3
$\begingroup$

Working inside a particular mathematical structure, I have no trouble giving rigorous definitions, nor deciding whether or not a definition is rigorous. For example, working inside $\mathbb{Z}$:

Definition. Let us say that $x \in \mathbb{Z}$ is prime iff $x$ is non-zero, $x$ does not divide $1$, and for all $y,z \in \mathbb{Z}$, we have that if $x \mid yz$, then $x \mid y$ or $x \mid z$.

So we've specified the meaning of "$x \in \mathbb{Z}$ is prime" without any serious ambiguity. However, in contexts that involve an "external" perspective on mathematical structures, I find it difficult to give truly rigorous definitions, or perhaps I just cannot tell which definitions are rigorous.

For a silly example:

Definition. Let $(G,*)$ and $(H,\diamond)$ denote groups. Then a homomorphism $(G,*) \rightarrow (H,\diamond)$ is a function $f : G \rightarrow H$ such that for all $g,g' \in G$, we have that $f(g*g') = f(g) \diamond f(g')$.

Notice that, according to the above definition, a homomorphism $(G,*) \rightarrow (H,\diamond)$ is a particular kind of function $f : G \rightarrow H$. I guess that, since $f$ is a function, or in other words, a $\mathbf{Set}$-arrow, this means that the domain and codomain of $f$ are sets, namely $\mathrm{dom}(f) = G$ and $\mathrm{cod}(f)=H.$ But (perhaps) I am imagining that the domain and codomain to be groups, so that $\mathrm{dom}(f) = (G,*)$ and $\mathrm{dom}(f) = (H,\diamond).$ So, has my definition truly captured the concept it was intended to? I don't know, but I find it slightly unsatisfying. And sure, there's more to mathematics than rigor and precision, but I like to have my cake and eat it: I like to be precise and intelligible.

Question. Does anyone know of an article purports to help you learn to think and write about mathematical structures in a way that is rigorous and precise?

$\endgroup$
4
  • 1
    $\begingroup$ In the same way that a morphism in $\mathbf{Set}$ can be regarded as a triple $(f,A,B)$ where $f\subset A\times B$ is a functional relation, I suppose you could regard a more general structure-preserving morphism as a triple $(f,\mathcal{A},\mathcal{B})$ where $\mathcal{A}$ is a structure with domain $A$, $\mathcal{B}$ a structure with domain $B$, and $f\subset A\times B$ a functional relation that preserves the structure. $\operatorname{dom}$ and $\operatorname{cod}$ just give the second and third coordinates, respectively, capturing all of the structure. $\endgroup$
    – Hayden
    Sep 28, 2014 at 16:50
  • $\begingroup$ @Hayden, that's a fair point. How would you suggest phrasing the definition, though? $\endgroup$ Sep 28, 2014 at 17:08
  • $\begingroup$ $1$ and $-1$ are prime? $\endgroup$ Sep 28, 2014 at 18:08
  • $\begingroup$ @archipelago, fixed. $\endgroup$ Sep 28, 2014 at 18:19

1 Answer 1

2
$\begingroup$

Well, as far as your actual example is concerned, your proposed definition seems quite sensible to me. You are trying to translate the following definition into natural language (more or less): given groups $(G,*)$ and $(H,\diamond)$,

$$ \mathbf{Grp}((G,*),\ (H,\diamond)):=\{((G,*),\ f,\ (H,\diamond)):\ (G,\ f,\ H)\in \mathbf{Set}(G,\ H)\ \wedge$$$$\wedge\ \forall g,g'\in G\ (f(g*g') = f(g) \diamond f(g'))\} $$

In other words, you are specifically defining the category of groups as a concrete category. Note also that, if $((G,*),\ f,\ (H,\diamond))\in \mathbf{Grp}((G,*),\ (H,\diamond))$, there is no confusion at all about what its domain and codomain are. Indeed, the domain and the codomain of an arrow in $\mathbf{Grp}$ are functions

$$ d_{\mathbf{Grp}},\ c_{\mathbf{Grp}}\colon Arr(\mathbf{Grp})\to Ob(\mathbf{Grp}), $$

and one puts $d_{\mathbf{Grp}}(((G,*),\ f,\ (H,\diamond))):= (G,*)$ and $c_{\mathbf{Grp}}(((G,*),\ f,\ (H,\diamond))):= (H,\diamond)$, whereas, with obvious notations, $d_{\mathbf{Set}}((G,\ f,\ H))= G$ and $c_{\mathbf{Set}}((G,\ f,\ H))= H$. In other words, you are simply considering different couples of functions, when describing the domain and the codomain of $f$ as a morphism of groups or as a morphism of sets.

In addition to this, I would just like to mention that, actually, one can define groups and morphisms of groups in a completely internal way, as the objects and the arrows of the category of group-objects in the category (topos) $\mathbf{Set}$ (see here). Similarly, a whole bunch of algebraic structures (basically all the fundamental ones, like monoids, rings, partial orders, excluding apparently fields though - see here again) can be described as algebraic structures internal to $\mathbf{Set}$.

To conclude, even if I can not give you references to answer your question, I would say that, quoting you and looking at my personal experience, a way "to help you learn to think and write about mathematical structures in a way that is rigorous and precise" is simply that of trying to define those structures as suitable categories and work internally to them.

EDIT: as archipelago pointed out, there was a little bias in my previous definition of $\mathbf{Grp}((G,*),\ (H,\diamond))$ as:

$$ \mathbf{Grp}((G,*),\ (H,\diamond)):=\{f\in \mathbf{Set}(G,\ H):\ \forall g,g'\in G\ (f(g*g') = f(g) \diamond f(g'))\}. $$

I have now fixed it in a way that should somehow work properly and have modified part of my original post accordingly.

$\endgroup$
2
  • 2
    $\begingroup$ Your definition is not completely rigorous. If you define a group homomorphism like that, the information of it's domain and codomain (including their group structure) is not part of the data. That really matters if one has a set theoretic map between two sets, which carry each two group structures, such that the map fulfills the defining equality of a group homomorphism for both of them. In that case the assignments $d_{Grp}$ and $c_{Grp}$ are not well defined. $\endgroup$ Sep 28, 2014 at 18:16
  • $\begingroup$ @archipelago I admit you had a point with that. I have edited the post. Still the "internal-approach" suggestion should work. $\endgroup$ Sep 28, 2014 at 21:15

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .