mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   FactorDB (https://www.mersenneforum.org/forumdisplay.php?f=94)
-   -   The "Hey YOU" Thread (https://www.mersenneforum.org/showthread.php?t=18489)

EdH 2016-10-23 20:56

[QUOTE=schickel;445622]The way to check that is go ahead and query. If it's new, you get the * indicator that it was freshly added.
...[/QUOTE]But, if I were to query, it would make the query itself a moot point, since any number queried would be in the db. I was merely thinking of looking at the ones that were, but didn't want to add any to the db for only that reason.

schickel 2016-10-23 21:55

[QUOTE=EdH;445631]But, if I were to query, it would make the query itself a moot point, since any number queried would be in the db. I was merely thinking of looking at the ones that were, but didn't want to add any to the db for only that reason.[/QUOTE]Oh, sorry. I thought you wanted to know if they had been inserted previous to your querying them. I would probably assume if the numbers have been around any significant length of time, they'll most likely be in there because someone will have looked them up.

There would actually be a way to not insert them if you could get close to your limit of page hits; after you hit the limit, it will show numbers that you query as not in the DB but also not insert them. Kind of useless, since you usually want to be inserting them anyway.

EdH 2016-10-23 22:27

[QUOTE=schickel;445633]Oh, sorry. I thought you wanted to know if they had been inserted previous to your querying them. I would probably assume if the numbers have been around any significant length of time, they'll most likely be in there because someone will have looked them up.

There would actually be a way to not insert them if you could get close to your limit of page hits; after you hit the limit, it will show numbers that you query as not in the DB but also not insert them. Kind of useless, since you usually want to be inserting them anyway.[/QUOTE]
Thanks! although it would be minuscule in the overall scheme, I didn't want to add to the already too-large number of 309 digit composites. I think the only times I hit my limit were when running yafu.pl and the limit cleared rather quickly.

Although xilman's numbers have significance, I don't like adding numbers that have no other significance than my curiosity.

henryzz 2016-10-23 22:32

[QUOTE=EdH;445630]From the distribution graph, I had estimated ~2000, but someone else might want to check that.[/QUOTE]

I should be able to remove that. I spending around a second ecming each of the 55000 numbers accessible. The factor rate should be over 5% based upon experiments. I am hoping for 3000 factors. Of course the few digits before will need redoing as well as my factors will leave smaller composites. Assuming one factor per number it will be necessary to go back around 30 digits. Currently my best factor is 25 digits. That will probably increase.
I might try and run this sort of ecm from the smallest numbers upward and see how far I can get. I would imagine I would get a higher factor rate normally than 309 digits due to rsa composites.

henryzz 2016-10-24 10:43

[QUOTE=henryzz;445635]I should be able to remove that. I spending around a second ecming each of the 55000 numbers accessible. The factor rate should be over 5% based upon experiments. I am hoping for 3000 factors. Of course the few digits before will need redoing as well as my factors will leave smaller composites. Assuming one factor per number it will be necessary to go back around 30 digits. Currently my best factor is 25 digits. That will probably increase.
I might try and run this sort of ecm from the smallest numbers upward and see how far I can get. I would imagine I would get a higher factor rate normally than 309 digits due to rsa composites.[/QUOTE]

Half done and 3700 factors. Largest 31 digits

EdH 2016-10-24 14:33

[QUOTE=henryzz;445659]Half done and 3700 factors. Largest 31 digits[/QUOTE]
And, yet, the distribution graph remains unchanged...

henryzz 2016-10-24 14:41

[QUOTE=EdH;445668]And, yet, the distribution graph remains unchanged...[/QUOTE]

I haven't submitted yet.

EdH 2016-10-24 22:52

I am wondering... how is/will henryzz's work affect chris2be8's work?

Also, I see a lot of the following:
[code]
[URL="http://www.factordb.com/index.php?id=1100000000874739246"][COLOR=#002099](137^104*17^51*226^14*188+1)/1434648233765[/COLOR][/URL]
[/code]Are these numbers formatted this way by the db for display, or are they entered into the db from a particular project?

Is there a reason for the spike at ~309 digits and the even larger amount at around 480-500 digits?

schickel 2016-10-25 01:04

[QUOTE=EdH;445713]I am wondering... how is/will henryzz's work affect chris2be8's work?

Also, I see a lot of the following:
[code]
[URL="http://www.factordb.com/index.php?id=1100000000874739246"][COLOR=#002099](137^104*17^51*226^14*188+1)/1434648233765[/COLOR][/URL]
[/code]Are these numbers formatted this way by the db for display, or are they entered into the db from a particular project?

Is there a reason for the spike at ~309 digits and the even larger amount at around 480-500 digits?[/QUOTE]I think someone is dumping numbers in. Don't know if they're from a "useful" project or not. I was looking at the PRPs over the weekend and thinking about clearing out the backlog <3000 digits again and saw a whole crap load of 19--3x digit primes appear (>25 in pretty much all the bins for size). They promptly vanished, but whenever I see that it concerns me that they're dumping number in faster than the prime-check workers can clear them. I would have to assume that the sudden spike in composites springs from the same effort.

The spikes in the "counts by type" graph do kind of look like someone is searching numbers near "rounds": decades, centuries, etc. Prime search of some kind?

chris2be8 2016-10-25 15:47

[QUOTE=EdH;445713]I am wondering... how is/will henryzz's work affect chris2be8's work?

[/QUOTE]

It won't affect me at all. I've checked for algebraic factors up to 324 digits, finding smallish factors by ECM won't create any more algebraic factors.

If it makes some more 309 digit numbers accessible I could do a special run to check them for algebraic factors. All I need to know is what range to search in factordb and when to start.

Chris

henryzz 2016-10-25 20:38

My work is submitted. Doesn't seem to have brought it below 550k numbers at 309 digits.


All times are UTC. The time now is 05:32.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.