missing?

tha · 2003-09-14, 08:57

Ah, a new subforum, can't resist walking in here.

OK, I got this line from the primenet stats page today :

Code:

  33200000   33299999                 184       2   229   598                  19

a couple of months ago it was

Code:

  33200000   33299999                 184           320   511                  12

If I add them up today it is 1032, whereas it was 1027. This happens at other lines as well. Everynow and then the sum adds one or two or loses one or two. Any idea what causes this?

YotN;

Tha

QuintLeo · 2003-09-14, 15:11

I would guess that the missing exponents are manual checkout ones, or possibly in a status that isn't recorded - perhaps they're "overdue for checkin" but haven't been reassigned yet?

GP2 · 2003-09-14, 16:45

[Sept 9 2003 data files used for the figures below]

Well, there's a total of 5797 primes in this range (33.2M-33.3M), but 3604 have known factors, leaving 2193 candidates. And sure enough, the "nofactor" file (derived from NOFACTOR.CMP) has 2193 entries in this range.

Of these 2193 exponents, there are 30 in LUCAS_V.TXT (verified-good results, matching double-checks) and there are 1459 in HRF3.TXT (only one LL test, or non-matching double-checks).

So 704 exponents in this range have no known factor and never had an LL test returned. Primenet shows that 231 assignments are still pending, so that means 473 exponents in this range have never been assigned by Primenet.

It's possible that those 473 exponents were assigned manually, or reserved by someone using Mlucas or Glucas. But most likely they have just never been assigned to anyone. After all these are very big exponents that take a long time, and only a small group of folks with fast machines are working on them.

So my guess is that the increase from 1027 to 1032 just means that 5 new exponents were requested in this range during this time.

Note that many exponents have trivial factors, so most of the factors in this range were eliminated a long time ago. By the time Primenet hands out factoring assignments, only non-trivial factors remain. So you can see only 184 new factors were found since the last database sync.

NickGlover · 2003-09-14, 19:13

The whole 33.2M range hasn't been assigned because all mersenne numbers in that range are not 10M digits. GIMPS started at M33219281.

Regarding the issue of increasing exponents, I think QuintLeo's first suggestion is probably right. Manual exponents (assigned by George via e-mail) do not show up on Primenet unless the person running them connects to Primenet. So in the time between tha's report downloads, I'm guessing five people connected to Primenet with manual 10M digit tests.

GP2 · 2003-09-14, 20:33

Quote:

Originally posted by NickGlover
The whole 33.2M range hasn't been assigned because all mersenne numbers in that range are not 10M digits. GIMPS started at M33219281.

Hmmm, OK let's redo that analysis:

In the range 33219281 - 33300000 there are:

4702 prime exponents
2978 of which have factors
1724 do not have factors

As before, of those 1724, there are 30 in LUCAS_V.TXT (verified-good results, matching double-checks) and there are 1459 in HRF3.TXT (only one LL test, or non-matching double-checks). There are no results in LUCAS_V.TXT or HRF3.TXT for 33.2M less than 33219281.

So 235 exponents in this range have no known factor and never had an LL test returned.

Now, the current status.txt (assigned exponents) shows 231 exponents currently assigned in this range. Again, there are no assignments for 33.2M less than 33219281.

However, of these, 5 assignments are for exponents that have already been LL tested at least once:

33222047
33228563
33234007
33285839
33298957

That leaves 9 exponents that have never been factored or LL tested and are not assigned by Primenet:

33229069
33236519
33238477
33240079
33243181
33249397
33255259
33290479
33297029

These 9 could be manual tests or reserved by Glucas / Mlucas. Or perhaps truly never assigned yet to anyone?

GP2 · 2003-09-14, 20:49

Quote:

That leaves 9 exponents that have never been factored or LL tested and are not assigned by Primenet:

33229069
33236519
33238477
33240079
33243181
33249397
33255259
33290479
33297029

These 9 could be manual tests or reserved by Glucas / Mlucas. Or perhaps truly never assigned yet to anyone?

Oops: the latest cleared.txt (cleared exponents) shows:

Code:

33238477  69     0xD12F51397BCA23__                01-Sep-03 14:16  S67532         C7364A0E9
33243181  69     0x84CEB8C84F756C__                13-Sep-03 09:47  Salz           Pinguin
33255259  69     0x9396794888AB68__                13-Sep-03 04:02  kom            Cel1_7G
33290479  69     0xA15113231F7710__                09-Sep-03 19:17  S62207         C16FB22CD
33297029  69     0x5DF171F39BC1CC__                10-Sep-03 08:58  S58328         C0D1BE822

These were cleared recently (note the files LUCAS_V.TXT and HRF3.TXT files are updated every week or so, so the data for them is from Sept 9... which doesn't explain that first exponent dated 01-Sep-03 though).

