Automated testing for all bases and k/n

kar_bon · 2008-10-28, 12:13

i wrote a script to automate the testing with LLR for Riesel sequnces base b.

Step by hand:
- test with LLR
- if prime/PRP is found, delete the sequence from test-file
- delete pairs tested by LLR
- create new test-file

all these steps are here in one script and a batch-file.

All you need:
- GAWK: a text-processing DOS-tool, which run special code-files (text like BASIC)
can be found here: http://sourceforge.net/project/showf...ckage_id=16431
only need 'gawk.exe' in bin-folder (v3.1.6, 352768 bytes)

- cLLR: the Command Line version of LLR
i use cLLR because there is an option 'StopOnSuccess': when a prime is found, LLR stops!
can be found here: http://jpenne.free.fr/

ToDo:
create a test-file like

Code:

108603139:M:1:2:258
82769 1024
124439 1024
132599 1024

put the four files (gawk.exe, cllr.exe, do.bat and read_res.awk) in a directory, include your test-file and call 'do' at a DOS prompt.

please read the file for instructions: there is a special name (created by srsieve) for the beginning test-file.

i included a test-file for the Riesel-Problem k*2^n-1 for k=1-509203 and n=1024-10k sieved to p=108603139.

the whole file was tested on my Pentium D DualCore at 2.8GHz in 70 minutes (see log-file).

any suggestions/errors?
please test and tell me you opinion.

the file can be downloded here:
www.rieselprime.org/dl/AutomatedLLR.zip

PS:
the phase from the beginning say n=1 upto ... i wrote a script for, too. but i have to make some changes before posting here. hope the next few weeks.

gd_barnes · 2008-10-29, 12:13

Quote:

Originally Posted by kar_bon

i wrote a script to automate the testing with LLR for Riesel sequnces base b.

Step by hand:
- test with LLR
- if prime/PRP is found, delete the sequence from test-file
- delete pairs tested by LLR
- create new test-file

all these steps are here in one script and a batch-file.

All you need:
- GAWK: a text-processing DOS-tool, which run special code-files (text like BASIC)
can be found here: http://sourceforge.net/project/showf...ckage_id=16431
only need 'gawk.exe' in bin-folder (v3.1.6, 352768 bytes)

- cLLR: the Command Line version of LLR
i use cLLR because there is an option 'StopOnSuccess': when a prime is found, LLR stops!
can be found here: http://jpenne.free.fr/

ToDo:
create a test-file like

Code:

108603139:M:1:2:258
82769 1024
124439 1024
132599 1024

put the four files (gawk.exe, cllr.exe, do.bat and read_res.awk) in a directory, include your test-file and call 'do' at a DOS prompt.

please read the file for instructions: there is a special name (created by srsieve) for the beginning test-file.

i included a test-file for the Riesel-Problem k*2^n-1 for k=1-509203 and n=1024-10k sieved to p=108603139.

the whole file was tested on my Pentium D DualCore at 2.8GHz in 70 minutes (see log-file).

any suggestions/errors?
please test and tell me you opinion.

the file can be downloded here:
www.rieselprime.org/dl/AutomatedLLR.zip

PS:
the phase from the beginning say n=1 upto ... i wrote a script for, too. but i have to make some changes before posting here. hope the next few weeks.

Outstanding work Karsten! Like I said in my PM response to you, I'll have to take a look at this. This is something we've definitely needed here for quite a while.

Gary

henryzz · 2008-10-30, 20:13

brilliant work

however i have found one bug
if you find a prime for k=2 then it removes sequences k=12, k=22 and so on when removing the k=2 sequence

this bug wouldnt affect the base 3 search currently though so it would be useful for that

a prp file containing all primes found would be nice

kar_bon · 2008-10-31, 10:34

Quote:

Originally Posted by henryzz

however i have found one bug
if you find a prime for k=2 then it removes sequences k=12, k=22 and so on when removing the k=2 sequence

this bug wouldnt affect the base 3 search currently though so it would be useful for that

a prp file containing all primes found would be nice

please give me your example you tested to eliminate this fault.

for the prime-file: yes, it would be better only the PRP in a file like LLR (only 'k n' pair in it, perhaps with header).
in the included "t17_b2_prim.txt" there are all primes with many other infos but not to handle so easy, you're right. will do that.
another point i think is to show the number of k's remain in the resultfile.

i think i can change this over the weekend.

