Leyland Primes (x^y+y^x primes)

[pxp]

Originally Posted by rogue

Reach out to pxp.

For individuals wishing to find Leyland PRPs that are not already known, let me state first of all that I can be of no help in the programming aspect of it. Secondly, it is my personal belief that all PRPs with fewer than 150000 decimal digits are already known. They are charted here. If you wish to find a new Leyland PRP, you will need to figure out how to tailor your search to exclude those "small" values. Finally, there are currently 85 Leyland PRPs in that chart with more than 149999 decimal digits, so you will try to avoid them. For example, all Leyland PRPs with at least 300000 decimal digits but fewer than 305000 decimal digits are believed to be known.

[rogue]

Originally Posted by pxp

For individuals wishing to find Leyland PRPs that are not already known, let me state first of all that I can be of no help in the programming aspect of it. Secondly, it is my personal belief that all PRPs with fewer than 150000 decimal digits are already known. They are charted here. If you wish to find a new Leyland PRP, you will need to figure out how to tailor your search to exclude those "small" values. Finally, there are currently 85 Leyland PRPs in that chart with more than 149999 decimal digits, so you will try to avoid them. For example, all Leyland PRPs with at least 300000 decimal digits but fewer than 305000 decimal digits are believed to be known.

Who is coordinating the ranges of x and y that have been searched? The original xyyx search did that and xyyxsieve was written based upon that. Searching purely by decimal length makes it harder for multiple people to work on the search concurrently.

Is anyone coordinating x^y - y^x? I don't think that you are searching terms of that form.

[kruoli]

Originally Posted by storm5510

A reference made to mtsieve makes no sense.

Are you running all canidates through OpenPFGW without sieving them first? That will take multiple times longer! I am sure we (as in "the forum") can help you in getting started with the sieving process first. But I concur with others that you should not feel bound by some random request. Imagine someone saying: Please do PRP/LL work for me until you find a new Mersenne prime. Bad idea.

[xilman]

Originally Posted by kruoli

Are you running all canidates through OpenPFGW without sieving it first? That will take multiple times longer!

Even 20 years ago, I wasn't that stupid!

Back then I used home-brew GMP code to test for PRPness after a preliminary sieve and a UBASIC implementation of APR-CL for proof up to about 1000 digits.

[storm5510]

Originally Posted by kruoli

Are you running all canidates through OpenPFGW without sieving it first? That will take multiple times longer! I am sure we (as in "the forum") can help you in getting started with the sieving process first.

But I concur with others that you should not feel bound by some random request. Imagine someone saying: Please do PRP/LL work for me until you find a new Mersenne prime. Bad idea.

Without sieving, yes. I've done pre-seiving before, but it was for Riesel. Just by watching OpenPFGW output on the screen, I can tell it needs to be done badly. What with, I haven't a clue. Something in the mtsieve collection possibly. There is no mtsieve binary in the group, yet folks keep referring to it as if there is.

[kruoli]

Keep in mind that what was asked is a totally absurd amount of work either way. And that it is off-topic.

[mathwiz]

Originally Posted by storm5510

mtsieve is an archive with 27 binaries inside. There must be an error here somewhere.

She's asking me to do it because I have a grasp how to do it. I have no problem with the ask.

All you are doing is feeding the troll and encouraging sweety to keep posting "please run <WORK>" for me. And in this case, <WORK> isn't even related to x^y+y^x, the actual topic of the thread.

[VBCurtis]

Originally Posted by storm5510

She's asking me to do it because I have a grasp how to do it. I have no problem with the ask.

Thing is, you *don't* have a grasp how to do it. Without sieving, the task is hopeless.

Sweety-
Time for your usual warning- posting about your pet projects in threads about other projects is cause for a ban. Stop it, for good.

[pxp]

Originally Posted by rogue

Who is coordinating the ranges of x and y that have been searched? The original xyyx search did that and xyyxsieve was written based upon that. Searching purely by decimal length makes it harder for multiple people to work on the search concurrently.

There hasn't been much of any coordination since Andrey Kulsha abandoned (in early 2017) his moderation of the project. To be fair, I don't think that by then there was a lot of activity. Kulsha had 1250 entries in his prime database. Since then, 1525 entries have been added... by six people, two people if we ignore more recent events. Those two people were Norbert and myself. Norbert was using (I believe) the old-fashioned system. I was working (strictly in Mathematica) by decimal length. When I thought that there was an overlap, I alerted Norbert to it. This is why he started posting results here as soon as he found something.

[kruoli]

If you could dump the work that has been done I can host a website for that.

[pxp]

Originally Posted by kruoli

If you could dump the work that has been done I can host a website for that.

Was this directed at me? If so, what do you mean by "the work that has been done" and by "dump"?