Twin prime gap bounds

[gnostique]

I'm working with prime and twin gap bounds and it appears that there may be an approach to establishing maximum bounds on the distance to the next pair (as per convention, a pair is denoted by the first member). It was possible to verify the results with the first 100k pairs but that's getting close to the limit of the capacity. We need to verify the maximum known relative gap between consecutive pairs i.e. (t_next - t) / t. According to the results, the ratio is bounded by a specific number so any data on large twin gaps or references to such would be much appreciated and I checked the related threads as well. I'll gladly share the details if the results are confirmed.

[CRGreathouse]

I think you're looking for A113275 and A113274.

[mart_r]

I'm still pondering about updating the twin prime gap sequences (there are six of them that I know of, so while submitting it all, my mailbox will get flooded...). I just can't decide how to go about the credits. Could I just give a link to the twin prime gap search thread?

@ CRGreathouse: you're an editor of OEIS, right? What if I just post the b-files here, would that be okay (and perhaps less work for everyone)?
Knowing myself, I probably spend hours checking whether I filled the forms right if I submit it the regular way.

[gnostique]

@CRGreathouse, thank you this was helpful. From these results it appears that the relative gap tends to decrease with the prime order and this is also in a good agreement with our results.

[gnostique]

Is it acceptable to post directly related research here for information / comments / feedback?

[mart_r]

Originally Posted by gnostique

Is it acceptable to post directly related research here for information / comments / feedback?

I guess so. It's what I do most of the time.
Here's the most recent data on twin prime gaps:
https://www.mersenneforum.org/showpo...&postcount=223
And the search goes on...

[gnostique]

Some recent results on estimates of twin prime gaps. Appreciate comments and feedback.

[CRGreathouse]

Originally Posted by mart_r

I'm still pondering about updating the twin prime gap sequences (there are six of them that I know of, so while submitting it all, my mailbox will get flooded...). I just can't decide how to go about the credits. Could I just give a link to the twin prime gap search thread?

@ CRGreathouse: you're an editor of OEIS, right? What if I just post the b-files here, would that be okay (and perhaps less work for everyone)?
Knowing myself, I probably spend hours checking whether I filled the forms right if I submit it the regular way.

Yes, I am an OEIS Editor-in-Chief. If you need help submitting something, feel free to PM me.

[mart_r]

Originally Posted by CRGreathouse

Yes, I am an OEIS Editor-in-Chief. If you need help submitting something, feel free to PM me.

That doesn't quite answer all of my questions, but since sending files via PM isn't allowed (AFAICT) I'm going to make a regular submission in the next couple of days. Or weeks. When I'm in the right mood to do so.


P.S.:
Here's a tidbit for further study:

Maximal gap between twin coprimes mod p#

 p#  gap  #inst. first instance after
 3#    6       1                6 ± 1
 5#   12       2               18 ± 1
 7#   30       2               72 ± 1
11#   42       4              900 ± 1
13#   66      12              732 ± 1
17#  108      20              702 ± 1
19#  150      20              660 ± 1
23#  204       4         76166568 ± 1
29#  258       2       1205437110 ± 1
31#  348       4       8813641452 ± 1
37#  528       2     544899485412 ± 1
41#  540      16   14416696724700 ± 1
43#  618       8  830330079152052 ± 1
47#  702     (?)
53#  816     (?)
59#  912+    (?)

Not yet sure about 47#+. But before I try to go any further and spending time that I currently don't have, I'm curious to know whether anyone has already investigated this.

[CRGreathouse]

If you'd like me to look anything over, post the A-number here or PM me once it's up. (If you're happy with how it looks, submit for review at the same time; if not, leave it in /draft while you have me look.)