A103314 conjecture

[Paul C. Leopardi]

David,
My comments below.
Best, Paul Leopardi
On Fri, 20 May 2005 02:55 am, David Wilson wrote:
> Right now, I am inclined to believe that if S is a zero-sum subset of the
> nth roots of 1, that n can be built up from (zero-sum) cyclically balanced
> subsets via the following operations:
>
> 1.  A U B, where A and B are disjoint.
> 2.  A - B, where B is a subset of A.
>
> If this were proved, it might allow us to make some headway against a
> general formula for A103314.
>
> Can anyone comment?

Sorry to be so thick, but:
1. By "n can be built up..." do you mean "S can be built up..." ?
2. I understand what you mean by zero-sum. By "cyclically balanced", I think 
you mean that the subset A is of the form

A = { exp(i 2\pi(j+k)/n} | k \in {0,...,m-1}, m|n } for some j \in {0,...,n-1}

Is this correct?