New Cunningham Tables are ready. Please see sample and comment

garo · 2005-08-01, 11:39

Ihave posted the 12+ table here: http://www.mersenneforum.org/showthread.php?p=47884

Please let me know if the format is okay and if you have any questions. I will post the rest up later today. The number in brackets after each digit level given the proportion of curves that have been completed to those required to have a (1-1/e) chance of missing a factor at that level.

akruppa · 2005-08-01, 12:11

Looks very good (and like a lot of work!) to me.

Alex

Mystwalker · 2005-08-01, 13:44

I like it.

One thing I think of:
Maybe changing "11M(45)" to "11M / 45 digits" or something like that reduces the risk that someone new tries to link the 45 to the "optimal work done" count, as both are inside paranthesis...

garo · 2005-08-01, 15:03

Yes Alex it was a lot of work :) But rogue helped me a lot by sending his tables.

@Mystwalker- done.

Some other notes/thoughts are being posted on the "Introduction" thread.

wblipp · 2005-08-01, 15:49

1. I'd drop the counts entirely, keeping only the percentages in the tables.

2. This phrase is correct, but unintuitive for most people:

"proportion of curves that have been completed to those required to have a (1-1/e) chance of missing a factor at that level."

The following phrase is also correct, and I think more intuitive:

"average number of times a factor at that level would have been found."

It could optionally be followed by:

"If a factor would have been found, on average, "x" times, the probability it would have been entirely missed is e^-x."

or perhaps:

"If a factor would have been found, on average, "x" times, Poisson estimates the probability it would have been entirely missed is e^-x. We consider a level "complete" when x=1."

William

garo · 2005-08-01, 16:43

@wblipp:
I thought about dropping the counts entirely, but that would have meant a loss of information. If I had dropped counts entirely in v1.0 for instance, I would have had a major problem doing the translation from ECM5 to ECM6 curves. More information is better and in my opinion leaving the curve counts in makes the tables more intuitive and makes book-keeping a whole lot easier.

For your second suggestion, I agree that the wording in my initial post of this thread is confusing. But please look at explanation I posted here http://www.mersenneforum.org/showthread.php?t=4440 and let me know if it is okay.

@ALL: The complete tables are now up.

Mystwalker · 2005-08-01, 20:17

Quote:

Originally Posted by garo

I thought about dropping the counts entirely, but that would have meant a loss of information.

I also just thought about that - I think dropping the curve count is no problem, as long as you just keep it somewhere else (maybe even in a public place).
I see a slight problem with the curve count anyway: Some people use non-standard B2 bounds, which affect higher/lower digit ranges. But my guess is that the calculations are not that much off...

garo · 2005-08-01, 22:33

The counts are not 100% accurate as it is. I had to make several conversions from ECM5 to ECM6 curves. usually, we ask people to report the B2 they used and akruppa obliges with the conversion factor

I would also like to hear from some newbies like OmbooHankvald on whether they prefer to have the curve counts up or not. It makes very little difference for me as I just need to remove a few variables from a print statement!

xilman · 2005-08-02, 09:06

Quote:

Originally Posted by garo

The counts are not 100% accurate as it is. I had to make several conversions from ECM5 to ECM6 curves. usually, we ask people to report the B2 they used and akruppa obliges with the conversion factor

I would also like to hear from some newbies like OmbooHankvald on whether they prefer to have the curve counts up or not. It makes very little difference for me as I just need to remove a few variables from a print statement!

I am not a newbie!

However, I'd like to express my opinion, which seems to be in line with Mystwalker's, that the curve counts should be kept around. I'll go further --- they should be kept around in an easily findable place.

Paul

garo · 2005-08-02, 16:52

Well there is always the option of posting two separate tables, one with the counts and the other with proportion done. I had thought about that as well. It's just that many tables run into 3 or 4 posts due to the 10k char limit and cutting and pasting is very tedious. Another option is for me to mail them to xilman and he can host them on his page and we can put a link to it. Alternatively, I can host them on my page as well.

