cp's OEIS Frontend

Conjecture: a(n) > 0 for all n > 0. In general, for any n consecutive primes p_k,...,p_{k+n-1}, there always exists a permutation q_k,...,q_{k+n-1} of p_k,...,p_{k+n-1} with q_{k+n-1} = p_{k+n-1} such that the n-1 numbers |q_k-q_{k+1}|, |q_{k+1}-q_{k+2}|,...,|q_{k+n-2}-q_{k+n-1}| are pairwise distinct. (In the case k = 2, this implies that a(n) > 0.)

Clearly there is no permutation a,b,c of 3,5,7 such that the three numbers |a-b|,|b-c|,|c-a| are pairwise distinct. Also, for {a,b} = {7,11}, the three numbers |5-a|,|a-b|,|b-13| cannot be pairwise distinct.

On Aug 31 2013, Zhi-Wei Sun proved the following extension of the general conjecture: Let a_1 < a_2 < ... < a_n be a sequence of n distinct real numbers in ascending order. Then there is a permutation b_1, ..., b_n of a_1, ..., a_n with b_n = a_n such that |b_1-b_2|, |b_2-b_3|, ..., |b_{n-1}-b_n| are pairwise distinct. In fact, when n = 2*k is even we may take (b_1,...,b_n) = (a_k,a_{k+1},a_{k-1},a_{k+2},...,a_2,a_{2k-1},a_1,a_{2k}); when n = 2*k-1 is odd we may take (b_1,...,b_n) = (a_k,a_{k-1},a_{k+1},a_{k-2},a_{k+2},..., a_2,a_{2k-2},a_1,a_{2k-1}).

On Sep 01 2013, Zhi-Wei Sun made the following conjecture: (i) For any n distinct real numbers a_1, a_2, ..., a_n (not necessarily in ascending or descending order), there is a permutation b_1, ..., b_n of a_1, ..., a_n with b_1 = a_1 such that the n-1 distances |b_1-b_2|, |b_2-b_3|, ..., |b_{n-1}-b_n| are pairwise distinct.

(ii) Let a_1, ..., a_n be n distinct elements of a finite additive abelian group G. Suppose that |G| is not divisible by n, or n is even and G is cyclic. Then there exists a permutation b_1, ..., b_n of a_1, ..., a_n with b_1 = a_1 such that the n-1 differences b_{i+1}-b_i (i = 1, ..., n-1) are pairwise distinct.

We believe that part (ii) of the new conjecture holds at least when G is cyclic, and it might also hold when the group G is not abelian.

Note that if g is a primitive root modulo an odd prime p, then for any j = 0,...,p-2 the permutation g^j, g^{j+1},...,g^{j+p-2} of the p-1 nonzero residues modulo p has adjacent differences g^{i+j+1}-g^{i+j} = g^{i+j}*(g-1) (i = 0, ..., p-3) which are pairwise distinct modulo p.

Conjecture: a(n) > 0 for all n > 3. In general, if a_1,...,a_n are n > 2 distinct elements of a finite additive abelian group G with n odd or |G| not divisible by n, then there exists a circular permutation b_1,...,b_n of a_1,...,a_n such that b_1+b_2, b_2+b_3, ..., b_{n-1}+b_n, b_n+b_1 are pairwise distinct.

Note that if g is a primitive root modulo a prime p > 3 then 1+g, g+g^2, ..., g^{p-3}+g^{p-2}, g^{p-2}+1 are pairwise distinct modulo p. So a(p) > 0 for any prime p > 3.

If n > 2 is odd, then 0+1, 1+2, ..., (n-2)+(n-1), (n-1)+0 are pairwise distinct modulo n, and hence the conjecture holds in the case {a_1,...,a_n} = G = Z/nZ.

Conjecture: a(n) > 0 except for n = 3.

If n is a power of two, then a(n) > 0 since the identical permutation 1,2,3,...,n meets the requirement. For any prime p > 3, we have a(p) > 0 since the permutation 1,...,(p-1)/2, (p+3)/2,(p+1)/2,(p+5)/2,...,p meets our purpose.

Let G(n) be the undirected simple graph with vertices 1,...,n which has an edge connecting two distinct vertices i and j if and only if i + j is relatively prime to n. Then, for any n > 2, the number a(n) is just the number of those Hamiltonian cycles in G(n) on which the vertices 1 and n are adjacent.

Let m be any integer relatively prime to n, and let i_k be the smallest positive residue of k*m modulo n. Then i_1, i_2, ..., i_n is a permutation of 1, ..., n with the n adjacent differences i_1-i_2, i_2-i_3, ..., i_{n-1}-i_n, i_n-i_1 all coprime to n.

On Sep 06 2013, the author's two former PhD students Hui-Qin Cao (from Nanjing Audit Univ.) and Hao Pan (from Nanjing Univ.) proved the conjecture fully.