Aliqueit.exe discussion

[gd_barnes]

Is there a way to make aliqueit stop when a certain size is reached?

Additionally is there a way to make it stop when a certain "hard" factor size (CSize) is reached? By hard factor size, I mean one in which survives the usual amount of ECM.

[Dubslow]

./aliqueit -h
Reading config file...
[...]
aliqueit v1.12 by Mikael Klasson 2009-2010, http://mklasson.com
usage: aliqueit [-i <index> <start_value>] [-f <factor>] [-s <start>] [-y] [-e] [-p] [-t] [-q]
                [-d <max_digits>] [-c <max_cofactor>] [-r <add_digits>] [-m <max_ecm_level>] [-b] <seq>
  Computes the aliquot sequence starting at <seq>.
  <factor> will be tested against the next sequence value.
    Multiple factors can be given with multiple -f.
  If no <index> is given the program continues where the elf file ends.
    In this case the elf file is also checked for correctness at startup.
  -y uses YAFU for all factoring after rho
  -e skips ecm on the current iteration and goes straight to qs/gnfs.
  -p runs at idle priority.
  -q quits after factoring the first number.
  -d quits after sequence reaches <max_digits> digits.
  -c quits after sequence passes ecm with a cofactor of <max_cofactor> or more digits.
  -b with -c and -d quits after a number meets both limits.
  -r with -c or -d (or both) increases max digit/cofactor by <add_digits> if no driver.
  -m ecm will only run to level <max_ecm_level> where '1'=20-digit factor, '2'=25, etc.
  -s submits the iterations <start>+ from <seq>'s elf file to Syd's DB.
    You'll need wget to use -s.
  -t quits after verifying elf file integrity.

(you want -d and -c, and possibly -b)

[gd_barnes]

Great! Thanks. I guess I should have read the aliqueit help file first.

[ChristianB]

I personally use "-c 120 -m 5" and a script to loop over my reserved sequences. I have used an additional "-r 5" before but this was not working as I expected (the limit was not increased to 125 digits but removed altogether) and I saw no way to fix it easily. So be careful with "-r".

[Mr. Odd]

I haven't used aliqueit in some time. When I tried to submit to FactorDB I got a 404 not found error. Do I need to update something?

[RichD]

Originally Posted by Mr. Odd

I haven't used aliqueit in some time. When I tried to submit to FactorDB I got a 404 not found error. Do I need to update something?

Perhaps this may be helpful.
http://www.mersenneforum.org/showpos...&postcount=510

[richs]

Originally Posted by Mr. Odd

I haven't used aliqueit in some time. When I tried to submit to FactorDB I got a 404 not found error. Do I need to update something?

This worked for me: http://www.mersenneforum.org/showpos...&postcount=515

[Stargate38]

I've noticed that Aliqueit doesn't detect certain drivers/guides properly:

2^5*3*7 (Shows up as "Guide: 2^5*3"; should say "Driver: 2^5*3*7")
2^7*3*5*17 (Shows up as "No driver"; should say "Guide: 2^7*(Insert divisor of 255 here)")
2^8*7*73 (Shows up as "No driver", "No 3", or "Large power of 2"; should say "Guide: 2^8*7*73")
2^30*2147483647 (Shows up as "No 3"; should say "Driver: 2^30*M31" or "Driver: 2^30*2147483647".)

Something similar should be done for 2^60*M61, 2^88*M89, 2^106*M107, and 2^126*M127. It should also auto-detect the factors of the next term, using the current driver/guide (i.e. 2^30*M31^2*p with p mod 4=1 will result in a number divisible by M31, but not 2^30). That would be useful when running sequences of the form 2^126*p*M127, so as to not have to put in "-q -f 170141183460469231731687303715884105727" (even with copy/paste) on every term of the sequence. That would save a LOT of time running such sequences.

[ChristianB]

Originally Posted by Stargate38

I've noticed that Aliqueit doesn't detect certain drivers/guides properly:

2^5*3*7 (Shows up as "Guide: 2^5*3"; should say "Driver: 2^5*3*7")
2^7*3*5*17 (Shows up as "No driver"; should say "Guide: 2^7*(Insert divisor of 255 here)")
2^8*7*73 (Shows up as "No driver", "No 3", or "Large power of 2"; should say "Guide: 2^8*7*73")
2^30*2147483647 (Shows up as "No 3"; should say "Driver: 2^30*M31" or "Driver: 2^30*2147483647".)

I've added detection for cases 1 and 3. I'm not sure what I should detect in case 2. Do you mean something like this:
2^7*3 should say "Guide: 2^7*3"
2^7*3*5 should say "Guide: 2^7*3*5"
2^7*5 should say "Guide: 2^7*5" ans so on with all combinations of the three divisors
OR
2^7*3*5*17 should say "Guide: 2^7*3*255"

Case 4 is a bit more tricky as the trial division that tests the driver/guide only checks the primes from 1 to 127. I would need to add a hardcoded list of mersenneprimes to test. But if this is only done for high powers of 2 I think it's worth doing. It would also help with the approach taken below.

Originally Posted by Stargate38

Something similar should be done for 2^60*M61, 2^88*M89, 2^106*M107, and 2^126*M127. It should also auto-detect the factors of the next term, using the current driver/guide (i.e. 2^30*M31^2*p with p mod 4=1 will result in a number divisible by M31, but not 2^30). That would be useful when running sequences of the form 2^126*p*M127, so as to not have to put in "-q -f 170141183460469231731687303715884105727" (even with copy/paste) on every term of the sequence. That would save a LOT of time running such sequences.

That sounds like a lot of logic to build in. What about just using the same hardcoded mersenneprimes as in the driver detection code and brute force it with every step?

The number of MP is also very small since we only need: M13, M17, M19, M31, M61, M89, M107, M127. Would that work?

[ChristianB]

I've gone ahead and implemented the easy stuff. Since there are a lot of changes compared to the original 1.12 from mklasson I decided to release a 1.13 with my changes so far (https://github.com/ChristianBeer/aliqueit/releases). Unfortunately my efforts in building a Windows version were not successful so I only provided a Linux binary. If someone gets a Windows build from my code before me just let me know and I can update the github page.

[chris2be8]

Originally Posted by Stargate38

It should also auto-detect the factors of the next term, using the current driver/guide (i.e. 2^30*M31^2*p with p mod 4=1 will result in a number divisible by M31, but not 2^30). That would be useful when running sequences of the form 2^126*p*M127, so as to not have to put in "-q -f 170141183460469231731687303715884105727" (even with copy/paste) on every term of the sequence. That would save a LOT of time running such sequences.

Or just use brute force by trying all the factors of term x to see if they divide term x+1. That should be simpler to code even if slightly slower.

Chris