Merry Christmas and a prime! (M50 related)

[DanielBamberger]

Originally Posted by masser

I think astronomy was mentioned earlier, but consider the 5 months between LIGO's first observation and the announcement.

https://en.wikipedia.org/wiki/First_...tational_waves

At eight days, I think George and company are doing pretty well.

Definitely! It's all "complaining on a high level".

Here are two of my favourite bad examples from astronomy (I hope this isn't too far off-topic).
Some conclusions you might draw: People are impatient; preventing leaks is hard; but releasing preliminary results can cost you news coverage. Some people draw very different conclusions, and the debate is ongoing!

LIGO's long proprietary period, combined with the high profile target, has caused some problems (including leaks, rumours, and gossip). Take this discovery it made last year: https://www.nature.com/news/rumours-...ghting-1.22482

The problem is that LIGO has a lot of false-positives, and checking for errors is a process that is very time consuming. There is little to no way to improve that.
The only alternative would be to release preliminary results; and that's a dangerous thing to do, given the likelyhood of a false alert.

Or take this example: https://en.wikipedia.org/wiki/Contro...very_of_Haumea The question "who discovered the dwarf planet Haumea?". A long proprietary period, combined with a leak, costs the discoverers the right to name the object.
Could the proprietary period have been shortened: Yes. But this again would have resulted in the release of preliminary results. It's not an easy decision to make.

[Madpoo]

Originally Posted by masser

...
At eight days, I think George and company are doing pretty well.

Yup, that.

I looked back at prior press releases to see what the gap was between discovery and press release (I'm going by the date listed in the press release):
M35 - 10 days
M36 - 8 days
M37 - 6 days
M38 - 29 days
M39 - 22 days
M40 - 8 days
M41 - 15 days
M42 - 9 days
M43 - 9 days (announced Christmas eve...not the best timing probably)
M44 - 7 days
M45 - 9 days (twin announcement with M47)
M46 - ?? - the press release is dated *before* the prime was found
M47 - 23 days (twin announcement with M45)
M48 - 11 days
M49 - 12 days (I think...the press release on the site doesn't have a date...weird)

The press releases *should* have the date the release goes out, not the date of discovery. The twin release didn't even list the discovery dates of each one which is unfortunate, but still available online anyway. And the lack of a "dateline" on M49 is unfortunate... I'll try to figure out when that went out and edit the copy on Primenet at least since it really ought to be there.

Anyway, as you can see, it does vary (widely in some cases) but what we're looking at for this one is well within the norm, so people need to keep calm and carry on.

EDIT: Oh... M46 was the weird one where it went unnoticed from April to June. The press release lists the April date (it *should* have the date of the press release, with the actual details in the body, but oh well).

[AG5BPilot]

If I might add an outsider's perspective...

At PrimeGrid, we have some projects where the algorithm is prone to producing false primes, and other projects where it's not. The projects where false prime are not likely are, in fact, so unlikely that we've never seen one at all. So either a particular type of test yields false primes frequently, or it never has.

For those tests that don't produce false primes, we often tease the audience about them. We're not going to end up with egg on our faces, and people enjoy guessing.

For tests that are prone to false positives, we don't say anything at all. Heck, we usually don't believe they're real ourselves until we've verified it. Even when it's coming from a previously 100% reliable computer. Only once it's independently verified do we say anything.

Obviously, we don't test Mersenne's, but I believe they'd fall into the "false primes are likely" category. We wouldn't say anything at all until it's verified.

Anyway -- congratulations, yet again!!!

[GP2]

Originally Posted by DanielBamberger

Or take this example: https://en.wikipedia.org/wiki/Contro...very_of_Haumea The question "who discovered the dwarf planet Haumea?". A long proprietary period, combined with a leak, costs the discoverers the right to name the object.

Actually, they did name it, in the end.

[DanielBamberger]

Originally Posted by GP2

Actually, they did name it, in the end.

Their name proposal was accepted, in what I'd call a a salomonic decision (it really cost them discovery credit, though, and to this day the discoverer of Haumea is undefined).

[rudy235]

I do not wish to relitigate that matter but while the name Haumea was indeed the one suggested by professor Brown, the minor planter center attributes the discovery to the Sierra Nevada observatory.

https://minorplanetcenter.net/db_sea...ject_id=136108

[DanielBamberger]

Originally Posted by masser

consider the 5 months between LIGO's first observation and the announcement.

https://en.wikipedia.org/wiki/First_...tational_waves

To wrap up the astronomy discussion:
5 months is a pretty short proprietary period, for astronomy standards. I have seen cases that were closer to 5 years (with really no reasonable way to cut that down).

I am now just looking forward to the GIMPS press release. Maybe tomorrow, maybe not.

[Mysticial]

Nice! Is this one going to be a new largest? Or do we have another case of out-of-order discovery?

[ewmayer]

Originally Posted by Mysticial

Nice! Is this one going to be a new largest? Or do we have another case of out-of-order discovery?

Happy new, year Alex - Search backward through the thread for mentions of FFT length of, and roundoff errors in, airsquirrels' gpuOwl verify run - comparing that to that to the ROEs/max-p of Prime95 at the same FFT length should tell you what you need to know.

[Batalov]

You cannot unsee this once you see it --

[ATH]

Alas, no twin prime, not surprisingly though:

2^7xxxxxxx-3 is not prime. RES64: xxxxxxxxxxxxxxxx. Wg8: F9C5D09F,00000000