Large number of filering passes

henryzz · 2012-07-31, 15:15

I was just surprised to notice that singleton removal just ran 95 passes on a c118. What is the highest anyone has had?

LaurV · 2012-07-31, 15:38

From the logs which I did not delete yet (I clear them regularly) it seems that I have every number between 9 and 21 many times, and i even saw a 33 and a 41. It does not seems to be any bigger.

(edit: I saw a 7 too, and I have no idea for which number is the 41, I have to put in correspondence the timing in the nfs.log where the 41 appears with the timing in the factor.log where the number to be factored appears, and I am too lazy to do that now, 22:45 here)

jasonp · 2012-07-31, 17:35

The closer you get to the crossover point between filtering failing and filtering succeeding but producing a terrible matrix, the larger the number of singleton-removal passes you need to do. It has something to do with the connectivity of the graph of large factors in the dataset; you have a huge number of relations that are connected to each other in groups, with a tiny bit of connectivity between the groups. Get rid of those few links and the whole dataset collapses into singletons that must be pruned, but add a few more relations and suddenly most of the data is not singletons anymore.

2012-07-31, 15:15	#1
henryzz Just call me Henry "David" Sep 2007 Cambridge (GMT/BST) 2³×3×5×7² Posts	Large number of filering passes I was just surprised to notice that singleton removal just ran 95 passes on a c118. What is the highest anyone has had?

2012-07-31, 15:38	#2
LaurV Romulan Interpreter Jun 2011 Thailand 22613₈ Posts	From the logs which I did not delete yet (I clear them regularly) it seems that I have every number between 9 and 21 many times, and i even saw a 33 and a 41. It does not seems to be any bigger. (edit: I saw a 7 too, and I have no idea for which number is the 41, I have to put in correspondence the timing in the nfs.log where the 41 appears with the timing in the factor.log where the number to be factored appears, and I am too lazy to do that now, 22:45 here) Last fiddled with by LaurV on 2012-07-31 at 15:43

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
checking very large number for primality.	WhoCares	Math	51	2017-04-20 17:17
Bug with particular large number	VolMike	YAFU	18	2012-04-09 21:39
Sieving a Large Number of k's	Dougal	Conjectures 'R Us	1	2010-06-17 21:48
Large number of Exponents Assigned	dave_0273	Data	2	2004-03-11 16:15
Is this a relatively large number?	MavsFan	Math	3	2003-12-12 02:23

2012-07-31, 17:35	#3
jasonp Tribal Bullet Oct 2004 3,541 Posts	The closer you get to the crossover point between filtering failing and filtering succeeding but producing a terrible matrix, the larger the number of singleton-removal passes you need to do. It has something to do with the connectivity of the graph of large factors in the dataset; you have a huge number of relations that are connected to each other in groups, with a tiny bit of connectivity between the groups. Get rid of those few links and the whole dataset collapses into singletons that must be pruned, but add a few more relations and suddenly most of the data is not singletons anymore.