Work over an ambient commutative ring with unity. Given a module $X$, write $TX$ for its submodule of torsion elements. Suppose we want to find the "submodule" of torsion-free elements of $X$. So basically, we're interested in $(X \setminus TX) \cup \{0\}$. If I understand correctly, this basically doesn't work, so instead we deal with the quotient module $X/TX$. Actually, I don't quite see why it doesn't work:


Q0. What are some examples of $\mathbb{Z}$-modules $X$ such that $(X \setminus TX) \cup \{0\}$ isn't a submodule?

Q1. I'm also interested in examples where $\mathbb{Z}$ is replaced by other commutative rings.

Q2. A truly complete answer would provide a characterization of those rings $R$ such that for all $R$-modules $X$, the set $(X \setminus TX) \cup \{0\}$ is an $R$-submodule. Obviously, every field has this property.


Consider $X = \mathbb{Z} \times \mathbb{Z}/2\mathbb{Z}$. Then both $(1,1)$ and $(-1,0)$ are in $(X \setminus TX) \cup \{0\}$, but their sum isn't. In fact let $R$ be any ring that has a module $M$ containing a nonzero torsion element $m$. Then in $R \times M$, both $(1, m)$ and $(-1,0)$ are torsion-free, but their sum isn't. In other words, there is no example other than rings for which no module has torsion, AKA fields (if $R$ is not a field let $I$ be a nontrivial ideal, then all of $R/I$ is torsion)

  • $\begingroup$ Its weird, the first thing I tried was $\mathbb{Z} \times \mathbb{Z}/2\mathbb{Z}$ but ended up concluding this wasn't an example. You're obviously right though. I think I'm visualizing it wrong. $\endgroup$ – goblin Jul 12 '16 at 12:53
  • $\begingroup$ @goblin Basically the problem is that torsionfree elements can differ by a torsion element. $\endgroup$ – Najib Idrissi Jul 12 '16 at 12:53
  • $\begingroup$ Thanks. I'll accept your answer later when I get home. The software won't let me do it right now. I think I was visualizing $(X \setminus TX)$ as the orthogonal complement of $TX$, which is why I was getting confused. $\endgroup$ – goblin Jul 12 '16 at 12:55

Let's ask when, given a module $M$ with a proper nonzero submodule $L$, the set $(M\setminus L)\cup\{0\}$ is a submodule.

Let $x\in M\setminus L$ and let $y\in L$, $y\ne0$. Then also $x+y\in M\setminus L$ and $-x\in M\setminus L$; on the other hand $$ (x+y)+(-x)=y\notin (M\setminus L)\cup\{0\} $$ Therefore the set is not even an additive subgroup.

So your set is a submodule only in the trivial cases when $TX=0$ or $TX=X$.


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.