4

There are two statements that my professor made today that I'm hoping I can get some more clarification on.

The first is that $\mathbb{Q}/\mathbb{Z}$ is an infinite quotient group where every element has finite order.

The second is that $\mathbb{R}/ \mathbb{Q}$ is also an infinite quotient group but every element except the identity has infinite order.

I'm having trouble even imagining an infinite quotient group...I'm familiar with groups like $\mathbb{Z} / n\mathbb{Z}$, but how would you even notate these other groups? I understand that in order for an element $xH$, where $H$ is the subgroup, to have finite order, $x^n$ must be in $H$ for some $n$. So if every element of $\mathbb{Q}/\mathbb{Z}$ is finite, does that imply that every rational number is in $\mathbb{Z}$? Obviously that is not true, but I'm having trouble figuring out where I'm going wrong.

  • 2
    every element of $\mathbb Q/\mathbb Z$ has finite order because any rational number can be added to itself some finite number of times to yield an integer – J. W. Tanner Oct 14 '20 at 21:56
  • 1
    Just to mention standard useful terminology: a group in which every element has finite order is called "torsion", or "periodic". Examples of torsion groups are finite groups, and also $\mathbf{Q}/\mathbf{Z}$ mentioned here. A group in which the only element of finite order is $1$ is called "torsion-free". Of course a group which is both torsion and torsion-free has to be trivial. Example of torsion-free groups are ${0}$, $\mathbf{Z}$, and also $\mathbf{R}/\mathbf{Q}$ mentioned here. Also, $\mathbf{R}/\mathbf{Z}$ is neither torsion, nor torsion-free. – YCor Oct 16 '20 at 15:38

3 Answers3

11

First, let's work purely algebraically:

The elements of $\mathbb{Q} / \mathbb{Z}$ are the cosets $q + \mathbb{Z}$ for each rational $q \in \mathbb{Q}$. So $q_1$ and $q_2$ become equal if they differ by an integer. So we are considering $\frac{1}{2}$ and $\frac{3}{2}$, for instance to be the same. So for any $q \in [n, n+1)$ we can shift it to be in $[0,1)$ by subtracting $n \in \mathbb{Z}$. So we see that $\mathbb{Q} / \mathbb{Z}$ is exactly like $\mathbb{Q} \cap [0,1)$, where we work "mod 1". When we add $\frac{2}{3} + \frac{1}{2}$, we get $\frac{7}{6}$, which we reduce to $\frac{1}{6}$.

Geometrically, this is like "winding $\mathbb{Q}$ around a circle". It is at this point that I am legally required to show you this picture:

the universal cover of S^1 by \mathbb{R}

We are wrapping $\mathbb{Q}$ into a circle, so that two points in $\mathbb{Q}$ lie on top of each other exactly when their difference is an integer.

When you internalize these two perspectives, you should see why every element of $\mathbb{Q} / \mathbb{Z}$ has finite order. After all, if we start with $\frac{a}{b}$, and we add it to itself $b$ times, we'll be left with

$$ \underbrace{\frac{a}{b} + \frac{a}{b} + \cdots + \frac{a}{b}}_{b \text{ times}} = b \frac{a}{b} = a = 0 \text{ (mod } 1) $$


Now let's move to $\mathbb{R} / \mathbb{Q}$. This one is extra tricky, because there is no constructive way to choose an element from each coset. (There is some set-theoretic subtletly regarding the axiom of choice

Much ink has been spilled trying to understand this group (see here and here just to start), but I will give one "elementary" way to understand what is happening (though trying to visualize this group is going to be a mess no matter what you do). At the very least, this will let us resolve the question about infinite order elements.

We can write $\mathbb{R}$ as a Vector Space over $\mathbb{Q}$, and so (assuming choice) we can find a basis for $\mathbb{R}$ as a vector space. Just like every finite dimensional vector space over $\mathbb{R}$ looks like $\mathbb{R}^n$ for its dimension $n$, every infinite dimensional vector space over $\mathbb{Q}$ looks like $\mathbb{Q}^\kappa$ for its dimension $\kappa$.

