3
$\begingroup$

First definitions: let $A, B \ \subset \mathbb{Z_{>0}}$ and $1\in A, 1\in B$. We define the relative density of $A$ with respect to $B$ to be $$rel(A, B) = \inf_n \frac{|A \cap [1,n]|}{| B \cap [1,n]|}$$ and we say a set $B$ is an asymptotic additive basis if there exists a postive integer $h$ such that each sufficiently large positive integer can be written as a sum of $h$ (not necessarily distinct) elements of $B$.

$$\text{If $rel(A,B) > 0,$ and $B$ is an asymptotic additive basis, does it follow that $A$ is, too?}$$

I have applications in mind and I would like to know if there is literature on this as it is a natural question to ask once you work with these concepts.

$\endgroup$
4
  • $\begingroup$ Do you want A to be a subset of B? it seems likely but you do not say it. $\endgroup$
    – user9072
    Commented Feb 20, 2016 at 0:25
  • 1
    $\begingroup$ I assume no inclusion. So, $A$ and $B$ are related only via the relative density. Maybe, it would be good to start with a subset and prove that A would be a thinner asymptotic basis. One obvious case is when A is a superset of B and we don't even need density.. $\endgroup$ Commented Feb 20, 2016 at 0:33
  • $\begingroup$ What if $B$ is the set of even numbers and $A$ is the set of even numbers with 2 replaced by 3? $\endgroup$ Commented Feb 20, 2016 at 1:30
  • $\begingroup$ @AnthonyQuas then $B$ does not contain $1$. $\endgroup$
    – user9072
    Commented Feb 20, 2016 at 1:32

1 Answer 1

4
$\begingroup$

Let $B$ be the set of squares. Let $A=\{1\}\cup \bigcup_k (2^{2^{2k}},2^{2^{2k+1}}]$. Then $|A\cap [1,2^{2^{2k}}]|= (2^{2^{2k-1}}-2^{2^{2k-2}})+(2^{2^{2k-3}}-2^{2^{2k-4}})+\ldots+1\approx \sqrt{2^{2^k}}$. These are the points up to which $A$ is sparsest, so that rel$(A,B)>0$.

Of course $B$ is an asymptotic basis of order 4. But because of the large gaps, $A$ is not an asymptotic basis of any order.

$\endgroup$
1
  • $\begingroup$ The sum above is $\approx \sqrt{2^{2^{2k}}}$ so, rel$(A,B)$ becomes $\frac{3}{4}$. (assuming $k$ starts with $0$). I have been convinced. Thank you! $\endgroup$ Commented Feb 20, 2016 at 16:07

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .