with whatever else becomes interesting.

gd_barnes · 2022-10-11, 20:08

I was wondering when you thought the tests would get too long. We would need to promote this list somehow with some of the heavier hitters to knock out a bigger portion of it. I thought getting everything C<150 done would be a good stopping point.

In the mean time, I finished ECM to t45 on the remaining C149s starting with 84^83. Out of 9 sequences I got one very lucky split: 96^100 split with a P53-P97. At 8 digits above the t-level, that's luckier than even the P47 I found for my 2nd pass of the t40 run on all sequences.

VBCurtis · 2022-10-11, 20:23

I'm game to do a few ~C150 jobs. Want to assign me a couple numbers and I'll knock them out within a week or so?

EdH · 2022-10-12, 00:30

Thanks! I think just posting the one(s) you're working on should be fine. I've kind of shied away from creating another reservation thread. But, if you'd like some suggestions, 30^110 and 33^106 don't have any 3 factor ATM, so they might be of interest for further than just index 1 work later on. 30^110 especially falls into that category, since it's also a matched parity sequence, so it should directly terminate, but it's higher than our cutoff and one of many for that base. I'll skip those two if I do decide to run some c150s. I actually might stop at the c150 cutoff, at least for a while.

VBCurtis · 2022-10-12, 04:45

Thanks for the suggestions- saves me the mental energy of parsing the thread and double-checking that nobody else is working it.
I'll start with 30^110. Pretty sure I'll get to 33^106 also, within a week.

gd_barnes · 2022-10-12, 07:06

Quote:

Originally Posted by VBCurtis

Thanks for the suggestions- saves me the mental energy of parsing the thread and double-checking that nobody else is working it.
I'll start with 30^110. Pretty sure I'll get to 33^106 also, within a week.

To let you know: All sequences in this thread for factors C>=150 have been ECM'd to t40 twice.

2022-10-08, 00:09	#90
gd_barnes "Gary" May 2007 Overland Park, KS 2²·11·269 Posts	The 2nd pass work included a 2nd pass of my own ECM effort too. There was only got a slightly higher percentage of 2nd pass splits on yours than mine. Breakdown: Bases < 100: 10/145 6.9% Bases > 100: 4/70 5.7% The big area was the bases in the 90s with 5 splits. Keep in mind that a large majority of these were lucky finds for factor sizes > 40. (11 out of 14) Some were completely outlandish factors for a t40. 2 of them in the 90s were a P45 and a P46! I always find it amusing when it finds such a large factor with so little effort. Here is a breakdown of factor sizes found for bases < 100: 38, 39, 39, 41, 41, 42, 42, 45, 46, 47(!) - bases in 90s So you only missed 3 factors < 40 digits. I'd say that's pretty good. Breakdown of factor sizes for bases > 100: 41, 42, 43, 46 The disadvantage of doing this is that it will reduce our t45 percentage of splits by a comparable amount. We may only get ~8-10% instead of ~14-16%. Last fiddled with by gd_barnes on 2022-10-08 at 00:15

2022-10-11, 20:08	#95
gd_barnes "Gary" May 2007 Overland Park, KS 11836₁₀ Posts	I was wondering when you thought the tests would get too long. We would need to promote this list somehow with some of the heavier hitters to knock out a bigger portion of it. I thought getting everything C<150 done would be a good stopping point. In the mean time, I finished ECM to t45 on the remaining C149s starting with 84^83. Out of 9 sequences I got one very lucky split: 96^100 split with a P53-P97. At 8 digits above the t-level, that's luckier than even the P47 I found for my 2nd pass of the t40 run on all sequences. Last fiddled with by gd_barnes on 2022-10-11 at 20:19

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
A new idea for OEIS "triangle read by rows" sequence	sweety439	sweety439	4	2022-05-28 06:20
Aliquot Sequence 18528 - Team Project?	EdH	Aliquot Sequences	45	2021-06-27 12:30
Is there a copy of "the" aliquot tree anywhere?	Dubslow	Aliquot Sequences	11	2016-11-02 05:05
Possible extention to the "GPU to 72 Tool" project?	chalsall	GPU to 72	332	2012-01-04 01:45
Collaborative mathematics: the "polymath" project	Dougy	Math	11	2009-10-21 10:04

2022-10-07, 23:55	#89
EdH "Ed Hall" Dec 2009 Adirondack Mtns 1487₁₆ Posts	All sounds good and interesting. I'm not surprised that your numbers are better than my GPU numbers. I haven't really been impressed with my GPU ECM efforts, but that machine isn't doing anything else. The stage 2 work does take away some sievers, but it's only a few. I've got one more c147 in work, but I don't know if it will finish before bed. If it does, I'll queue the c148s. I might be somewhat busy this weekend. Not sure yet.

2022-10-08, 02:37	#91
EdH "Ed Hall" Dec 2009 Adirondack Mtns 5·1,051 Posts	I think my ecmpi cluster is much better than the GPU setup, even though I have nowhere near the nodes to equal the GPU nodes. But the ecmpi cluster is the same cores that do sieving, so it's one or the other or split them for less effectiveness. I'm going to queue up the c148s and see how far they get.

2022-10-11, 02:20	#92
EdH "Ed Hall" Dec 2009 Adirondack Mtns 5×1,051 Posts	I've queued the first three c149s. They'll probably be finished tomorrow.

2022-10-11, 06:43	#93
gd_barnes "Gary" May 2007 Overland Park, KS 2²·11·269 Posts	I'll ECM to t45 the 9 remaining C149s starting with 84^83.

2022-10-11, 13:52	#94
EdH "Ed Hall" Dec 2009 Adirondack Mtns 5·1,051 Posts	We've taken a good chunk out of the original list, but those left are going to be a little more work than I may be interested in on a steady basis. It looks like I'm taking just about 5.5 hours per c149. Maybe I'll work them in with whatever else becomes interesting.

2022-10-11, 20:23	#96
VBCurtis "Curtis" Feb 2005 Riverside, CA 2³·3·5·47 Posts	I'm game to do a few ~C150 jobs. Want to assign me a couple numbers and I'll knock them out within a week or so?

2022-10-12, 00:30	#97
EdH "Ed Hall" Dec 2009 Adirondack Mtns 5×1,051 Posts	Thanks! I think just posting the one(s) you're working on should be fine. I've kind of shied away from creating another reservation thread. But, if you'd like some suggestions, 30^110 and 33^106 don't have any 3 factor ATM, so they might be of interest for further than just index 1 work later on. 30^110 especially falls into that category, since it's also a matched parity sequence, so it should directly terminate, but it's higher than our cutoff and one of many for that base. I'll skip those two if I do decide to run some c150s. I actually might stop at the c150 cutoff, at least for a while.

2022-10-12, 04:45	#98
VBCurtis "Curtis" Feb 2005 Riverside, CA 2³·3·5·47 Posts	Thanks for the suggestions- saves me the mental energy of parsing the thread and double-checking that nobody else is working it. I'll start with 30^110. Pretty sure I'll get to 33^106 also, within a week.