Aliqueit.exe discussion

[ChristianB]

I've done nothing to enable compilation with gmp-6.2. I was in the process of testing it myself but then yafu segfaulted and I tried to recompile which doesn't work with the version I have. I hadn't had time to update yafu. But my gmp-ecm 6.4.4 works well with GMP 6.2.0.

I now have several GMP and GMP-ECM versions in different directories that I first have to clean up before doing proper testing again. I also need to get my yafu working too.

In the meantime I figured out how to adapt to your ecmpi style logs. Have a try with this code: https://github.com/ChristianBeer/ali...cm_factors.zip it only has the log changes not the b2scale changes (yet).

[EdH]

Originally Posted by ChristianB

I've done nothing to enable compilation with gmp-6.2. I was in the process of testing it myself but then yafu segfaulted and I tried to recompile which doesn't work with the version I have. I hadn't had time to update yafu. But my gmp-ecm 6.4.4 works well with GMP 6.2.0.

I now have several GMP and GMP-ECM versions in different directories that I first have to clean up before doing proper testing again. I also need to get my yafu working too.

In the meantime I figured out how to adapt to your ecmpi style logs. Have a try with this code: https://github.com/ChristianBeer/ali...cm_factors.zip it only has the log changes not the b2scale changes (yet).

I d/led and compiled earlier today on a GMP-6.2.0 machine with no issue. Maybe something you did, "fixed" a minor trouble spot? I'll try the latest sometime soon and see how it runs.

Thanks for all the work!

[EdH]

Originally Posted by ChristianB

. . .
I also need to get my yafu working too.
. . .

Did you catch the change to msieve that is needed for GMP-6.2.0? It's documented here.

[EdH]

@Chris: I've tried having the aliqueit_ecm_temp.log replaced and appended with the ecmpi output and neither worked. Was I supposed to try something different?

Here is the relevant part of the output:

. . .
# 004: B1 = 1000000
# 004: #curves = 736
# 128: curve 0 found factor 1928088817363669499071639609 using sigma -1:2018415605
# 136 curves done (18.5%)
Results: 665195623909608793118604237027505628448713991780912365716587033760914298407196872483651634174320560969033 = 345002583863926781058856507114499367584791600904112246231579507936566437211537 * 1928088817363669499071639609

[ChristianB]

Originally Posted by EdH

@Chris: I've tried having the aliqueit_ecm_temp.log replaced and appended with the ecmpi output and neither worked. Was I supposed to try something different?

No you don't need to change anything. The new code works with the sample provided by you in this post https://www.mersenneforum.org/showpo...&postcount=582

What I essential do in each line that contains "Factor found" is to look for " * " and also extract the number after that.

[EdH]

@Chris:
I must confess to not doing further study in making the alternate version read my ecmpi output, but I would at least like to mention that I do like the new feedback from factordb when a sequence is submitted:

DB reports 98 new factors.

It's nice to see this note.

Thanks for all the work.

[Mr. Odd]

Ed - thanks for all your work supporting these applications, I finally got aliqueit compiled and running. One thing I can't figure out - is there a way to have it automatically submit new terms to the FactorDB as it goes along, or is the only way after you've stopped it?

[EdH]

Originally Posted by Mr. Odd

Ed - thanks for all your work supporting these applications, I finally got aliqueit compiled and running. One thing I can't figure out - is there a way to have it automatically submit new terms to the FactorDB as it goes along, or is the only way after you've stopped it?

The short answer is, "No - Aliqueit won't update factordb until stopped and told to."

But, there are longer answers which can update along the way:

1. While running Aliqueit to build a sequence, you can invoke another instance to update the db. I do this often and have written a bash script to help me in my larger jobs. A script could be set up to automatically do updates on a schedule.

2. Aliqueit could alternately be set up in a script to complete one line at a time and submit it to the db.

3. Although not, using Aliqueit, there are scripts that work directly with the db to retrieve the last composite of a sequence, factor it (usually with YAFU) and return the factors to the db. Unfortunately, I can't find the source page, but there is one called ali.pl that, I believe, was written by yoyo. If you would like more info on it, I can try to find it, or you might check with yoyo.

[firejuggler]

I seem to remember a gui for aliquiet, specifically Aliwin, that allowed you to update the db on the go. circa 2010-2012... let me find it again. https://www.mersenneforum.org/showthread.php?t=13365

[bur]

I am calculating aliquot sequences for the powers of integers project and wanted to limit the time spend on a sequence by using -m 5. Unfortunately this doesn't work in combination with -y, it seems that this parameter is not passed to yafu since yafu still tries ECM up to 1/3 the number of digits of the composite which can be quite large.

So I edited yafu.ini to include pretest=40 but it is ignored, maybe because of the way aliqueit invokes yafu.

Is there any way to achieve the result of -m 5 but in combination with -y?

[LaurV]

the pretest_ratio in yafu ini is in percent of the size (i.e. for a C150, using 33.3334 will factor to t50), and you need plan=custom to use the custom percent