So... only 4 exponents that are unfactored, un-LL-tested, and unassigned by Primenet:

33229069
33236519
33240079
33249397

GP2 · 2003-09-14, 21:36

Quote:

Originally posted by GP2
Oops: the latest cleared.txt (cleared exponents) shows:

Code:

33238477 69 0xD12F51397BCA23__ 01-Sep-03 14:16 S67532 C7364A0E9 33243181 69 0x84CEB8C84F756C__ 13-Sep-03 09:47 Salz Pinguin 33255259 69 0x9396794888AB68__ 13-Sep-03 04:02 kom Cel1_7G 33290479 69 0xA15113231F7710__ 09-Sep-03 19:17 S62207 C16FB22CD 33297029 69 0x5DF171F39BC1CC__ 10-Sep-03 08:58 S58328 C0D1BE822

These were cleared recently (note the files LUCAS_V.TXT and HRF3.TXT files are updated every week or so, so the data for them is from Sept 9... which doesn't explain that first exponent dated 01-Sep-03 though).

I investigated further with regards to the exponent dated 01-Sep-03.

It turns out that this exponent 33238477 is one of only 15 that are in cleared.txt (with a date before 09-Sep-03) but were never entered into the data files. An amazing coincidence that this current thread just happened to stumble on such an exponent.

See the new Cleared exponents that never made it into data files thread.

2003-09-14, 08:57	#1
tha Dec 2002 865₁₀ Posts	missing? Ah, a new subforum, can't resist walking in here. OK, I got this line from the primenet stats page today : Code: 33200000 33299999 184 2 229 598 19 a couple of months ago it was Code: 33200000 33299999 184 320 511 12 If I add them up today it is 1032, whereas it was 1027. This happens at other lines as well. Everynow and then the sum adds one or two or loses one or two. Any idea what causes this? YotN; Tha

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
are my GHz days missing	wildrabbitt	Lounge	7	2017-02-01 11:39
Missing bit levels?	NBtarheel_33	Data	6	2016-05-31 15:27
Missing top 10?	R.D. Silverman	GMP-ECM	3	2011-03-18 21:35
The missing dollar	Andi47	Puzzles	4	2008-09-11 14:56
A missing identity	fivemack	Factoring	4	2008-03-04 05:04

2003-09-14, 15:11	#2
QuintLeo Oct 2002 Lost in the hills of Iowa 1C0₁₆ Posts	I would guess that the missing exponents are manual checkout ones, or possibly in a status that isn't recorded - perhaps they're "overdue for checkin" but haven't been reassigned yet?

2003-09-14, 16:45	#3
GP2 Sep 2003 5035₈ Posts	[Sept 9 2003 data files used for the figures below] Well, there's a total of 5797 primes in this range (33.2M-33.3M), but 3604 have known factors, leaving 2193 candidates. And sure enough, the "nofactor" file (derived from NOFACTOR.CMP) has 2193 entries in this range. Of these 2193 exponents, there are 30 in LUCAS_V.TXT (verified-good results, matching double-checks) and there are 1459 in HRF3.TXT (only one LL test, or non-matching double-checks). So 704 exponents in this range have no known factor and never had an LL test returned. Primenet shows that 231 assignments are still pending, so that means 473 exponents in this range have never been assigned by Primenet. It's possible that those 473 exponents were assigned manually, or reserved by someone using Mlucas or Glucas. But most likely they have just never been assigned to anyone. After all these are very big exponents that take a long time, and only a small group of folks with fast machines are working on them. So my guess is that the increase from 1027 to 1032 just means that 5 new exponents were requested in this range during this time. Note that many exponents have trivial factors, so most of the factors in this range were eliminated a long time ago. By the time Primenet hands out factoring assignments, only non-trivial factors remain. So you can see only 184 new factors were found since the last database sync.

2003-09-14, 19:13	#4
NickGlover Aug 2002 Richland, WA 2²×3×11 Posts	The whole 33.2M range hasn't been assigned because all mersenne numbers in that range are not 10M digits. GIMPS started at M33219281. Regarding the issue of increasing exponents, I think QuintLeo's first suggestion is probably right. Manual exponents (assigned by George via e-mail) do not show up on Primenet unless the person running them connects to Primenet. So in the time between tha's report downloads, I'm guessing five people connected to Primenet with manual 10M digit tests.