Picking a specific exponent to TF?

137ben · 2012-05-25, 19:59

Under advanced->p-1, you can select a specific exponent to p-1 factor. There is a similar option for ECM factoring and for LL tests, but not for TF. Is there a way to select a specific range and exponent to TF?

Dubslow · 2012-05-25, 20:29

I don't know about using the menus, but you can certainly modify worktodo.txt with a given assignment. Put something like this in it:

Code:

Factor=N/A,exponent,start-bit-level,end-bit-level

So if I wanted to do something like factor M61253499 from 69 to 71 bits, it'd be

Code:

Factor=N/A,61253499,69,71

Just be sure to exit Prime95 before modifying worktodo.txt. (That means Test/Exit, not just closing the window.)
See here for more info.

Uncwilly · 2012-05-26, 01:19

Be sure to check the status of the exponent like this:

http://www.mersenne.org/report_factoring_effort/?exp_lo=67890127&exp_hi=&bits_lo=0&bits_hi=999&txt=0&exassigned=1&B1=Get+Data
exp_lo = the starting exponent for your range
exp_hi = the ending exponent for your range
bits_lo & bits_hi = start and stop bit levels for your query
txt: 0 = show as a table, 1 = show as a comma delimited list (useful to make a worktodo)
exassigned: 0 = show all, 1 = show only unassigned (useful to make a worktodo)

petrw1 · 2012-05-26, 02:01

Quote:

Originally Posted by Uncwilly

exassigned: 0 = show all, 1 = show only unassigned (useful to make a worktodo)

No offense, but please be sure to only select UNassigned exponents.
It is frowned upon to complete work assignments that others have assigned to them and intend to complete and get credit for themselves.

137ben · 2012-05-26, 02:55

Hmm...
I'm trying to find a factor for M1061, and while there had been enormous amounts of ECM factoring, it has only been trial factored up to 2^63. I tried entering the line into the worktodo.txt as suggested, but I got an error in the communications thread telling me to use ECM instead of trial factoring for that particular exponent. Why?

dabaichi · 2012-05-26, 03:05

Quote:

Originally Posted by 137ben

Hmm...
I'm trying to find a factor for M1061, and while there had been enormous amounts of ECM factoring, it has only been trial factored up to 2^63. I tried entering the line into the worktodo.txt as suggested, but I got an error in the communications thread telling me to use ECM instead of trial factoring for that particular exponent. Why?

There is little hope to find a factor for M1061 via trial factoring. Also TF M1061 from 63 bits to 64 bits takes 1850 GHz-Days. So it's very inefficient to trial factor M1061.

Uncwilly · 2012-05-26, 03:06

Quote:

Originally Posted by 137ben

I'm trying to find a factor for M1061, and while there had been enormous amounts of ECM factoring, it has only been trial factored up to 2^63.

Threads about M1061
http://www.mersenneforum.org/showthread.php?t=14822
http://www.mersenneforum.org/showthread.php?t=14236
http://www.mersenneforum.org/showthread.php?t=3192
http://www.mersenneforum.org/showthread.php?t=6148
http://www.mersenneforum.org/showthread.php?t=12130
http://www.mersenneforum.org/showthread.php?t=16645

Dubslow · 2012-05-26, 03:16

Quote:

Originally Posted by 137ben

Hmm...
I'm trying to find a factor for M1061, and while there had been enormous amounts of ECM factoring, it has only been trial factored up to 2^63. I tried entering the line into the worktodo.txt as suggested, but I got an error in the communications thread telling me to use ECM instead of trial factoring for that particular exponent. Why?

The enormous amounts of ecm are enough so that we can say there is a very very very small chance that there are any factors less than 60 digits, which is 200 bits. TF is hopeless. (I'm guessing about the ECM, but it's at least 55 digits, and probably more than 60. I'll search the threads posted.)

Edit: http://mersenne.org/report_ecm/?ecm_...ecmnof_hi=2500
You can see that ECM to t60 is complete; that means there's a exp(-1)~=37% chance there is a factor of 60 digits (not bits). This also means that there is practically zero chance of a 55 digit or smaller factor.Thread ID 6148 in Uncwilly's post is also about getting it to t60. There has probably been some more ECM done beyond that. Note that Number Field Sieve has been started; sieving will likely be done in a month or two, and after that, it will be a few more months of linear algebra; we should see its factors (regardless of ECM/TF) before Christmas.

cheesehead · 2012-05-26, 10:16

Quote:

Originally Posted by 137ben

Hmm...
I'm trying to find a factor for M1061, and while there had been enormous amounts of ECM factoring, it has only been trial factored up to 2^63.

Do you understand that the time needed to TF from a given bit level to the next higher one is _inversely_ proportional to exponent size?

That is, TFing M1061 from 2^63 to 2^64 requires checking 50,000 times as many candidate factors as TFing M50xxxxxx from 2^63 to 2^64 does. Each candidate check may take, say, only 40% as long for M1061 as for M50xxxxxx, but that still leaves a factor of 20,000 in elapsed time.

137ben · 2012-05-26, 19:16

Ah, I see, that's what I was missing. Now I see why we haven't been TFing it. Thanks for the help.

2012-05-25, 19:59	#1
137ben Mar 2012 7 Posts	Picking a specific exponent to TF? Under advanced->p-1, you can select a specific exponent to p-1 factor. There is a similar option for ECM factoring and for LL tests, but not for TF. Is there a way to select a specific range and exponent to TF?

2012-05-25, 20:29	#2
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 3×29×83 Posts	I don't know about using the menus, but you can certainly modify worktodo.txt with a given assignment. Put something like this in it: Code: Factor=N/A,exponent,start-bit-level,end-bit-level So if I wanted to do something like factor M61253499 from 69 to 71 bits, it'd be Code: Factor=N/A,61253499,69,71 Just be sure to exit Prime95 before modifying worktodo.txt. (That means Test/Exit, not just closing the window.) See here for more info.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Test a Specific Exponent in Prime95. When it is 100% done, does it mean I find a new prime number?	king	Information & Answers	5	2018-02-21 18:15
Test a specific exponent with Prime95	Karl M Johnson	Software	12	2015-10-12 15:56
GNU specific macros	R.D. Silverman	Programming	17	2015-09-29 03:44
Cherry Picking	davieddy	Lounge	1	2013-01-08 16:40
Request specific exponent from primenet question	Unregistered	Information & Answers	19	2008-05-30 08:34

2012-05-26, 01:19	#3
Uncwilly 6809 > 6502 """"""""""""""""""" Aug 2003 101×103 Posts 10010100010101₂ Posts	Be sure to check the status of the exponent like this: http://www.mersenne.org/report_factoring_effort/?exp_lo=67890127&exp_hi=&bits_lo=0&bits_hi=999&txt=0&exassigned=1&B1=Get+Data exp_lo = the starting exponent for your range exp_hi = the ending exponent for your range bits_lo & bits_hi = start and stop bit levels for your query txt: 0 = show as a table, 1 = show as a comma delimited list (useful to make a worktodo) exassigned: 0 = show all, 1 = show only unassigned (useful to make a worktodo)

2012-05-26, 02:55	#5
137ben Mar 2012 7 Posts	Hmm... I'm trying to find a factor for M1061, and while there had been enormous amounts of ECM factoring, it has only been trial factored up to 2^63. I tried entering the line into the worktodo.txt as suggested, but I got an error in the communications thread telling me to use ECM instead of trial factoring for that particular exponent. Why?

2012-05-26, 19:16	#10
137ben Mar 2012 7 Posts	Ah, I see, that's what I was missing. Now I see why we haven't been TFing it. Thanks for the help.