Here I ask the following question in additive combinatorics.

QUESTION: Let $A$ be any finite subset of an additive abelian group $G$ with $|A|=n>3$. Can we write $A$ as $\{a_1,\ldots,a_n\}$ so that all the $n$ sums \begin{gather*}a_1+a_2+a_3,\ a_2+a_3+a_4,\ \ldots, \\ a_{n-2}+a_{n-1}+a_n,\ a_{n-1}+a_n+a_1,\ a_n+a_1+a_2\end{gather*} are pairwise distinct?

Note that if $G=\{a_1,\ldots,a_n\}$ is an addtive abelian group with $|G|=n>3$ then $2(a_1+\ldots+a_n)=0$ and hence \begin{align*}&(a_1+a_2+a_3)+\ldots+(a_{n-2}+a_{n-1}+a_n) \\&+(a_{n-1}+a_n+a_1)+(a_n+a_1+a_2) \\=&3(a_1+\ldots+a_n)=a_1+\ldots+a_n.\end{align*} This is the main motivation of my question.

In 2013 I conjectured that the question always has a positive answer, and showed this when $G$ is torsion-free (cf. http://arxiv.org/abs/1309.1679). However, I'm even unable to answer the question for $G=\mathbb Z/p\mathbb Z$ with $p$ prime.

Any comments are welcome!


Let $A$ consist of $6$ elements of $(\mathbb Z_2)^n$ that sum to $0$. Then $a_1 + a_2 + a_3 = a_4 + a_5 + a_6$, no matter how you permute $A$.

  • $\begingroup$ thank you for your counterexample for $G=\mathbb Z_2^n$. I believe that the conjecture holds if $G$ is cyclic or it contains no involution. $\endgroup$ – Zhi-Wei Sun Jul 12 '18 at 13:31

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.