So $\mathbb{R} \cong \mathbb{Q}^{\mathfrak{c}}$ (here $\mathfrak{c} = |\mathbb{R}| = \text{dim}(\mathbb{R}$) over $\mathbb{Q}$). Then when we quotient out by $\mathbb{Q}$, we can think about this as killing off one of the $\mathfrak{c}$ copies of $\mathbb{Q}$. Of course, since $\mathfrak{c}$ is infinite, we have (set theorists look the other way) "$\mathfrak{c} - 1 = \mathfrak{c}$". So we see that:

$$ \mathbb{R} / \mathbb{Q} \cong \mathbb{Q}^{\mathfrak{c}} / \mathbb{Q} \cong \mathbb{Q}^{\mathfrak{c}} \cong \mathbb{R} $$

Thus, by some nonconstructive black magic, $\mathbb{R} / \mathbb{Q} \cong \mathbb{R}$. But there is no way to describe the isomorphism more than we already have.

Like I said, though, this is enough information to answer your question: Since every nonzero element of $\mathbb{R}$ has infinite order, so too does every nonzero element of $\mathbb{R}/\mathbb{Q}$.


I hope this helps ^_^

HallaSurvivor
  • 38,115
  • 4
  • 46
  • 87
  • 2
    You make really good content :) – Shaun Oct 14 '20 at 22:28
  • 1
    genuinely, thank you <3 That's really sweet – HallaSurvivor Oct 14 '20 at 22:34
  • This is fantastic! It's exactly what I was looking for! I do have one question though. I just wrote a proof about how R has two elements of finite order: -1 and 1. What fundamental thing am I missing? Because clearly there can't be an isomorphism between two groups with differing amounts of finite elements... – regionalsky Oct 14 '20 at 22:39
  • 1
    The case of $\Bbb{R}/\Bbb{Q}$ is not "extra tricky". You don't need AC to see that every non-zero element of the additive group $\Bbb{R}/\Bbb{Q}$ has infinite order. – Rob Arthan Oct 14 '20 at 22:40
  • @regionalsky You're mistaking the group $\mathbb{R} \setminus {0}$ with multiplication and the group $\mathbb{R}$ with addition. Notice $-1 \times -1 = 1$ (the multiplicative unit), whereas $-1 + -1 \neq 0$ (the additive unit). – HallaSurvivor Oct 14 '20 at 22:40
  • 1
    Ah. Thank you!! – regionalsky Oct 14 '20 at 22:41
5

Mark's comment explains the case of $\mathbb{Q}/\mathbb{Z}$. For the case $\mathbb{R}/\mathbb{Q}$, you just have to note that every nontrivial element of $\mathbb{R}/\mathbb{Q}$ is of the form $x+\mathbb{Q}$ where $x$ is an irrational number. Thus, if there exists a positive integer $n$ such that $(x+\mathbb{Q})^{n}=nx+\mathbb{Q}=0$, then there exist $a,b\in\mathbb{Z}$, $b\neq0$ such that $nx=\frac{a}{b}$. But then we get $x=\frac{a}{nb}\in\mathbb{Q}$, a contradiction. Therefore, every nontrivial element of $\mathbb{R}/\mathbb{Q}$ has infinite order.

J. W. Tanner
  • 60,406
2

You have to remember what the operation is. The notation $x^n$ is used in general groups. However, if $\mathbb{Q}$ and $\mathbb{Z}$ then the operation is addition, not multiplication. So every element in $\mathbb{Q}/\mathbb{Z}$ having a finite order means that for every rational number $a\in\mathbb{Q}$ there is some $n\in\mathbb{N}$ such that $na=a+...+a\in\mathbb{Z}$. And this is indeed true. If $a=\frac{p}{q}$ for $p\in\mathbb{Z}$ and $q\in\mathbb{N}$ then $qa\in\mathbb{Z}$.

Mark
  • 39,605