Factoring database

[schickel]

Originally Posted by wblipp

Why is there interest in finding the complete factorization of 136^35+n for n odd 1 to one million, and 2^1024+n, n 1 to 1000? It appears both of these sets have been dumped into factordb this week.

I figured our friend has been (hyper)active again....

[lorgix]

Fatal error:  Call to a member function __getgmp() on a non-object in /apache/htdocs/factor.php on line 538

[lorgix]

Originally Posted by lorgix

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

Btw, are all scan levels necessary? The first is ridiculously fast. And ECM to low bounds is exhausted with only a few curves, I think.

Anyone have an opinion on this?

[cmd]

Originally Posted by lorgix

Looking at HP₁₀(49) I clicked 'Show huge graph'.

That brought me to http://factordb.com/aliquot.php?type=10&aq=7&big=1

And this image was displayed;

25064

[bchaffin]

Originally Posted by lorgix

Anyone have an opinion on this?

I think the DB already does some work in the background on all composites. I have a script which slowly walks through open aliquot sequences to keep my local list up to date, and it often finds that, for example, a c110 has "spontaneously decayed" into a p15 and a c95, without adding any terms to the sequence. Surely this is not a human going to the trouble of submitting a partial factorization (is it?). So it seems to me that the DB must be doing a little testing which occasionally finds factors up to about 20 digits.

[Syd]

I just put a small update online:

-now it does P-1 to B1=25k on every incoming number <400 digits
-the first click on the scan-button runs 2 levels at once
-some small bugfixes (HP 49 bug, certificate bug, some more)
-snfs poly generator (just as an experiment)
-next prime above N
and some more.

Hope everything is working (and I left no debugging code in )

Originally Posted by bchaffin

I think the DB already does some work in the background on all composites. I have a script which slowly walks through open aliquot sequences to keep my local list up to date, and it often finds that, for example, a c110 has "spontaneously decayed" into a p15 and a c95, without adding any terms to the sequence. Surely this is not a human going to the trouble of submitting a partial factorization (is it?). So it seems to me that the DB must be doing a little testing which occasionally finds factors up to about 20 digits.

I ran a few curves with very low limits on all the composites below 150 digits, just to get the very small factors, maybe that found the p15.

[lorgix]

Nice changes!

Weird btw: search for 99,738 on the page below.

http://factordb.com/show_config.php

[schickel]

Syd, would it be possible to add another stats query like the "stat_1.php" for composites, but list the PRPs instead?

I noticed the DB being laggy while uploading some sequence results and a little snooping shows that someone has apparently dumped in 1,000s of mid-20 digit PRPs that are taking a while to clear......I found 3 pages of 1000 at the 21-digit level.

[Edited to add: Make that 10s of 1,000s. There are >10,000 PRPs between 22 and 26 digits right now.]

[wblipp]

Using the "More Information" button, you can see that the PRPs were not entered directly. They come from factoring a large number of composites recently entered. They do not seem to have any particular structure.

[schickel]

Originally Posted by wblipp

Using the "More Information" button, you can see that the PRPs were not entered directly. They come from factoring a large number of composites recently entered. They do not seem to have any particular structure.

It looks like they cleared out faster than I thought they were.

I had uploaded <10 lines of a sequence and the ~40 digit factors were still marked as PRP after 10 minutes or so. It looks like the queue has cleared now.....I just panicked thinking we had someone flooding the DB with useless numbers again.

[Syd]

Originally Posted by lorgix

Nice changes!

Weird btw: search for 99,738 on the page below.

http://factordb.com/show_config.php

Thank you! These limits were determined by experiment, so they are not 100% accurate. But they dont need to be.

Originally Posted by schickel

Syd, would it be possible to add another stats query like the "stat_1.php" for composites, but list the PRPs instead?

Sure:

http://factordb.com/stat_1.php?prp