Tried out aliqueit.exe: ggnfs failing

[Greebley]

I discovered the aliqueit.exe program after following links on aliquot sequences. A very nice program. Might be able to help out too by extending some sequences. Ran into one issue mentioned toward end of post.

My only snafu is that my version of perl (5.0) was too old. Might want to mention that it needs a more recent one than that.

To test I wanted things outside the 'normal range':
I started with 1 google which immediately hits the most persistent "driver" of all and plummets to terminate in only 68 steps (does 2^0*1 meet the formal definition of a driver?) Its in the database now.

I then ran 1000152, the first sequence outside the 1-1000000 range that has potential. Much of it was already there, but I added to it up to 2196th iteration. At that point, ggnfs seems to run on it but it terminates and no factor seems to be found. Am I doing something wrong or have it set up wrong? I will paste the output below.

I put up to 2195 into the database if that helps.

<snip>
using previously found factor 29
c99: running 481 ecm curves at B1=1e6... <I canceled this step as it failed before>

Msieve v. 1.41
Thu Jun 04 10:28:03 2009
random seeds: bee3f918 b22c12da
factoring 612866359018419653283913545567941385719456545986288356312819216189730082882576641503539782906338843 (99 digits
)
searching for 15-digit factors
commencing number field sieve (99-digit input)
commencing number field sieve polynomial selection
time limit set to 0.33 hours
searching leading coefficients from 1 to 2898
deadline: 10 seconds per coefficient
coeff 60-600 64944 84427 84428 109756 lattice 1096
p 49956 84427 84430 142689 lattice 1096
batch 3032 84431
p 29559 49955 142692 241151 lattice 384
batch 1893 49957
save 1.606243e-009 -5.039069 19934.328509 3.272592e-009
save 1.606243e-009 -5.039069 19934.328509 3.272592e-009
save 1.586705e-009 -5.164989 23615.145883 3.245656e-009
save 1.586705e-009 -5.164989 23615.145883 3.245656e-009
save 1.614073e-009 -5.963712 42108.926159 3.261236e-009
save 1.550538e-009 -5.205398 26370.488442 3.198021e-009
save 1.627606e-009 -5.939588 43016.881015 3.265852e-009
save 1.693507e-009 -5.828912 36224.397046 3.359229e-009
save 1.667356e-009 -5.783249 36383.813052 3.303308e-009
save 1.726816e-009 -5.529360 26418.265475 3.407377e-009
save 1.596116e-009 -5.332439 28107.888915 3.210109e-009
save 1.742756e-009 -6.143119 43119.427847 3.400932e-009
deadline: 10 seconds per coefficient
coeff 660-1200 71637 93128 93129 121067 lattice 1430
p 71637 93128 93130 121070 lattice 1430
batch 1863 93131
p 42388 71636 121071 204612 lattice 846
batch 2589 71647
save 1.580569e-009 -5.625658 32294.955827 3.215647e-009
save 1.591554e-009 -5.122078 21137.864984 3.232050e-009
save 1.591554e-009 -5.122078 21137.864984 3.232050e-009
save 1.567773e-009 -4.954847 19159.093240 3.214142e-009
