• For an abelian group $G$, $H$ a finite subgroup of $G$, $x_1, \dots, x_r \in G$ and $L_1, \dots, L_r \in \mathbb N$, let: $P(x ; L) = P(x_1, \dots, x_r ; L_1, \dots, L_r) = \{l_1x_1 + \dots + l_rx_r \mid |l_i| \leq L_i \; \forall i\}$ be a progression of rank $r$. Then: $H + P(x;L) = \{h + p \mid h \in H, \; p \in P(x;L)\}$ is a coset progression of rank r in $G$.

  • Given $x \in \mathbb R / \mathbb Z$, define $|| x ||_{\mathbb R / \mathbb Z} = |\hat x|$, where $\hat x$ is the representative of $x$ in $\mathbb R / \mathbb Z$ with $\hat x \in [-\frac{1}{2}, \frac{1}{2}]$

  • For a finite abelian group $G$, let $\Gamma = \{\gamma_1, \dots, \gamma_r\} \subset \hat G = Hom(G, \;\mathbb R / \mathbb Z)$ and $\rho \in [0, \frac{1}{2}]$. Define the Bohr Set of rank $r$: $B(\Gamma, \rho) = \{g \in G \mid ||\gamma_i(g)||_{\mathbb R / \mathbb Z} \leq \rho, \; \forall i\}$.

  • Given two subsets $A,B$ of some abelian groups $G,H$, a function $f:A \rightarrow B$, we say that $f$ is a Freiman $m$-Homomorphism, if whenever $a_1, \dots, a_m, a_{m+1}, \dots, a_{2m} \in A$ are such that $a_1 + \dots+ a_m = a_{m+1} + \dots +a_{2m}$, then we have $f(a_1) + \dots + f(a_m) = f(a_{m+1}) + \dots + f(a_{2m})$

I am asked to show that every coset progression of rank $r$ is Freiman isomorphic to a Bohr set of rank $r$.

My immediate thought is that since the ranks are related, we must be "constructing" the $\gamma_i$'s using the $x_i$'s.

Additionally, if we had some condition similar to linear independence between the $x_i$'s, then I can see how we may be able to step through a product of finite, cyclic groups to achieve the isomorphism. However, in a situation where there is some sort of "linear relation" with the $x_i$'s, how does this not affect the rank of our final Bohr set?

I am struggling to see how we might be able to construct a finite group, with which we may find some $\Gamma$ and $\rho$ to give us the desired result. Given this, perhaps we can instead start with an arbitrary Bohr Set of rank r, and show that the Freiman isomorphic image is a coset progression of rank $r$, but I believe this is actually false, though I can't seem to find that in my notes currently.

How should I approach this question? How can I effectively construct a finite abelian group, even though it is possible that some of the $x_i$'s have infinite order? Would it be sensible instead then to use $H$ as the starting point, in which case how might I be able to deal with the fact that our $\gamma_i$'s may no longer have the $x_i$'s in their domain?

Any help that may be offered would be very much appreciated, thank you.


Your Answer

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

Browse other questions tagged or ask your own question.