2005-08-01, 11:39	#1
garo Aug 2002 Termonfeckin, IE 2767₁₀ Posts	New Cunningham Tables are ready. Please see sample and comment Ihave posted the 12+ table here: http://www.mersenneforum.org/showthread.php?p=47884 Please let me know if the format is okay and if you have any questions. I will post the rest up later today. The number in brackets after each digit level given the proportion of curves that have been completed to those required to have a (1-1/e) chance of missing a factor at that level.

2005-08-01, 16:43	#6
garo Aug 2002 Termonfeckin, IE 2,767 Posts	@wblipp: I thought about dropping the counts entirely, but that would have meant a loss of information. If I had dropped counts entirely in v1.0 for instance, I would have had a major problem doing the translation from ECM5 to ECM6 curves. More information is better and in my opinion leaving the curve counts in makes the tables more intuitive and makes book-keeping a whole lot easier. For your second suggestion, I agree that the wording in my initial post of this thread is confusing. But please look at explanation I posted here http://www.mersenneforum.org/showthread.php?t=4440 and let me know if it is okay. @ALL: The complete tables are now up. Last fiddled with by garo on 2005-08-01 at 16:57

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Extensions to Cunningham tables	Raman	Cunningham Tables	87	2012-11-14 11:24
Extended Cunningham tables	Zeta-Flux	Factoring	2	2008-03-03 18:34
Cunningham Tables @mersenneforum.org v2.0	garo	Cunningham Tables	3	2006-07-04 08:00
New ECM Effort Tables for Cunningham Composites Ready	garo	Factoring	12	2005-09-06 07:53
A question about Cunningham tables	T.Rex	Factoring	14	2005-05-27 00:27

2005-08-01, 12:11	#2
akruppa "Nancy" Aug 2002 Alexandria 2,467 Posts	Looks very good (and like a lot of work!) to me. Alex

2005-08-01, 13:44	#3
Mystwalker Jul 2004 Potsdam, Germany 3×277 Posts	I like it. One thing I think of: Maybe changing "11M(45)" to "11M / 45 digits" or something like that reduces the risk that someone new tries to link the 45 to the "optimal work done" count, as both are inside paranthesis...

2005-08-01, 15:03	#4
garo Aug 2002 Termonfeckin, IE 2,767 Posts	Yes Alex it was a lot of work :) But rogue helped me a lot by sending his tables. @Mystwalker- done. Some other notes/thoughts are being posted on the "Introduction" thread.

2005-08-01, 15:49	#5
wblipp "William" May 2003 New Haven 2371₁₀ Posts	1. I'd drop the counts entirely, keeping only the percentages in the tables. 2. This phrase is correct, but unintuitive for most people: "proportion of curves that have been completed to those required to have a (1-1/e) chance of missing a factor at that level." The following phrase is also correct, and I think more intuitive: "average number of times a factor at that level would have been found." It could optionally be followed by: "If a factor would have been found, on average, "x" times, the probability it would have been entirely missed is e^-x." or perhaps: "If a factor would have been found, on average, "x" times, Poisson estimates the probability it would have been entirely missed is e^-x. We consider a level "complete" when x=1." William

2005-08-01, 22:33	#8
garo Aug 2002 Termonfeckin, IE 2,767 Posts	The counts are not 100% accurate as it is. I had to make several conversions from ECM5 to ECM6 curves. usually, we ask people to report the B2 they used and akruppa obliges with the conversion factor I would also like to hear from some newbies like OmbooHankvald on whether they prefer to have the curve counts up or not. It makes very little difference for me as I just need to remove a few variables from a print statement!

2005-08-02, 16:52	#10
garo Aug 2002 Termonfeckin, IE 101011001111₂ Posts	Well there is always the option of posting two separate tables, one with the counts and the other with proportion done. I had thought about that as well. It's just that many tables run into 3 or 4 posts due to the 10k char limit and cutting and pasting is very tedious. Another option is for me to mail them to xilman and he can host them on his page and we can put a link to it. Alternatively, I can host them on my page as well.