Partitioning the Positive Integers

I want to partition the set of positive integers { 1 , 2 , 3 , } \{1,2,3,\ldots\} into subsets such that each subset satisfies the property that if a number x x is in it, 2 x 2x is not.

What is the minimum possible number of subsets that I can get?

2 3 4 There is no minimum

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.

9 solutions

Jam M
Nov 20, 2018

Every positive integer has a unique representation 2 k m 2^k m , where k N { 0 } k \in \mathbb{N} \cup \{0\} and m m is odd. If A 1 = { 2 k m k is even and m is odd } A_1 = \{ 2^k m \; | \; \mbox{k is even and m is odd} \} and A 2 = { 2 k m k is odd and m is odd } A_2 = \{ 2^k m \; | \; \mbox{k is odd and m is odd} \} , then A 1 A 2 = N A_1 \cup A_2 = \mathbb{N} and A 1 A 2 = A_1 \cap A_2 = \emptyset . Furthermore, since multiplying by 2 2 changes the parity of k k , x A i 2 x ∉ A i x \in A_i \Rightarrow 2x \not\in A_i

I think there's a typo in the index of A 2 A_2

Henry U - 2 years, 6 months ago

Thanks. I fixed it.

Jam M - 2 years, 6 months ago

Proof of the first statement?

Orlando Moreno - 2 years, 6 months ago

Log in to reply

Trivial corollary of the fundamental theorem of arithmetic.

Jesse Nieminen - 2 years, 6 months ago
Binky Mh
Dec 3, 2018

It only requires 2 2 subsets:

Go through the numbers in order. It doesn't matter which subset the odd numbers are put into. Simply place each even number n n into whichever subset doesn't contain n 2 \frac{n}{2} .

Jeremy Galvagni
Nov 21, 2018

You can prove n=2 is possible inductively. Once you've placed the numbers 1,2,...,2m you can place 2m+1 in either set and 2m+2 in whichever set m+1 is not in.

Induction does not work here. Induction can only proved that a partition exists for an arbitrarily large finite set, not infinite sets like this one. See https://math.stackexchange.com/questions/98093/why-doesnt-induction-extend-to-infinity-re-fourier-series. This same logic applies to invalidate Binky's proof.

Alex Li - 2 years, 6 months ago

Log in to reply

I don't understand. If x is always in the opposite subset to 2x & x/2, how can the desired property not be satisfied?

Binky MH - 2 years, 6 months ago

Log in to reply

Well, your argument is basically induction as well. You show that once you construct a 2 subsets up to some number X-1, you can put X into one of the other two subsets. That reasoning is correct. However, then you continue that logic 'until you get to infinity', but this somehow does not work.

I tried to come up with a example similar to yours where it will not work for clarity, though I'm not sure how helpful it is.

Claim: There is a odd number that is greater than all but at most one even numbers.

Proof: If we go through the numbers in order, when we add an odd number, this number is greater than all even numbers, and when we add an even number, the last odd number is still greater than all but this even number.

The logic in this proof is also sound besides the 'going through numbers in order', but it is clear that it is incorrect, since any odd number x is less than the even numbers x+1 and x+3.

Alex Li - 2 years, 6 months ago

Log in to reply

@Alex Li But in your scenario, the relationship between x and an infinite number of other numbers affects whether the claim is true of x. In the actual problem, there are a finite number of relationships which affect whether it is true of x.

Binky MH - 2 years, 6 months ago

@Alex Li And if this kind of logic is fallible, how can an infinite series of numbers be placed into multiple subsets of non-finite length in the first place?

Binky MH - 2 years, 6 months ago

Log in to reply

@Binky Mh I could not come up with a scenario perfectly like this one, because, well, this one happened to be true. There may be a way to add some clauses to the process of induction to make your argument work in this specific case, though I don't think I've ever seen induction used on infinite sets so I'm not too sure. Jam M provides a proof which does not require any induction, by defining the whole set at once.

Alex Li - 2 years, 6 months ago

Log in to reply

@Alex Li I suggest you look at transfinite induction. Essentially the same as ordinary induction but you also have to show that the limit as n tends to infinity is also valid (actually it's not that simple and requires some quite deep set theory to demonstrate but that's broadly true).

Richard Farrer - 2 years, 6 months ago

@Alex Li So it seems the issue stems from the fact that I implied the categorisation, rather than explicitly stating it.

Binky MH - 2 years, 6 months ago

Some nice counterexamples on that link. Let me see if I understand.
Induction can't be used to place all the numbers one at a time. As in Russell's train example, there is no last car. You shouldn't use induction for infinite sets.

Induction more for proving formulas that work for any finite number.

