Do there exist a nontrivial group homomorphism from $(\mathbb{Q}^*,\cdot)$ and $(\mathbb{Z},+)?$

103 Views Asked by At

My Thought:

Let $\phi:\mathbb{Q}^* \to \mathbb{Z}$ be a nontrivial homomorphism. Then I have observed the following properties:

  1. $\phi(p^n)=n\cdot\phi(p) \ \forall n \in \mathbb{N}$.

  2. $\phi(p^{-n})=n\cdot\phi(p^{-1}) \ \forall n \in \mathbb{N}$.

So $\phi$ can be completely determined by its values in $p$ and $p^{-1}$ for all prime $p$.

Is there some other direction to think and get some thing else? Please provide me some hints. Thank you.

1

There are 1 best solutions below

7
On

Yes, there are nontrivial group homomorphisms. In fact, they can be nicely classified.

Consider the function $\phi_2 : \mathbb{Q}^* \to \mathbb{Z}$, where $\phi_2(q)$ is defined as the unique integer such that $q \cdot 2^{- \phi_2(q)}$ can be written as $\frac{a}{b}$ with $a, b$ odd.

Then $\phi_2$ is a group homomorphism.

Note that it's possible to generalise $\phi_2$ to $\phi_p$ for any prime number $p$. So this gives us a broad class of nontrivial group homomorphisms.

In fact, it can be shown that any group homomorphism $\mathbb{Q}^* \to \mathbb{Z}$ must be of the form $\sum\limits_{n = 0}^\infty a_n \phi_{p_n}$, where $\{p_n\}_{n \in \mathbb{N}}$ is the sequence of prime numbers and $a_n$ is a sequence of integers. Note that this infinite sum makes sense because when we compute $\sum\limits_{n = 0}^\infty a_n \phi_{p_n}(a / b)$, only finitely many values $\phi_{p_n}(a/b)$ are nonzero.

With some more analysis, it turns out that $\mathbb{Q}^*$ is isomorphic to $(\mathbb{Z}/2 \mathbb{Z}) \oplus \bigoplus\limits_{n \in \mathbb{N}} \mathbb{Z}$, where the injections send the nontrivial element of $\mathbb{Z}/2\mathbb{Z}$ to $-1$ and the $n$th injection $\mathbb{Z} \to \mathbb{Q}^*$ sends $i$ to $(p_n)^i$. This makes classifying the group homomorphisms $\mathbb{Q}^*$ to $\mathbb{Z}$ quite simple, since we have

$\begin{equation} \begin{split} Hom(\mathbb{Q}^*, \mathbb{Z}) &\simeq Hom((\mathbb{Z}/2 \mathbb{Z}) \oplus \bigoplus\limits_{n \in \mathbb{N}} \mathbb{Z}, \mathbb{Z}) \\ &\simeq Hom(\mathbb{Z}/2 \mathbb{Z}, \mathbb{Z}) \times Hom(\bigoplus\limits_{n \in \mathbb{N}} \mathbb{Z}, \mathbb{Z}) \\ &\simeq Hom(\mathbb{Z}/2 \mathbb{Z}, \mathbb{Z}) \times \prod\limits_{n \in \mathbb{N}} Hom(\mathbb{Z}, \mathbb{Z}) \end{split} \end{equation}$

and since $Hom(\mathbb{Z} / 2 \mathbb{Z}, \mathbb{Z}) \simeq 0$, and since $Hom(\mathbb{Z}, \mathbb{Z}) \simeq \mathbb{Z}$, we see that the final term is isomorphic to $\prod\limits_{n \in \mathbb{N}} \mathbb{Z} \simeq \mathbb{Z}^\mathbb{N}$.