Factoring database

[Andi47]

Originally Posted by Syd

No, it was not intended!!

nice to hear that!

[WraithX]

Originally Posted by Syd

-ecm/p+1/p-1-results are now also saved. Just copy the whole gmp-ecm output inside the results box

-added a little tool to calculate ecm group orders
(http://factordb.com/groupcalc.php)

The calculator for the group order works great for a couple of test numbers I entered in. And it is pretty quick too. What do you use to calculate the group orders? Is it Pari/GP, something written in GMP, or something else?

Also, when submitting gmp-ecm output, what all should be included? Should it be:
---------------------------------------------------
GMP-ECM 6.3 [configured with GMP 5.0.1] [ECM]
Input number is 189620700613125325959116839007395234454467716598457179234021
Using B1=3000000, B2=5706890290, polynomial Dickson(6), sigma=4282141129
********** Factor found in step 2: 282174488599599500573849980909
Found probable prime factor of 30 digits: 282174488599599500573849980909
Probable prime cofactor 671998030559713968361666935769 has 30 digits
---------------------------------------------------
Or more? Or less?

Also, do you save what version of gmp-ecm was used? And, if someone submits an ecm/p-1/p+1 for an existing number (that either does or does not already have an existing ecm/p-1/p+1 entry), does the new entry overwrite/update the existing factor? Or do you just save this info for new numbers?

[Syd]

Originally Posted by WraithX

The calculator for the group order works great for a couple of test numbers I entered in. And it is pretty quick too. What do you use to calculate the group orders? Is it Pari/GP, something written in GMP, or something else?

Also, when submitting gmp-ecm output, what all should be included? Should it be:
---------------------------------------------------
GMP-ECM 6.3 [configured with GMP 5.0.1] [ECM]
Input number is 189620700613125325959116839007395234454467716598457179234021
Using B1=3000000, B2=5706890290, polynomial Dickson(6), sigma=4282141129
********** Factor found in step 2: 282174488599599500573849980909
Found probable prime factor of 30 digits: 282174488599599500573849980909
Probable prime cofactor 671998030559713968361666935769 has 30 digits
---------------------------------------------------
Or more? Or less?

Also, do you save what version of gmp-ecm was used? And, if someone submits an ecm/p-1/p+1 for an existing number (that either does or does not already have an existing ecm/p-1/p+1 entry), does the new entry overwrite/update the existing factor? Or do you just save this info for new numbers?

I use a part of the sage-library to calculate it, its the fastest i could find.

I only save the sigma, b1/b2 bounds, submission time and the user (if logged in). There can only be one ecm/p+1/p-1-result per prime, and thats the one submitted first. I store this not only for new primes, also on already existing numbers.

You can either paste the gmp-ecm output inside the report box (needs input number, prime found + parameters in it) or open the ecm-window on this number and enter the parameters manually. I tested this for quite a few numbers/outputs, so hopefully its bugfree

[lorgix]

Doesn't work for factors found with Brent-Suyama's extension, right?

[Syd]

Originally Posted by lorgix

Doesn't work for factors found with Brent-Suyama's extension, right?

No, not yet.

[lorgix]

OK. Sounds like you're working on that(?)

Would it be realistic, btw, to do some basic ecm/p+1/p-1 automatically? Like TF.


p.s. Great work with the db, I really appreciate it!

[mnh001]

For numbers less than 50 digits it's easy enough to grab thousands at once to batch process but for numbers greater than 50 digits, is there any way to get large quantities of them at once? Or just one at a time? Might be a nice feature to put in sometime.

[ugly2dog]

Any chance that factors that were submitted for sequences be redone on your end? I had a hard drive crash and lost several days of data.

[Syd]

Originally Posted by lorgix

OK. Sounds like you're working on that(?)

Thats something I have to read more about first.

Would it be realistic, btw, to do some basic ecm/p+1/p-1 automatically? Like TF.

On small numbers - of cause. At least P-1, B1=20k for all numbers <400 digits sounds okay to me. But this needs some more changes - give me a week or so.

p.s. Great work with the db, I really appreciate it!

Thank you

Originally Posted by mnh001

For numbers less than 50 digits it's easy enough to grab thousands at once to batch process but for numbers greater than 50 digits, is there any way to get large quantities of them at once? Or just one at a time? Might be a nice feature to put in sometime.

You could either use listtype.php - or - the script I just placed within the download-section. Max. is 5k at once. Enough? But at the moment you will just get some of the random numbers someone spammed in there ..

Originally Posted by ugly2dog

Any chance that factors that were submitted for sequences be redone on your end? I had a hard drive crash and lost several days of data.

Redone? If you submitted them, you can get em back. Or did I not understand your question?

[Syd]

Originally Posted by cmd

_.

100k useful numbers?!?

[EdH]

Originally Posted by Syd

...But at the moment you will just get some of the random numbers someone spammed in there ...

I hope those aren't residual from my sociable number search around 10^15 a week or so ago. I was using AliWin/Aliqueit and was bumping the db to see if the sequences might already exist, which caused it to initiate and run the sequences based on the primary number. Then the sequences took off on their own generating more composites. I stopped querying the db as soon as I realized it was running away with the sequences. I would have thought that they would have cleared by now, but maybe not.

Sorry, if I was that source of the "random" numbers. It was neither foreseen nor intentional.

I wonder if it might not actually be good to have the db initiate an aliquot sequence simply by querying for a non-existent one...