save 1.567773e-009 -4.954847 19159.093240 3.214142e-009
save 1.600586e-009 -4.985889 18713.612014 3.269583e-009
save 1.600586e-009 -4.985889 18713.612014 3.269583e-009
p 19293 42387 204616 449546 lattice 296
batch 2186 42391
deadline: 10 seconds per coefficient
coeff 1260-1800 75201 97761 97762 127090 lattice 1700
p 75201 97761 97764 127092 lattice 1700
batch 1930 97771
p 44496 75200 127094 214794 lattice 1006
batch 2706 75209
deadline: 10 seconds per coefficient
coeff 1860-2400 77689 100995 100996 131294 lattice 1930
p 77689 100000 102002 131295 lattice 1892
batch 1909 110456
p 45969 77688 131297 221893 lattice 1142
batch 2786 77689
save 1.534045e-009 -4.543784 18772.368793 3.177196e-009
save 1.534045e-009 -4.543784 18772.368793 3.177196e-009
save 1.610399e-009 -5.161535 22836.021562 3.262402e-009
save 1.632166e-009 -5.982672 32626.071714 3.252775e-009
save 1.610399e-009 -5.161535 22836.021562 3.262402e-009
save 1.772914e-009 -4.594570 11824.140387 3.448070e-009
save 1.772914e-009 -4.594570 11824.140387 3.448070e-009
save 1.650978e-009 -5.546875 26591.509169 3.298165e-009
save 1.680207e-009 -4.861616 18679.087991 3.365381e-009
save 1.841639e-009 -5.219852 12117.159947 3.523059e-009
save 1.576742e-009 -4.678259 19031.559756 3.234236e-009
save 1.647138e-009 -4.666045 14531.711056 3.336433e-009
save 1.567307e-009 -4.551315 12660.167734 3.219876e-009
save 1.614412e-009 -4.824186 20262.136586 3.264610e-009
save 1.614412e-009 -4.824186 20262.136586 3.264610e-009
save 1.680207e-009 -4.861616 18679.087991 3.365381e-009
save 1.576742e-009 -4.678259 19031.559756 3.234236e-009
save 1.794759e-009 -5.123591 17865.303031 3.512836e-009
save 1.647138e-009 -4.666045 14531.711056 3.336433e-009
save 1.616090e-009 -5.063949 15054.175026 3.276872e-009
save 1.567307e-009 -4.551315 12660.167734 3.219876e-009
save 1.574579e-009 -4.787392 17026.786509 3.256090e-009
save 1.540355e-009 -4.578158 13967.534174 3.196901e-009
save 1.540355e-009 -4.578158 13967.534174 3.196901e-009
deadline: 10 seconds per coefficient
coeff 2460-2880 79442 103274 103275 134257 lattice 2084
p 79442 100000 106657 134258 lattice 1954
batch 1719 112210
p 47006 79441 134260 226902 lattice 1233
batch 2800 79451
integrator failed 1.870373e-016 1.813653e-016
integrator failed 1.870373e-016 1.813653e-016
polynomial selection complete
R0: -12693016316435177922
R1: 12624079087
A0: -55155820712101931819305
A1: 73131720286124033583
A2: -18974146219940727
A3: 2175575125389
A4: 123362464
A5: 1860
skew 12117.16, size 1.841639e-009, alpha -5.219852, combined = 3.523059e-009
elapsed time 00:08:22
WARNING: gnfs failed to find a factor. This really shouldn't happen.
I'll just run ecm till the end of time or a factor turns up...
Let's hope you don't run out of disk space before either of those.
^C^C running 0 auto-increasing ecm curves...
D:\aliquot\aliqueit107>