Jeremy Galvagni - 2 years, 6 months ago

Log in to reply

Yep, that's exactly the point I was trying to make.

Alex Li - 2 years, 6 months ago
Praveen Kumar
Dec 5, 2018

Let's assume there will be 2 subsets required !

  1. In first let's take all odd nos.

    We can't take doubles so let's take.
    quadruples(4th multiples) of all odds.

    And similarly, 16th , 64th, 128th.......
    multiples of odds.
    (because we can't take 8th multiples as it'll.
    be double of 4th multiples and so on! )


  2. Our 2nd set is already created as it is the set of
    remaining nos. It'll contain all doubles of odds, 8th multiples.
    of Odds and so on. Hence both set Fulfil the mentioned condition.

Will this partition contain numbers like 15, 35, 210 etc.? Kindly elucidate. Because if it does not, then it does not contain all the numbers to be partitioned?

Giri V K - 2 years, 6 months ago

Log in to reply

Yeah bro! Thanx and I've eddited my answer I replaced primes with odd nos. I hope it's correct now

Praveen Kumar - 2 years, 6 months ago
Robert DeLisle
Dec 8, 2018

Using the fundamental theorem of arithmetic, each positive integer can be uniquely represented in the form ( 2 k ) m (2^k)m where m is odd (being the product of all the prime factors that are not 2, or 1 in the case of powers of 2) and k = 0,1,2,... .

Define two partitions with k = 0,1,2,... and m is an odd positive integer: P 1 P_1 = { ( 2 k ) m (2^k)m | k is even} and P 2 P_2 = { ( 2 k ) m (2^k)m | k is odd}

For each n = ( 2 k ) m (2^k)m , 2n = ( 2 k + 1 ) m (2^{k+1})m and by definition in the opposite partition from n, since k and k+1 have opposite odd/even parity.

Affan Morshed
Dec 7, 2018

This is my method which is not nearly as simple and easy as Binky Mh's solution, and is quite unnecessarily long, but I decided to write down my own solution anyways(for practice. Also do not that creating this solution was much easier than communicating it (in fact, it took a very little amount of time)). Note that, for this solution, 0∈N.We will also be calling the 2 sets G and W.The way you can do this is to first partition the set {x:x∈N} into an infinite series of mutually exclusive sets {x*(2^n):n∈N} by first creating such a set for x=1, than repeating the cycle of creating a new of such set for the lowest x not in any previous set, forever. This works as every x will always have a unique prime factorization which will never include 2 (as otherwise x/2 would have already been in a previous set), and each set is just a list of numbers who's prime factorization is that of x with additions (this term is very confusing, but I am too lazy to think up a better term) of 2, so will be different as their prime factorization will be different as their prime factorization excluding 2 will be different, it is also quite obvious how the lowest possible value not in any previous set will incorporate all possible values over the infinite cycles. For each set {x(2^n):n∈N}, we take the smallest value and put it into G, and than repeat the cycle where the next smallest value (which will be double the one we previously put in, hence making it necessary to put into the other set (but as it is the other set, and this is the only value double the previous value, we will validate the previous value)) into the set (either G or W) which is not in where you just put the previous smallest value (which is now gone as you put it into the other set) forever, the infinite nature of this means every value will be included in W or (exclusive) G, but the double of this value will be in the other set which is not your set, and we will be doing this for all the series of nth_term=x(2^n)/set {x(2^n):n∈N} so hence partitioning all positive integers.

谦艺 伍
Dec 5, 2018

Partition the set of all positive integers into many geometric series with ratio 2. Consider 2 sets: A and B. For each geometric series, starting from its initial element, we put each element alternately into set A and B. Then we are done.

William G.
Dec 3, 2018

Basically separate the odd and even numbers.

This doesn't work, all doubles of even numbers are in the same set as their halves.

Paul Evans - 2 years, 6 months ago
K T
Dec 2, 2018

We need two 2 subsets. Put any odd number x in either set, at will. The constraint only demands that 2x be in the other subset, but we could put 4x together with x, and any mx where m is a power of 4. Put 2x,8x,32x...(2mx) into the other.

The set is infinite, so you will need infinite subsets. For each positive integer x there is a value 2x that lies within the same infinite set of positive integers. Also these infinities have a one to one correspondence, so these infinities would require infinite subsets to separate.

Atlantean Dreamer - 2 years, 5 months ago

Yes, both subsets are each infinite in size, But we only need 2 sets.

K T - 2 years, 5 months ago

The simplest way is put all odd numbers, together with all 4-folds, 16-folds, 64-folds of odd numbers ... in set A and all 2-folds, 8-folds, 32-folds of odd numbers, ... in set B.

K T - 2 years, 5 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...