[step:Prove that $\mathbb{Q}/\mathbb{Z}$ is an injective cogenerator of abelian groups]
Let $D$ denote the abelian group $\mathbb{Q}/\mathbb{Z}$.
[claim:$D$ is injective as a $\mathbb{Z}$-module]
[/claim]
[proof]
Let $A$ be an abelian group, let $B \subset A$ be a subgroup, and let $\varphi: B \to D$ be a group homomorphism. We prove that $\varphi$ extends to a group homomorphism $\Phi: A \to D$.
Let $\mathcal{P}$ be the set of pairs $(C,\psi)$ such that $C$ is a subgroup of $A$ with $B \subset C$ and $\psi: C \to D$ is a group homomorphism satisfying $\psi|_B=\varphi$. Order $\mathcal{P}$ by declaring $(C,\psi) \le (C',\psi')$ when $C \subset C'$ and $\psi'|_C=\psi$. Every chain in $\mathcal{P}$ has an upper bound obtained by taking the union of the subgroups in the chain and the uniquely induced homomorphism on that union. By [Zorn's lemma](/theorems/1226), choose a maximal element $(C,\psi)$ of $\mathcal{P}$.
We show $C=A$. Suppose not, and choose $a \in A \setminus C$. Let $C' := C+\mathbb{Z}a$ be the subgroup generated by $C$ and $a$. Define
\begin{align*}
N := \{n \in \mathbb{N} : n a \in C\}.
\end{align*}
If $N=\varnothing$, then every element of $C'$ has a unique form $c+n a$ with $c \in C$ and $n \in \mathbb{Z}$. Choose any element $u \in D$, and define $\psi': C' \to D$ by
\begin{align*}
\psi'(c+n a) := \psi(c)+n u.
\end{align*}
The uniqueness of the representation makes $\psi'$ well-defined, and direct addition of representatives shows that $\psi'$ is a group homomorphism extending $\psi$, contradicting maximality.
It remains to treat the case $N \ne \varnothing$. Let $m := \min N$. Then $m a \in C$. We claim that every relation $c+n a=0$ with $c \in C$ and $n \in \mathbb{Z}$ satisfies $m \mid n$. Indeed, if $n=qm+r$ with $q \in \mathbb{Z}$ and $0 \le r < m$, then
\begin{align*}
r a = n a - q m a \in C.
\end{align*}
By the minimality of $m$, this forces $r=0$. Since $D=\mathbb{Q}/\mathbb{Z}$ is divisible, there exists $u \in D$ such that
\begin{align*}
m u = \psi(m a).
\end{align*}
Define $\psi': C' \to D$ by
\begin{align*}
\psi'(c+n a) := \psi(c)+n u.
\end{align*}
To verify well-definedness, suppose $c+n a=c'+n'a$. Then $(c-c')+(n-n')a=0$, so $m \mid n-n'$. Write $n-n'=qm$. Since $(n-n')a=-(c-c')$, we have $qma=c'-c$. Therefore
\begin{align*}
\psi(c)-\psi(c')+(n-n')u
&= \psi(c-c')+qm u \\
&= \psi(c-c')+q\psi(m a) \\
&= \psi(c-c'+qma) \\
&= \psi(0) \\
&=0.
\end{align*}
Thus $\psi(c)+nu=\psi(c')+n'u$, so $\psi'$ is well-defined. It is a group homomorphism by direct addition of representatives and extends $\psi$, again contradicting maximality. Hence $C=A$, and $D$ is injective as a $\mathbb{Z}$-module.
[/proof]
[claim:$D$ separates points of abelian groups]
[/claim]
[proof]
Let $A$ be an abelian group and let $a \in A$ be nonzero. Let $C:=\mathbb{Z}a$ be the cyclic subgroup generated by $a$.
If $a$ has infinite order, define $\theta: C \to D$ by
\begin{align*}
\theta(n a) := n\left(\frac{1}{2}+\mathbb{Z}\right).
\end{align*}
Then $\theta(a)=\frac{1}{2}+\mathbb{Z} \ne 0$.
If $a$ has finite order $m \ge 2$, define $\theta: C \to D$ by
\begin{align*}
\theta(n a) := \frac{n}{m}+\mathbb{Z}.
\end{align*}
This is well-defined because $m a=0$ and $m(\frac{1}{m}+\mathbb{Z})=0$, and $\theta(a)=\frac{1}{m}+\mathbb{Z} \ne 0$.
In either case, the injectivity of $D$ extends $\theta$ to a homomorphism $\Theta: A \to D$. Since $\Theta(a)=\theta(a)\ne 0$, the group $D$ separates points.
[/proof]
[/step]