Going From Large To Small

Consider infinite sums of reciprocals

n = 1 1 a n \displaystyle \sum _{ n=1 }^{ \infty }{ \dfrac { 1 }{ { a }_{ n } } }

where a n {a}_{n} are the terms in any of the following infinite sets

1) ( 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , ) = n \;\; (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12,\ldots)=n
2) ( 1 , 5 , 9 , 13 , 17 , 21 , 25 , 29 , 33 , 37 , 41 , 45 , ) = 4 n + 1 \;\; (1, 5, 9, 13, 17, 21, 25, 29, 33, 37, 41, 45,\ldots) = 4n+1
3) ( 2 , 3 , 5 , 7 , 11 , 13 , 17 , 19 , 23 , 29 , 31 , 37 , ) = n \;\; (2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37,\ldots) = n th prime number
4) ( 2 , 3 , 7 , 11 , 19 , 23 , 31 , 43 , 47 , 59 , 67 , 71 , ) = \;\; (2, 3, 7, 11, 19, 23, 31, 43, 47, 59, 67, 71,\ldots) = after 2 , 3 2, 3 , primes of form 4 m + 3 4m+3
5) ( 3 , 5 , 5 , 7 , 11 , 13 , 17 , 19 , 29 , 31 , 41 , 43 , ) = \;\; (3, 5, 5, 7, 11, 13, 17, 19, 29, 31, 41, 43,\ldots)= consecutive twin primes
6) ( 1 , 10 , 11 , 100 , 101 , 102 , 1000 , 1001 , 1002 , 1003 , 10000 , 10001 , ) \;\; (1, 10, 11, 100, 101, 102, 1000, 1001, 1002, 1003, 10000, 10001,\ldots)
7) ( 5 , 7 , 11 , 13 , 11 , 13 , 17 , 19 , 101 , 103 , 107 , 109 , ) = \;\; (5, 7, 11, 13, 11, 13, 17, 19, 101, 103, 107, 109,\ldots)= consecutive prime quadruplets

Twin Primes are pairs of primes of the form ( p , p + 2 ) \; (p, p+2)
Prime Quadruplets are sets of four primes of the form ( p , p + 2 , p + 6 , p + 8 ) \; (p, p+2, p+6, p+8)

All of these may or may not include some terms (not necessarily consecutive) that are in an arithmetic progression of any arbitrary* or infinite length ( both of which are not the same thing! ), but none of the full sequences are an arithmetic-geometric progression with a common ratio r > 1 r>1 .

For your convenience, the numerical values of each for the first 12 terms are

1) 3.10321 \;\; 3.10321\ldots
2) 1.67290 \;\; 1.67290\ldots
3) 1.59272 \;\; 1.59272\ldots
4) 1.28596 \;\; 1.28596\ldots
5) 1.26989 \;\; 1.26989\ldots
6) 1.22481 \;\; 1.22481\ldots
7) 0.82811 \;\; 0.82811\ldots

Some of them may or may not be divergent (although most of you will recognize 1) as being famously divergent), and some of them may or may not be convergent. In any case, the two kinds have been sorted, with all the divergent ones on top and all the convergent ones, if any, on bottom.

Which is the first series listed here that is convergent , going from top 1) to bottom 7) ?

Enter the list number as your answer. If "none", enter 0.

Note: Any arbitrary [integer] means any integer which is not infinite. So, for example, adding any arbitrary integer number of finite terms will always result in a finite total. Whereas, \infty is not an integer number.

1 2 3 4 5 6 7 They are all divergent

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

1 solution

Michael Mendrin
Jan 25, 2017

Bron proved that the infinite sum of reciprocals of twin primes converges, and that is listed as 5)

Dirichlet proved that the infinite sum of reciprocals of primes of the form 4 m + 3 4m+3 diverges, which is listed as 4)

Sets like the ones listed here are each either called a large set if the infinite sum of the reciprocals of the terms diverges, or called a small set if the same converges. See Large Set . Hence "going from large to small" with that list of candidate sets when sorted that way.

I'll add more comments later.

I find it baffling that the infinite sum of reciprocals of primes of the form a + n d a + nd for a n y any coprime positive integer pair ( a , d ) (a,d) will diverge, and yet the twin prime reciprocal series converges. Primes never cease to amaze me.

Brian Charlesworth - 4 years, 4 months ago

Log in to reply

Well, this subject is interesting and fairly fecund. I've always been curious to know just where is the "boundary" between divergent and convergent series. Erdos has a conjecture about that, still an unsolved problem, which is that any such large set contains arithmetic progression of numbers of any arbitrary length. But that does not mean that no small set wouldn't have this property. See 6) , which obviously converges. If Erdo's conjecture is true, then there exists arithmetic progressions of prime numbers of any arbitrary length. How about that.

Michael Mendrin - 4 years, 4 months ago

Log in to reply

O.k., so the Green-Tao theorem is a special case, and Erdos' conjecture covers all large sets. Your question got me thinking also of that boundary between convergent and divergent series. For a given divergent series, does there exist a "minimum" infinite subsets A A that you can filter out to end up with a convergent series. I don't know what metric you would use to establish m i n ( A ) min(A) , as its sum would be necessarily infinite. Subtracting \infty from \infty is fraught with peril.

Brian Charlesworth - 4 years, 4 months ago

Log in to reply

@Brian Charlesworth That's right, the Green-Tao theorem is a special case. But what we don't see very much is the idea of branched sequences, so what about divergence or convergence of such branched sequences? Think of a fractal tree, where each segment is its own sequence. How would one even express that as a sum? The classic sum notation seems woefully inadequate to describe it.

Maybe I'll post a problem having to do with a branched sequence. I got the idea while trying to work on a problem to post here, which I gave up because I realized nobody will ever try to tackle it.

Michael Mendrin - 4 years, 4 months ago

Log in to reply

@Michael Mendrin That would be a cool question. Two thoughts on that theme:

  • with two sequences { a n } \{a_{n}\} and { b n } \{b_{n}\} such that one was convergent and one divergent, we then form a set of hybrid sequences { a b n } \{ab_{n}\} where the n n th term is chosen randomly between a n a_{n} and b n b_{n} . Some of these hybrids would converge and others diverge; I suspect almost all would diverge, but a boundary would be present amidst the hybrids;

  • the random harmonic series converges with probability 1 1 , but it makes me think of your idea of branched sequences. (Divergence is a tail event so happens with probability 0 0 by Kolmogorov's zero-one law, but I don't know if that necessarily rules out the presence of a boundary.) I did post a question based on this here .

Brian Charlesworth - 4 years, 4 months ago

It's even more amazing that we do not know whether there are finite or infinite pairs of twin primes, yet we've calculated the sum of their reciprocals!

Sharky Kesa - 4 years, 4 months ago

Log in to reply

Yes, quite amazing, and the same goes for Sophie Germain primes , (although the estimate, ( 1.54 \approx 1.54 ), for the reciprocal sum of these doesn't have a name attached to it). Here is a list of other well-known series and their reciprocal sums.

Brian Charlesworth - 4 years, 4 months ago

Log in to reply

@Brian Charlesworth Here's an eyebrow-raising quote from that paper

No coincidence that the constant value of the twin primes = 0,6601611815 is practically very near to the spin of the black hole before mentioned

Twin Primes and Black Holes have a connection ? ( ! )

Michael Mendrin - 4 years, 4 months ago

Log in to reply

@Michael Mendrin Haha I always suspected that there was a connection. :)

Brian Charlesworth - 4 years, 4 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...