0
$\begingroup$

The place from where I am studying Abstract Algebra recommends the following as the most general understanding of a group, the rest reducible to this.

For a set $T$, the set of all bijections $Sym(T)$ from $T$ to $T$ constitutes a group under the product operation being the composition of these bijections. Identity is then a bijection onto itself.

But when thinking of a group in terms of bijections or transformations, how do we reconcile this understanding to understand the Integers being a group under addition? What is the bijection here? Am I missing something here?

  • 0
    I think your confusion has to do with the nature of the elements in each set. The elements of $Sym(T)$ are bijections, the elements of $\mathbb Z$ are integers. – 2017-02-21
  • 1
    This recommendation is probably be a reference to [Cayley's Theorem](https://en.wikipedia.org/wiki/Cayley's_theorem). Note, however, that while it's always true, it's not always the most intuitive way to think of some specific groups. – 2017-02-21
  • 0
    I think 'd study abstract algebra somewhere else. For good explanations and a more practical motivation, read Fraleigh's Abstract Algebra. – 2017-02-21
  • 0
    @victoria I am looking for a MOOC, or any free online course for this. But I shall look Fraleigh's up. – 2017-02-21

3 Answers 3

0

The idea is not to think of a group in terms of bijections. It so happens that the elements of $\mathrm{Sym}(T)$ are bijections. The elements of $\mathbb Z$ are integers. Although there is a relationship between the two, the source you are studying from probably mentions $\mathrm{Sym}(T)$ as a good general exemple of group. In which case, the focus is on the group structure and properties, whatever the nature of its elements is.

3

That definition is either wrong (as written) or rather misleading. Here is the usual definition:

A group is a set $G$ with a law of composition $\cdot: G^2 \to G$ such that

  • Identity: there is an element $1\in G$ which satisfies $g\cdot 1 = 1\cdot g=g$ for all $g\in G$
  • Inverse: for every $g\in G$ there is an element $h\in G$ such that $g\cdot h=h\cdot g = 1$ (this element is unique and is usually written $g^{-1}$)

It is sometimes convenient to write a group as a quadruple $(G,\cdot,-^{-1},1)$.

An example of a group is the group of bijections $\mathrm{Sym}(T)$ on a set $T$, with the law of composition given by function composition. The identity is exactly the identity map $\mathrm{id}_T$ and the inverse of a bijection $f: T \to T$ is exactly the inverse map $f^{-1}: T \to T$.

There are certainly groups which are not of the above form. One way of seeing this is that for $|T|>2$ the group $\mathrm{Sym}(T)$ is not Abelian, meaning that the law of composition is not commutative. This is because given distinct elements $x,y,z\in T$, we can define bijections $f_1,f_2: T\to T$ by $$f_1(x)=y,f_1(y)=x, f_1(w)=w \quad \text{for $w\in T\setminus\{x,y\}$}$$ $$f_1(y)=z,f_1(z)=y, f_1(w)=w \quad \text{for $w\in T\setminus\{y,z\}$}$$ In other words, $f_1$ switches $x$ and $y$ and leaves everything else fixed, and $f_2$ switches $y$ and $z$ and leaves everything else fixed. Then $$(f_1\circ f_2)(y) = f_1(f_2(y))=f_1(z)=z$$ $$(f_2\circ f_1)(y) = f_2(f_1(y))=f_2(x)=x$$ are unequal, so $f_1\circ f_2 \neq f_2 \circ f_1$.

However, $\mathbb{Z}$ with addition is Abelian, so it can't be of the form $\mathrm{Sym}(T)$ for any set $T$.

What is true, however, is that every group is (isomorphic) to a subgroup of $\mathrm{Sym}(T)$ for some set $T$. This is known as Cayley's Theorem.

By subgroup of a group $(G,\cdot,-^{-1},1)$ we mean a subset $H\subset G$ for which $H$ is closed under $\cdot$ and taking inverses, and contains $1$.

Given a group $(G,\cdot,-^{-1},1)$, consider $\mathrm{Sym}(G)$. We then associated each element $g\in G$ to the function $\tau_g: G \to G$ defined by $\tau_g(h) = g\cdot h$. This is invertible, since $\tau_g \circ \tau_{g^{-1}} = \mathrm{id}_G$. More generally, $\tau_g \circ \tau_h = \tau_{g\cdot h}$. This allows us to treat $G$ like a subgroup of $\mathrm{Sym}(G)$.

  • 0
    As far as I understand, within the scope of group theory, isomorphic groups are identical. So if all groups can be reduced to a subgroup of a $Sym(T)$, wouldn't the claim I put up in the question be correct with the added caveat of the subgroup. What I meant by a general form was that in this way, all the product operations defined with different groups can be viewed as compositions of bijections, allowing a consistent picture of a group. Otherwise, I am unable to deal with a general meaning of this product operation. – 2017-02-21
  • 0
    Yes, with the added caveat of being a subgroup, it is true. While it might be a useful way for understanding abstract groups in general, viewing groups as 'groups of symmetries', personally this context is not how I see groups and can get in the way of dealing with more general structures, where the word 'group' shows up but thinking in terms of 'symmetries' isn't conducive to intuition. I just think of it as a nice binary operation. There are many binary operations, plus, times, max, composition, concatenation, etc, and groups represent the nicest of those. – 2017-02-21
2

There is a correspondence between integers and a certain class of bijections of $\mathbb{Z}$, as follows. For $n \in \mathbb{Z}$, define $f_n:\mathbb{Z} \to \mathbb{Z}$ by $f_n(k) = k+n$. Then $f_n$ is a bijection. Geometrially, $f_n$ translates $\mathbb{Z}$ to the right by $n$. Then the association $n \mapsto f_n$ is a way to think of each integer as a bijection $\mathbb{Z} \to \mathbb{Z}$.