There exists such a subset $A$ of the reciprocals of natural numbers $\{\frac{1}{n} \ |\ n \in \mathbb N\}$ that any real number $x$ on the interval $[0,1]$ can be expressed as sum of members of some subset $B_x$ of $A$ in such a way that all elements of $B_x$ are distinct, i.e. without any repeating terms in the sum. Call this as the property.

An example of a set $A$ having the property would be the inverses of powers of $2$. This is easy to see, as every number $x$ on the interval $[0,1]$ can be expressed in base two, using any power of $2$ only once, and these numbers form the set $B_x$.

Thus the set of reciprocals of any subset of $\mathbb N$, that has the powers of $2$ as a subset, also has the property.

There are also subsets of reciprocals of natural numbers that do not have the property. Take inverses of powers of $3$ as an example. If we try to express $\frac{1}{5}$ in this way, we note that $$\sum _{n=2}^{\infty } \frac{1}{3^n} = \frac{1}{6}<\frac{1}{5}$$ and $$\frac{1}{3}>\frac{1}{5}.$$ We conclude that we cannot avoid repeating a term.

Question 1) Consider the reciprocals of prime numbers. Does this set have the property?

Question 2) Clearly the sum over all the members of $A$ has to be greater or equal to $1$ for the set $A$ to have the property. Is this enough? What are the necessary and sufficient conditions for a set to have the property?


(1) Yes, because the sum of all of them diverges. There was a question recently where it was shown that this is a sufficient condition.

(2) it is not sufficient because if A includes the reciprocals of {2,3,4} and then reciprocals of powers of 1000 then there are number less than $\frac{1}{4}$ which are not expressible. I conjecture that being able to express just $1$ as the sum of elements of $A$ is sufficient, if $A$ is infinite and you express $1$ as an infinite sum.

  • $\begingroup$ 1/2+1/3+1/6 = 1. You need more than that. $\endgroup$
    – user24142
    Apr 22 '15 at 20:22
  • $\begingroup$ @user24142: 1 should also be expressible as an infinite sum $\endgroup$
    – wlad
    Apr 22 '15 at 20:34
  • $\begingroup$ That feels like a possible equivalent condition. $\endgroup$
    – user24142
    Apr 22 '15 at 20:35
  • $\begingroup$ @ user3491648: Can you give me the recent question you mention in (1), please? $\endgroup$
    – Piquito
    Apr 25 '15 at 0:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.