Flatlander · 2008-10-31, 13:35

Maybe this could work with Phrot it someone could compile it with a StopOnPRP option? (The PRPs would have to be checked after the run, of course.)
Phrot is about 30% faster than LLR for the base-3 stuff I am working on.

kar_bon · 2008-10-31, 13:46

Quote:

Originally Posted by Flatlander

Maybe this could work with Phrot it someone could compile it with a StopOnPRP option? (The PRPs would have to be checked after the run, of course.)
Phrot is about 30% faster than LLR for the base-3 stuff I am working on.

would be quite a good chance for speeding up the checks.

BTW: i tested with cLLR only the PRP so i think i have to include a prime check with pfgw, too.

henryzz · 2008-10-31, 17:15

run this attachment through the script and i think u will see what i mean
12*15^1-1 is prime, in the sieve file and is the first prime for that k yet this prime is not found because the k=12 sequence is removed when the k=2 sequence is removed
if u look in the t17_b15_prim.txt afterwards you will notice that after k=2 is removed an enormous amount is removed

it is possible to avoid this bug by having all k values the same length

Flatlander · 2008-10-31, 19:26

Quote:

Originally Posted by kar_bon

would be quite a good chance for speeding up the checks.

BTW: i tested with cLLR only the PRP so i think i have to include a prime check with pfgw, too.

StopOnPRP is now available:
http://mersenneforum.org/showpost.ph...2&postcount=89

kar_bon · 2008-11-03, 09:42

Quote:

Originally Posted by henryzz

run this attachment through the script and i think u will see what i mean
12*15^1-1 is prime, in the sieve file and is the first prime for that k yet this prime is not found because the k=12 sequence is removed when the k=2 sequence is removed
if u look in the t17_b15_prim.txt afterwards you will notice that after k=2 is removed an enormous amount is removed

it is possible to avoid this bug by having all k values the same length

many even k-values will be removed because for n=1 they're primes!

look in the file b15_n1_prm.txt:
there're k=2 and 12 found as prime for n=1

in the resultfile b15_n1_res.txt these are trivially factored!

the script extract all k's not prime/PRP from this res-file: the chars from beginning to '*' (multiplication) will be written in the next test-file.
so all seems ok!

kar_bon · 2008-11-03, 09:45

Quote:

Originally Posted by Flatlander

StopOnPRP is now available:
http://mersenneforum.org/showpost.ph...2&postcount=89

yes, i saw this. but i need a WIN-version!

henryzz · 2008-11-03, 15:02

Quote:

Originally Posted by kar_bon

many even k-values will be removed because for n=1 they're primes!

look in the file b15_n1_prm.txt:
there're k=2 and 12 found as prime for n=1

in the resultfile b15_n1_res.txt these are trivially factored!

the script extract all k's not prime/PRP from this res-file: the chars from beginning to '*' (multiplication) will be written in the next test-file.
so all seems ok!

12*15^1-1 does not appear in the t17_b15_prim.txt
here is the t17_b15_prim.txt for the file i attached a few days ago

Code:

2008/11/03 14:54:15 found: 2 1 (6215 pairs remain, 5701 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:15 found: 4 1 (2950 pairs remain, 3264 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:15 found: 6 1 (1169 pairs remain, 1780 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:15 found: 10 1 (941 pairs remain, 227 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:16 found: 18 1 (753 pairs remain, 187 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:16 found: 30 1 (651 pairs remain, 101 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:16 found: 38 1 (545 pairs remain, 105 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:16 found: 70 1 (455 pairs remain, 89 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:16 found: 88 1 (380 pairs remain, 74 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:16 found: 198 1 (366 pairs remain, 13 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:16 found: 378 1 (354 pairs remain, 11 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:16 found: 390 1 (339 pairs remain, 14 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:16 found: 500 1 (332 pairs remain, 6 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:17 found: 558 1 (323 pairs remain, 8 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:17 found: 578 1 (318 pairs remain, 4 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:17 found: 580 1 (309 pairs remain, 8 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:17 found: 590 1 (297 pairs remain, 11 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:17 found: 598 1 (287 pairs remain, 9 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:17 found: 758 1 (274 pairs remain, 12 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:17 found: 780 1 (261 pairs remain, 12 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:17 found: 798 1 (248 pairs remain, 12 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:17 found: 808 1 (240 pairs remain, 7 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:18 found: 898 1 (228 pairs remain, 11 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:18 found: 900 1 (219 pairs remain, 8 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:18 found: 908 1 (205 pairs remain, 13 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:18 found: 950 1 (198 pairs remain, 6 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:18 found: 958 1 (186 pairs remain, 11 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:18 found: 978 1 (169 pairs remain, 16 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:18 found: 980 1 (160 pairs remain, 8 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:18 found: 998 1 (153 pairs remain, 6 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:19 found: 90 3 (120 pairs remain, 32 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:19 found: 98 3 (102 pairs remain, 17 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:19 found: 158 3 (89 pairs remain, 12 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:19 found: 508 3 (79 pairs remain, 9 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:19 found: 800 3 (66 pairs remain, 12 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:19 found: 150 5 (60 pairs remain, 5 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:19 found: 178 5 (52 pairs remain, 7 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:20 found: 878 5 (42 pairs remain, 9 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:20 found: 850 13 (36 pairs remain, 5 pairs deleted, 1 pairs LLR tested)
2008/11/03 14:54:20 found: 350 55 (23 pairs remain, 5 pairs deleted, 8 pairs LLR tested)
2008/11/03 14:54:20 found: 80 79 (7 pairs remain, 10 pairs deleted, 6 pairs LLR tested)
2008/11/03 14:54:21 found: (0 pairs remain, 6 pairs deleted, 1 pairs LLR tested)

there are no 2s in the k/n pairs found prime
also why did removing k=2 remove 5701 pairs
does this explain it better i have done an awful job of explaining myself

2008-10-28, 12:13	#1
kar_bon Mar 2006 Germany 101101011000₂ Posts	Automated testing for all bases and k/n i wrote a script to automate the testing with LLR for Riesel sequnces base b. Step by hand: - test with LLR - if prime/PRP is found, delete the sequence from test-file - delete pairs tested by LLR - create new test-file all these steps are here in one script and a batch-file. All you need: - GAWK: a text-processing DOS-tool, which run special code-files (text like BASIC) can be found here: http://sourceforge.net/project/showf...ckage_id=16431 only need 'gawk.exe' in bin-folder (v3.1.6, 352768 bytes) - cLLR: the Command Line version of LLR i use cLLR because there is an option 'StopOnSuccess': when a prime is found, LLR stops! can be found here: http://jpenne.free.fr/ ToDo: create a test-file like Code: 108603139:M:1:2:258 82769 1024 124439 1024 132599 1024 put the four files (gawk.exe, cllr.exe, do.bat and read_res.awk) in a directory, include your test-file and call 'do' at a DOS prompt. please read the file for instructions: there is a special name (created by srsieve) for the beginning test-file. i included a test-file for the Riesel-Problem k*2^n-1 for k=1-509203 and n=1024-10k sieved to p=108603139. the whole file was tested on my Pentium D DualCore at 2.8GHz in 70 minutes (see log-file). any suggestions/errors? please test and tell me you opinion. the file can be downloded here: www.rieselprime.org/dl/AutomatedLLR.zip PS: the phase from the beginning say n=1 upto ... i wrote a script for, too. but i have to make some changes before posting here. hope the next few weeks.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Automated LLR testing with LLRnet	mdettweiler	No Prime Left Behind	24	2011-11-04 19:20
Automated PRP using LLRNet	axn	Sierpinski/Riesel Base 5	73	2008-11-26 03:46
Automated primality testing with LLRnet	mdettweiler	Conjectures 'R Us	18	2008-03-04 00:06
Automated P-1 thoughts.	nucleon	Marin's Mersenne-aries	3	2004-03-25 02:45
Semi-automated P-1 testing	GP2	Marin's Mersenne-aries	2	2003-09-29 19:01

2008-10-30, 20:13	#3
henryzz Just call me Henry "David" Sep 2007 Cambridge (GMT/BST) 2³·3·5·7² Posts	brilliant work however i have found one bug if you find a prime for k=2 then it removes sequences k=12, k=22 and so on when removing the k=2 sequence this bug wouldnt affect the base 3 search currently though so it would be useful for that a prp file containing all primes found would be nice

2008-10-31, 13:35	#5
Flatlander I quite division it "Chris" Feb 2005 England 31·67 Posts	Maybe this could work with Phrot it someone could compile it with a StopOnPRP option? (The PRPs would have to be checked after the run, of course.) Phrot is about 30% faster than LLR for the base-3 stuff I am working on.