Talk:Successive prime differences
Different results for 6,4,2
I am getting slightly different results for the last group than the Python example and am finding it hard to see where I may be wrong. I am finding 337 groups. Am I just mising something? --Thundergnat (talk) 01:20, 27 April 2019 (UTC)
It's my birthday soon so I googled my age and found:
- It's a prime.
- It's a twin prime.
I searched Rosetta Code and found that there was no twin prime task! (I had expected that someone would have already started it). I resolved to wait until closer to my birthday then put up a twin primes task and left it at that.
A few days later I started to think of what a generalisation around the idea of twin primes would be and hit on a difference; then multiple differences; then really liked how my solution to generating a sliding group of <count> items from a list actually did come from the Python fundamentals:
zip(*(lst[n:] for n in range(count)))
I finished the code and played with the differences then firmed up what the task details would become. I wrote the task and added extra explanations and emphasis to try and help the reader grasp the details, then went to bed.
Today I've just done a search of the primes generated from differences of
2, 4 on OEIS to find that it is known to some degree, but expressed differently and not as generally as here - I guess recreational maths peeps think alike :-)
- It is very well studied, but you must state it slightly differently. Let P2 be the infinite sequence of successive primes (p2_a,P2_b) such that P2_b-P2_a=2. and P4 be the similar infinite sequence (P4_a,P4_b) such that P4_b-P4_a = 4. The your generalization to P2P4 as 3 successive primes with Pa,Pb,Pc with Pb-Pa=2 and Pc-Pb=4 is a search through P2 and P4 to find P2_b=P4_a. An interesting study would be to compute over a large range the length of P2 and P4 and thus predict the length of P2P4. For a given range should the length of P2P4 be the same as P4P2?--Nigel Galloway (talk) 13:28, 27 April 2019 (UTC)
- I was going to add a twin prime task (and cousin prime task, a difference of four), but was somewhat preempted with addition of the sexy prime task (a difference of six), so I dithered a bit. There are other named difference primes such as devil (also called beast), centennial, and millennial primes. However, having a Rosetta Code task just for twin primes would make the code a lot cleaner and simpler, not to mention faster. This would've made the task solutions more easier to compare (and I think more useful for people who wanted a clean and robust code for just concerning the generation of twin primes). Plus it would be easier to find when people are looking for a simple twin prime generator. -- Gerard Schildberger (talk) 21:25, 27 April 2019 (UTC)