Let $\mathbb{N}$ denote the set of positive integers. For any prime $p\in\mathbb{N}$ let $p\mathbb{N} = \{np: n\in \mathbb{N}\}$. Is there a partition ${\cal P}$ of $\mathbb{N}\setminus\{1\}$ such that for all $B \in {\cal P}$ and every prime $p\in\mathbb{N}$ we have $|B \cap p\mathbb{N}|=1$?
-
$\begingroup$ Oops, I missed the every prime condition earlier. Start assembling sets of coprime integers. If you can't add a number to one of the existing sets, start a new set. In particular, each set will have at most one power of a given prime. Gerhard "Thinks Jumping Primes Are Cooler" Paseman, 2020.05.11. $\endgroup$– Gerhard PasemanCommented May 11, 2020 at 19:33
-
$\begingroup$ @GerhardPaseman, if each member of $B$ contains only one element, then, for each $B$, $\lvert B \cap p\mathbb N\rvert$ will equal $0$ for most primes $p$. On edit: it seems that your revised construction is just what I described, isn't it? $\endgroup$– LSpiceCommented May 11, 2020 at 19:34
-
1$\begingroup$ Yes, I misread, and so seriously edited my comment above. It may be possible to give a slicker proof than what you have posted as an answer, but I'm not seeing it. Gerhard "More Prime Powers To You" Paseman, 2020.05.11. $\endgroup$– Gerhard PasemanCommented May 11, 2020 at 19:42
2 Answers
Recursively define a sequence of $B$'s as follows. Initially, each is empty. At each step $n > 1$, place $n$ in the first $B$ that contains only elements coprime to $n$. Clearly, for each prime $p$, there is no $B$ that contains two distinct multiples of $p$. Now fix a prime $p$ and a natural number $N > 1$, and consider the first $B$ that contains no multiple of $p$ after step $N$ has completed. The first power $p^k$ of $p$ that is larger than $N$ cannot be placed in any earlier $B$ (since all have a multiple of $p$), so it will be placed in $B$ if no multiple $p d$ of $p$ with $N < p d < p^k$ has been.
As @StevenStadnicki points out, it's interesting to investigate the structure of these sets. (I started to do it by hand, and found it sort of addictive.) Here's some Haskell code to allocate the first $N$ numbers (doubtless both inefficient and unidiomatic, but it seems to work):
insert N [] = [(N, [N])]
insert N ((c,bs):bss) = if gcd c N == 1 then (N*c,N:bs):bss else (c,bs):(insert N bss)
insertTo 1 = []
insertTo N = insert N $ insertTo (N - 1)
One runs it as
map snd $ insertTo 1000
(for example), whose output starts
[[997,991,983,977,971,967,953,947,941,937,929,919,911,907,887,883,881,877,863,859,857,853,839,829,827,823,821,811,809,797,787,773,769,761,757,751,743,739,733,727,719,709,701,691,683,677,673,661,659,653,647,643,641,631,619,617,613,607,601,599,593,587,577,571,569,563,557,547,541,523,521,509,503,499,491,487,479,467,463,461,457,449,443,439,433,431,421,419,409,401,397,389,383,379,373,367,359,353,349,347,337,331,317,313,311,307,293,283,281,277,271,269,263,257,251,241,239,233,229,227,223,211,199,197,193,191,181,179,173,167,163,157,151,149,139,137,131,127,113,109,107,103,101,97,89,83,79,73,71,67,61,59,53,47,43,41,37,31,29,23,19,17,13,11,7,5,3,2],
[961,841,529,361,289,169,121,49,25,9,4],
[667,323,143,35,6],[899,437,221,77,15,8],
[713,247,187,21,10],[551,391,91,55,12],
[851,493,209,65,27,14],[299,133,85,33,16],
[377,253,119,95,18],[703,527,319,161,39,20],
[989,779,629,403,203,45,22],
[893,731,533,407,217,115,24],
[943,817,341,259,125,51,26],
[799,481,451,145,57,28],
[901,611,589,473,287,30] …
-
2$\begingroup$ Running through this construction by hand, $B_1$ is (of course) the primes; $B_2$ appears to be the squares of the primes (and I suspect a proof would be pretty straightforward), and of course $2n\in B_n$. The structure of the $B$s after that point appears to get much more complicated, although for instance $m=12n+3\in B_{(m+1)/2}$ and $m=12n+9\in B_{(m-1)/2}$... $\endgroup$ Commented May 12, 2020 at 0:42
-
$\begingroup$ Indeed, I found it sort of fun to do it by hand, although it gets boring allocating each even number its own set. I'll put the Haskell code I used to generate examples in the post. $\endgroup$– LSpiceCommented May 12, 2020 at 1:26
Place each $n$ that is not a prime power into its own $B=B(n)$. Then fill the rest of $B(n)$ with powers of primes that do not divide $n$.