[10metreh]

Originally Posted by Greebley

To test I wanted things outside the 'normal range':
I started with 1 google which immediately hits the most persistent "driver" of all and plummets to terminate in only 68 steps (does 2^0*1 meet the formal definition of a driver?) Its in the database now.

If 2 disappears, the sequence will probably terminate quickly.

I then ran 1000152, the first sequence outside the 1-1000000 range that has potential. Much of it was already there, but I added to it up to 2196th iteration. At that point, ggnfs seems to run on it but it terminates and no factor seems to be found. Am I doing something wrong or have it set up wrong? I will paste the output below.

I put up to 2195 into the database if that helps.

<snip>
using previously found factor 29
c99: running 481 ecm curves at B1=1e6... <I canceled this step as it failed before>

What exactly happened? What was the failure?

<snipped log>
WARNING: gnfs failed to find a factor. This really shouldn't happen.
I'll just run ecm till the end of time or a factor turns up...
Let's hope you don't run out of disk space before either of those.
^C^C running 0 auto-increasing ecm curves...
D:\aliquot\aliqueit107>

My only snafu is that my version of perl (5.0) was too old. Might want to mention that it needs a more recent one than that.

Ah-ha! That's it. factMsieve.pl is known not to work with 5.2 and older. The newest version (5.10 I think) should be fine.

BTW, that poly does not seem too good. I have a C99 in progress with an e value of 4.15e-9. 3.52e-9 is probably quite a bit slower - have a go at setting use_msieve_polyfind = false in aliqueit.ini and see if you come out with something better.

[schickel]

Originally Posted by Greebley

I discovered the aliqueit.exe program after following links on aliquot sequences. A very nice program. Might be able to help out too by extending some sequences. Ran into one issue mentioned toward end of post.

Hello and welcome to our little group! Glad you could find us. (Just out of curiosity, how hard was it to find us, and what prompted your search?)

My only snafu is that my version of perl (5.0) was too old. Might want to mention that it needs a more recent one than that.

I suppose a note to the effect that you need a working GGNFS installation could be added to the README file.

As far as your problem goes, everything up to this point in the log looks normal:

Msieve v. 1.41
Thu Jun 04 10:28:03 2009
random seeds: bee3f918 b22c12da
factoring 612866359018419653283913545567941385719456545986288356312819216189730082882576641503539782906338843 (99 digits
)
searching for 15-digit factors

This indicates that everything appears to operating normally. The program is searching for a polynomial to use for the GNFS run.

polynomial selection complete
R0: -12693016316435177922
R1: 12624079087
A0: -55155820712101931819305
A1: 73131720286124033583
A2: -18974146219940727
A3: 2175575125389
A4: 123362464
A5: 1860
skew 12117.16, size 1.841639e-009, alpha -5.219852, combined = 3.523059e-009
elapsed time 00:08:22
WARNING: gnfs failed to find a factor. This really shouldn't happen.
I'll just run ecm till the end of time or a factor turns up...
Let's hope you don't run out of disk space before either of those.
^C^C running 0 auto-increasing ecm curves...
D:\aliquot\aliqueit107>

Having failed at this point, it's hard to tell exactly what happened from this end. A couple of questions: have you completed a GNFS factorization before? Can you try pasting the line after gnfs_cmd = in the aliqueit.ini file?

Again, welcome to the group!

[Greebley]

I found the aliquot sequences interesting mathematically, and was following the links from other aliquot pages and found one to the forum.

I pasted the start of a run below so you could see the set up. It has the ggnfs line (and describes a second issue).


I have not successfully run ggnfs yet, which is why I think my settings are likely wrong. Is there a command I can give that will call ggnfs on a value known to work? I could test that and see what the result is.

These are the files I downloaded:
aliqueit107.zip
ecm622_win32_intel.zip
ggnfs-svn340-win32-p4.zip (do I want the svn360 version?)
msieve141_win32.zip
yafu-1.10-win32bin.zip

I am running vista, my system type is 32-bit Operating System. and the processor is:
Intel(R) Core(TM)2 Duo CPU E4600 @ 2.40GHz 2.40 GHz
2 GB of ram.

----------
I also got another problem with yafu (see bottom)

