Very Prime Riesel and Sierpinski k

[robert44444uk]

Originally Posted by Thomas11

This sounds quite interesting to me. So you may count on me for such an effort...

Thomas 11, it will take a little time to develop 1000 new VPS but if you join in on the Sierpinski side, then this should double the effort!! I suggest you download Robert G's software, available from 3 October post, and also look at Oct 24 post on progress to date on Sierpinski side, and go from there.

I am running 2.5 cores approx on this exercise.

[robert44444uk]

Originally Posted by kar_bon

oh, yes, you're right. i was only referring to the Riesel-side of my collection of primes and Payem-numbers!

it's time to make a collection like mine for the '+'-side! somebody here to do so?

You will find Sierpinski VPS found before Robert's program in my post of 29 Sep

There is a superior Riesel - see post of 29 Sept (Reisel)

Regards

Robert

[Thomas11]

Originally Posted by robert44444uk

... I suggest you download Robert G's software, available from 3 October post, and also look at Oct 24 post on progress to date on Sierpinski side, and go from there.

Okay, I've got the code and it compiled well on my 64 bit Linux machine.

For the beginning I will take (according to your 24 Oct 09 post):
S 58 from Iteration=0, i=98849

Will take more, once I got some feeling about the software (cpu and memory utilization, running times, ...) and the theory behind it.

[Thomas11]

After just two hours I already found some nice candidates:

S 201456540759 58 100/6929 105/10000 K=27547987231931215530195
S 581338538697 58 100/9084 105/10000 K=79494597599651554004685
S 2246005540369 58 100/8809 103/10000 K=307127937945434173384245
S 125445708657 58 100/9608 100/10000 K=17153956716241359070485

Now I caught fire...

Also taking:
S 60 from Iteration=0, i=24786

[robert44444uk]

Originally Posted by Thomas11

Okay, I've got the code and it compiled well on my 64 bit Linux machine.

For the beginning I will take (according to your 24 Oct 09 post):
S 58 from Iteration=0, i=98849

Will take more, once I got some feeling about the software (cpu and memory utilization, running times, ...) and the theory behind it.

I did not find it necessary to adjust any of the settings for S 58, I think this and 60 are about optimal for quick finds. 52 is just too slow, because a lot of close candidates.

When you post, don't forget to check your candidates against the original Sierpinski list which contains the cumulative list of 194 finds before Robert's software. Of the 4 you posted the first two are known 20145... is 152/137581 and 58133...is 118/25170. The other two are new. And I will add these to the master list of new finds and credit you.

The duplication issue should cease once you are past the first 3 iterations for 58 and 60 given that the largest finds were 838422520523 for 58 and 933294678535 for 60.

4 in a couple of hours is awesome. My poor laptops and no electricity here in Bangladesh make for snail-like progress.

Don't forget to save the record tables either. You have to check these every time you stop the program, otherwise you lose that record for ever. You never know, you might find an 16/16 in there. I am keeping records for each E level.

[Thomas11]

Okay, after one day I've got 23 VPS candidates for S 58, out of which (only) 4 were duplicates. So here are just the new ones (including the two reported yesterday):

S 52930116711 58 100/7489 104/10000 K=7237879563729757669155
S 125445708657 58 100/9608 100/10000 K=17153956716241359070485
S 251671885031 58 100/9867 100/10000 K=34414637764300621582755
S 1045666111343 58 100/8726 100/10000 K=142988639513076872029515
S 1461196621251 58 100/6837 105/10000 K=199809972482937566175855
S 1526585622539 58 100/8345 101/10000 K=208751530626465272827095
S 1712013118665 58 100/9006 103/10000 K=234107641063399884677325
S 1729977121403 58 100/9969 100/10000 K=236564111904189367255815
S 1780903684577 58 100/8389 102/10000 K=243528017403598252982085
S 1880663844267 58 100/9973 100/10000 K=257169627624053528204535
S 2166054501341 58 100/8704 101/10000 K=296195107499597269840305
S 2246005540369 58 100/8809 103/10000 K=307127937945434173384245
S 2354360292597 58 100/8357 101/10000 K=321944807726132334214185
S 2562085263199 58 100/9853 100/10000 K=350349965564828800966395
S 3241817377221 58 100/9958 100/10000 K=443299301077406965467705
S 3787165959141 58 100/7042 103/10000 K=517872485522463490009305
S 4392705084017 58 100/9672 100/10000 K=600676369763058021003285
S 4679042300103 58 100/9973 100/10000 K=639831285970023930669315
S 5622377417391 58 100/8369 101/10000 K=768826768909295041500555

Meanwhile I'm running all E<=106 (58, 60, 66, 82, 100, and 106) and will probably extend this to even higher ones during the weekend.

Just one question about the in.txt file:
I noticed that you (Robert) posted also some "near misses". So should I change the "vpscount" to somewhat lower than 100 (perhaps 95)?

And another question:
Do I need to check whether, let's say, a "S 58" may be already listed as a "S 60" (or higher) candidate?

Will post more results (e.g. for S 60) soon...

[robert44444uk]

Originally Posted by Thomas11

Okay, after one day I've got 23 VPS candidates for S 58

I envy your processing power! Hail to thee!!! I will credit your results to the master list. We will get to 1000 new VPS quite quickly I think.

Originally Posted by Thomas11

Just one question about the in.txt file:
I noticed that you (Robert) posted also some "near misses". So should I change the "vpscount" to somewhat lower than 100 (perhaps 95)?

No, we will have problems with running even 100/10000 up to higher numbers. We should let these 99/10000 lie. The discussion around 99/10000 was before Robert G's program.

Originally Posted by Thomas11

And another question:
Do I need to check whether, let's say, a "S 58" may be already listed as a "S 60" (or higher) candidate?

Yes, you should factorise y and look to see if the factorisation contains the next prime or even 2 primes in M(x). If so, divide y by the prime factor(s) and check to make sure it is not duplicated. All y values posted should be efficient in this respect. Robert's program does not carry out this step.

It is not so important for the racing tables if a y is not perfectly stated. There are only so many racing records, and values can be adjusted by hand if necessary.

[robert44444uk]

By the way, when we take these numbers to higher n, we need to do at least prp-3 checking on the first 10000 n. Robert's program (I understand) gambles with prp-1. It is worth noting that I have not found any count discrepancies between prp-3 and prime, neither have I discovered any discrepancies between prp-1 and prp-3.

Also for lower n, we must recognise that k> 2^n and adjust tests accordingly for these low n.

[robert44444uk]

Originally Posted by Thomas11

S 52930116711 58 100/7489 104/10000 K=7237879563729757669155

Can you also post date, iteration number, and i value and date found. Would like to get a full record for statistical purposes.

[Thomas11]

Originally Posted by robert44444uk

Can you also post date, iteration number, and i value and date found. Would like to get a full record for statistical purposes.

Meanwhile the number of VPS sequences from my end has increased to 116, out of which 80 are new. They are attached to this post, sorted by E and y.
By factorizing the y values I've found that two of them are indeed candidates of the next higher E series (one 60 -> 66, and the other one 66 -> 82). They are indicated in the file (at the rightmost column).

I was quite excited when I hit a "116/10000", but unfortunately this one was already discovered earlier (S 2158430601663 66). So the highest unknown one is a still quite nice 110/10000 (S 5475497492533 58).

BTW.: I haven't found any VPS for E=130 and E=138 so far. I know that they become quite rare with increasing E. But I'm a bit concerned about the "smith_check" levels. Should I relax those constrains a bit? E.g. there is a Riesel 97/10000 candidate I found earlier, which would have been thrown out already at the "10 50" level but otherwise would have survived the "smith_check" at all other (higher) levels.

[Thomas11]

I've just hit my first VPS for E=130:
S 179782224211057 130 100/7663 104/10000
K=11806316649721727826033357267756435645
iteration=62 I=82756 Fri Nov 13 12:30:36 2009