Yeah, I can't run that. Not even close. I got a Cursor subscription and have been using Grok 4.5. It seems to be the best bang for the buck.
I had it port some math code to CUDA so I could get some larger results. It spit out a number in an hour that would have taken days on my CPU and my CPU isn't slow.
Now I need about 2 days to get the next number 😛 and 4 months to get the one after that and 20 years to get the one after that.
I need more compute!!!
For science!!!
Just to nerd snipe you. The numbers l am looking for are:
The largest gap between successive pairs of twin coprimes in the nth wheel sieve.
Basically you find all the numbers less than the the first n primes multiplied together, p_n#, and coprime to it. Then isolate the twins, then find their central numbers. E.G. 12 for (11,13) and 18 for (17,19) then you find the gap between them. 6 in the example above. For 2*3*5=30 the centers are 0,12,18,30. So the largest gap is 12
The wheels are symmetric so you really only need to search the first half taking care to double the middle gap. So in our example above I count to 15, noting the gap of 12 then count 3 more to get to 15, double that to get 6. 12 is bigger than 6 so keep that. No need to do 30-18, that's just 12 again.
Same with larger wheels n=4, 2*3*5*7=210 and so on. Blows up fast. I have done to n=14.
Why? Well any element of the set < p_n^2 is prime if the maximum gap is < p_n^2 then no matter how well orchestrated the primes are in eliminating twin elements then there will always be some twin elements that are prime.
So I am trying to figure out what grows faster max_gap or p_n^2. So far the ratio p_n^2/max_gap is 3.2 or something and somewhat consistent. But still a long way to infinity.
Daniel Wigton
npub1w4…q30zp
2026-07-21 12:39:34 CEST
in reply to nevent1q…ae64
Author Public Key
npub1w4jkwspqn9svwnlrw0nfg0u2yx4cj6yfmp53ya4xp7r24k7gly4qaq30zpPublished at
2026-07-21 12:39:34 CESTEvent JSON
{
"id": "ce5d8cb353ebdf92dd2182949d3cb58eed4d413e6f23e334b97d0518c5c2bde8",
"pubkey": "75656740209960c74fe373e6943f8a21ab896889d8691276a60f86aadbc8f92a",
"created_at": 1784630374,
"kind": 1,
"tags": [
[
"alt",
"A short note: Yeah, I can't run that. Not even close. I got a Cu..."
],
[
"e",
"14d372e106e763a101b96dc11660013d3cf1687c74b1fea42057d75341de01fa",
"wss://nostr.wine/",
"root",
"d28413712171c33e117d4bd0930ac05b2c51b30eb3021ef8d4f1233f02c90a2b"
],
[
"e",
"f1d60dd28715a98ac5df65989905cfc362430b91fd55c798362ecbb7a7fd08d5",
"wss://theforest.nostr1.com/",
"",
"75656740209960c74fe373e6943f8a21ab896889d8691276a60f86aadbc8f92a"
],
[
"e",
"62d73e72b5d6d2a89121992bbcd4ffd765b3db3ec95855510aedc262fabffb5d",
"wss://theforest.nostr1.com/",
"reply",
"d28413712171c33e117d4bd0930ac05b2c51b30eb3021ef8d4f1233f02c90a2b"
],
[
"p",
"75656740209960c74fe373e6943f8a21ab896889d8691276a60f86aadbc8f92a",
"wss://theforest.nostr1.com/"
],
[
"p",
"d28413712171c33e117d4bd0930ac05b2c51b30eb3021ef8d4f1233f02c90a2b",
"wss://relay.damus.io/"
],
[
"client",
"Amethyst"
]
],
"content": "Yeah, I can't run that. Not even close. I got a Cursor subscription and have been using Grok 4.5. It seems to be the best bang for the buck.\n\nI had it port some math code to CUDA so I could get some larger results. It spit out a number in an hour that would have taken days on my CPU and my CPU isn't slow.\n\nNow I need about 2 days to get the next number 😛 and 4 months to get the one after that and 20 years to get the one after that.\n\nI need more compute!!!\n\nFor science!!!\n\nJust to nerd snipe you. The numbers l am looking for are: \n\nThe largest gap between successive pairs of twin coprimes in the nth wheel sieve.\n\nBasically you find all the numbers less than the the first n primes multiplied together, p_n#, and coprime to it. Then isolate the twins, then find their central numbers. E.G. 12 for (11,13) and 18 for (17,19) then you find the gap between them. 6 in the example above. For 2*3*5=30 the centers are 0,12,18,30. So the largest gap is 12\n\nThe wheels are symmetric so you really only need to search the first half taking care to double the middle gap. So in our example above I count to 15, noting the gap of 12 then count 3 more to get to 15, double that to get 6. 12 is bigger than 6 so keep that. No need to do 30-18, that's just 12 again. \n\nSame with larger wheels n=4, 2*3*5*7=210 and so on. Blows up fast. I have done to n=14.\n\nWhy? Well any element of the set \u003c p_n^2 is prime if the maximum gap is \u003c p_n^2 then no matter how well orchestrated the primes are in eliminating twin elements then there will always be some twin elements that are prime.\n\nSo I am trying to figure out what grows faster max_gap or p_n^2. So far the ratio p_n^2/max_gap is 3.2 or something and somewhat consistent. But still a long way to infinity.",
"sig": "36fc5ab1d22ec97702d8b2894965f268563b23a271e50821e09c15650d19daaec8dedd04f124b3d54029d72d2d1277371ab103ebc6fc8feedf15ee3cdfb0b57d"
}