D:\aliquot\aliqueit107>aliqueit 232800
Reading config file...
ggnfs_clean_cmd = del cols*.* deps*.* factor.easy lpindex*.* rels*.* numFF.txt numrels.txt tmp.job spmat sp-index test.f
b matsave matsave.bak test.dat*
null_device = nul
makedir_cmd = mkdir
prefer_yafu = true
ecm_cmd = ecm
yafu_cmd = yafu
msieve_cmd = msieve
ggnfs_cmd = c:\perl\perl-5.10.0\perl.exe "d:\aliquot\gnfs\factMsieve.pl"
use_msieve_polyfind = true
detect_merge = false
trial_cutoff = 10000
gnfs_cutoff = 99
hide_output_limit = 60
neat_factor_limit_ecm = 45
neat_factor_limit_pm1 = 45
neat_factor_limit_pp1 = 40
result_file_prefix = alq_
log_file = aliqueit.log
save_log = true
ecm_tempfile = aliqueit_ecm_temp.log
ecm_depth = 45 10, 50 15, 55 20, 60 30, 65 40, 70 60, 73 90, 76 120, 79 150, 82 180, 83 200, 86 300, 90 400, 93 500, 96
600, 99 700, 102 800, 106 900, 110 1000, 114 1100, 118 1200, 122 1300
big_ecm_cutoff = 65
qs_k = 0.448
qs_m = -11.26
gnfs_k = 0.235
gnfs_m = 9.4
b1scale_ecm = 1.0
b2scale_ecm = 1.0
b1scale_pm1 = 1.0
b2scale_pm1 = 1.0
b1scale_pp1 = 1.0
b2scale_pp1 = 1.0
Precalcing primes for trial factoring...
seq = 232800
index = 0
value = 232800 (6 digits)
...
460 . c22 = 1522608819208455890462 = 2 * 1979 * 32957 * 2608559 * 4474703
461 . c21 = 762529205502959354338 = 2 * 29 * 15959 * 823801946694179
WARNING: factor doesn't divide n
WARNING: factor doesn't divide n
......
WARNING: factor doesn't divide n
^C9: running qs (yafu)...

The screen spammed the warning alternating the 'running qs (yafu)' message seemingly forever.

[schickel]

Originally Posted by Greebley

I found the aliquot sequences interesting mathematically, and was following the links from other aliquot pages and found one to the forum.

Glad you could find us. I've thought to myself a couple of times that I should add a link to the wikipedia article.....keep forgetting, though.

I pasted the start of a run below so you could see the set up. It has the ggnfs line (and describes a second issue).

D:\aliquot\aliqueit107>aliqueit 232800
Reading config file...
ggnfs_clean_cmd = del cols*.* deps*.* factor.easy lpindex*.* rels*.* numFF.txt numrels.txt tmp.job spmat sp-index test.fb matsave matsave.bak test.dat*

So far, so good.

ggnfs_cmd = c:\perl\perl-5.10.0\perl.exe "d:\aliquot\gnfs\factMsieve.pl"

This may be where your problem is. A couple of things to check:

- Is the path to your perl.exe correct?

- Did you set the paths in factMsieve.pl also? The ones to be concerned with are $GGNFS_BIN_PATH= this should point to the folder where the GGNFS binaries are located and $SYS_BIN_PATH= this should point to where "cat.exe" and "zip.exe" reside.

value = 232800 (6 digits)
...
460 . c22 = 1522608819208455890462 = 2 * 1979 * 32957 * 2608559 * 4474703
461 . c21 = 762529205502959354338 = 2 * 29 * 15959 * 823801946694179
WARNING: factor doesn't divide n
WARNING: factor doesn't divide n
......
WARNING: factor doesn't divide n
^C9: running qs (yafu)...

The screen spammed the warning alternating the 'running qs (yafu)' message seemingly forever.

This may be a problem that has cropped up a couple of times. there are some snall numbers which will fail to be factored by one or the other of msieve or yafu. If that happens, it gets caught in a loop, since it never recieves the factors of the number. the WARNING: factor doesn't divide n results from the continued perusal of the log file and trying to divide out the leading factors again; once they're divided out, they can't be divided out again. One way to kick start it is to factor the number externally, then feed the largest factor in by starting aliqueit this way:

aliqueit <seq> -e -f <factor>

This tells it to start the sequence, skip the ECM portion, and use the factor passed in on the command line.

If msieve or yafu fails, it should fall over to the other. Are the paths to both in aliqueit.ini set correctly?

[schickel]

Originally Posted by Greebley

I have not successfully run ggnfs yet, which is why I think my settings are likely wrong. Is there a command I can give that will call ggnfs on a value known to work? I could test that and see what the result is.

These are the files I downloaded:
aliqueit107.zip
ecm622_win32_intel.zip
ggnfs-svn340-win32-p4.zip (do I want the svn360 version?)
msieve141_win32.zip
yafu-1.10-win32bin.zip

I probably should have handled this and the previous message in the other order.....oh well.

You've got everything you need to bring terror to the heart of aliquot sequences everywhere. Just a couple of things to make sure you have everything laid out properly. If aliqueit is humming along mostly happy, you have the ECM executable in the right place, most likely. The next thing to check is yafu and msieve. Did you put a copy of both in the aliqueit folder? Also, as indicated in the previous post, did you change the paths in factmsieve.pl? (Actually, if it started the factorization, you don't have to worry, the path to perl.exe is set properly.) Starting the process and then failing is most likely an indication that it can't find the binaries for the GNFS portion.

Try this, create a file called test.n with this line in it (don't worry, we're not going to let it try to factor this):

n: 78541005650722049653086072205420438110929719018560861960885801681433105948145094453468591465656683293302279256523

Then, from the aliqueit directory, try calling the factmsieve script this way:

perl factmsieve.pl test

If it compalins about not finding the binaries, check the $GGNFS_BIN_PATH= line in factmsieve.pl.

[henryzz]

he has already said frank that he only has perl 5.0
factmsieve.pl needs 5.2 to run correctly
i expect what he is experiencing is what happens with a too early version of perl

[mklasson]

Originally Posted by Greebley

skew 12117.16, size 1.841639e-009, alpha -5.219852, combined = 3.523059e-009
elapsed time 00:08:22
ggnfs output should be here
WARNING: gnfs failed to find a factor. This really shouldn't happen.
I'll just run ecm till the end of time or a factor turns up...
Let's hope you don't run out of disk space before either of those.
^C^C running 0 auto-increasing ecm curves...
D:\aliquot\aliqueit107>

It doesn't seem as if ggnfs ever starts. If it did, you would probably get _some_ output from it at the indicated position. At least an error msg or two. My guess is your perl.exe path is wrong. Hopefully you've already solved the problem with the other guys' help.

Originally Posted by schickel

This may be a problem that has cropped up a couple of times. there are some snall numbers which will fail to be factored by one or the other of msieve or yafu. If that happens, it gets caught in a loop, since it never recieves the factors of the number.

That's a good reason for keeping both yafu and msieve available. If one accidentally fails the other will most likely crack it without problem. The endless looping should only happen if both fail (but see below).

Originally Posted by schickel

the WARNING: factor doesn't divide n results from the continued perusal of the log file and trying to divide out the leading factors again; once they're divided out, they can't be divided out again.

Unless I've forgotten how it works that specific reason is only applicable when you first start aliqueit, and not in the middle of a run. In this case I would guess yafu reports an invalid factor (see http://www.mersenneforum.org/showpos...&postcount=177), which fools aliqueit into thinking everything went ok and never giving msieve a chance to factor the number instead. The factor is then removed from n (or, as it was wrong, discarded and the warning msg printed), and yafu is run on the remaining composite (which in this case is the original number). So the same thing happens again and endless looping ensues. I should fix aliqueit to give the other qs program a go if a faulty factor was found...

[Greebley]

I created a aliqueit/bin directory, put that into my path and then copied ecm-<blah>.exe to ecm.exe in the bin. This seems to work as it runs ecm and yafu. Not sure if I have tried msieve yet, but msieve.exe is in the bin directory as well.

For perl it was definitely messed up. I installed strawberry perl 5.10 instead and changed the path in factMseive.pl. This is working much better.

However as show below it is having problems with cat.exe. I do have a cat.exe (and other unix tidbits) in c:\bin

D:\aliquot\ggnfs>perl factMsieve.pl test.n
-> ___________________________________________________________
-> | This is the factMsieve.pl script for GGNFS. |
-> | This program is copyright 2004, Chris Monico, and subject|
-> | to the terms of the GNU General Public License version 2.|
-> |__________________________________________________________|
-> This is client 1 of 1
-> Using 1 threads
-> Working with NAME=test...
-> Error: Polynomial file test.poly does not exist!
-> Found n=7854100565072204965308607220542043811092971901856086196088580168143310594814509445346859146565668329330227925
6523.
-> Attempting to run polyselect...
-> Selected default polsel parameters for 113 digit level.
-> Selected default factorization parameters for 112 digit level.
-> Selected lattice siever: d:/aliquot/ggnfs/gnfs-lasieve4I13e.exe
-> Searching leading coefficients from 1 to 1000.
=> "d:/aliquot/ggnfs/pol51m0b.exe" -b test.polsel.Falcon.4100 -v -v -p 4 -n 7.90E+016 -a 0 -A 1 > test.polsel.Falcon.41
00.log
-> =====================================================
-> Best score so far: 0.000000e+000 (goodScore=5.310000e-010)
-> =====================================================

-> Searching leading coefficients from 1001 to 2000.
=> "d:/aliquot/ggnfs/pol51m0b.exe" -b test.polsel.Falcon.4100 -v -v -p 4 -n 7.90E+016 -a 1 -A 2 > test.polsel.Falcon.41
00.log
=> "d:/aliquot/ggnfs/pol51opt.exe" -b test.polsel.Falcon.4100 -v -v -n 2.90E+015 -N 2.43E+013 -e 5.31E-010 > test.polse
l.Falcon.4100.log
'"/cat.exe"' is not recognized as an internal or external command,
operable program or batch file.
'"/cat.exe"' is not recognized as an internal or external command,
operable program or batch file.
-> =====================================================
-> Best score so far: 0.000000e+000 (goodScore=5.310000e-010)
-> =====================================================

-> Searching leading coefficients from 2001 to 3000.
=> "d:/aliquot/ggnfs/pol51m0b.exe" -b test.polsel.Falcon.4100 -v -v -p 4 -n 7.90E+016 -a 2 -A 3 > test.polsel.Falcon.41
00.log
=> "d:/aliquot/ggnfs/pol51opt.exe" -b test.polsel.Falcon.4100 -v -v -n 2.90E+015 -N 2.43E+013 -e 5.31E-010 > test.polse
l.Falcon.4100.log

[schickel]

Originally Posted by Greebley

I created a aliqueit/bin directory, put that into my path and then copied ecm-<blah>.exe to ecm.exe in the bin. This seems to work as it runs ecm and yafu. Not sure if I have tried msieve yet, but msieve.exe is in the bin directory as well.

For perl it was definitely messed up. I installed strawberry perl 5.10 instead and changed the path in factMseive.pl. This is working much better.

Great, we're almost there.

However as show below it is having problems with cat.exe. I do have a cat.exe (and other unix tidbits) in c:\bin

That's the problem, then. The location of cat.exe is set with the variable $SYS_BIN_PATH=, rather than the first one. (You'll find it just after the 3 "use" lines....) Set that to ":/bin" and try again.

[Greebley]

I had to do one final thing and that was copy over the more recent msieve.exe into the ggnfs directory - the one that came with ggnfs was
old.

Once I did that it successfully factored using the general number field seive.

So to summarize, in order to get ggnfs working I needed to:

1) Make sure I had a recent (past 5.2) version of perl
2) Change the aliqueit.ini file entry for ggnfs_cmd to point to ggnfs's factMSieve.
3) Update factMSieve.pl ggnfs_bin_path (top)
4) Make sure I had cat.exe.
5) Update factMSieve.pl sys_bin_path (further down)
6) For the 340 version of ggnfs (not sure about 360), I needed to copy over a later version of mSeive into the ggnfs_bin_path directory.

I think that was it. Might be worth adding to web page or docs of aliqueit.