MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Please excuse me if the question is too vague or uninteresting.

Let $\mathcal{C}$ be a category and $A$ an object in $\mathcal{C}$. Motivated by the equivalence of Dedekind-finiteness and finiteness for sets (assuming Choice), let us define that an object $A$ is $\mathcal{C}$-infinite if there is an object $A'$ in $\mathcal{C}$ corresponding to a proper subobject of $A$ and an isomorphism $f: A \to A'$ in $\mathcal{C}$.

In the category Set, we have the property that ($B$ is Set-infinite and $B \subseteq A) \implies A$ is Set-infinite. The analogous property does not hold (with this definition of infinity) over arbitrary categories. For instance, the additive group $\mathbb{Q}$ has $\mathbb{Z}$ as a proper subgroup, is finite in the category Grp (since any homomorphism $f: \mathbb{Q} \to \mathbb{Q}$ must actually be an isomorphism), but $\mathbb{Z}$ is Grp-infinite by $g:\mathbb{Z} \to 2\mathbb{Z}$ with $z \mapsto 2z$.

In which categories would a version of this transitivity property hold? What about other properties depending on a notion of infinity? Or determining whether a given object in a category is infinite or not?

A result somewhat related to this notion is that a Banach space isomorphic to all its infinite-dimensional closed subspaces is isomorphic to a separable Hilbert space. Any such Banach space is obviously Ban-infinite.

I don't know much Category Theory or Logic, but this has been troubling me for a while (I'm an undergraduate student). Thanks in advance for your contributions.

share|cite|improve this question
Is the notion you give equivalent to the notion "There is a monomorphism $A\to A$ which is not epic", and if not why is the one you specify preferable to the one I suggest here? – Asaf Karagila Aug 18 '14 at 6:07
Even in very well-behaved categories (like toposes) there are many (inequivalent) ways to say when objects are finite. See e.g. and for definitions, interrelations, and properties of these various definitions. – Chris Heunen Aug 18 '14 at 9:42
Alternatively, in monoidal categories like that of Hilbert spaces there is a way to characterize finiteness (finite dimensionality) via reflexivity, see the finite-dimensional Hilbert spaces are precisely the dualizable objects. (Not precisely what you're asking, but might be worth pointing out.) – Chris Heunen Aug 18 '14 at 9:46
@Chris: This reflects nicely from axiom of choice related definition of finiteness. John Truss had a nice paper about seven definitions of finiteness, and the relations between them in the absence of choice; and there are other definitions as well. – Asaf Karagila Aug 18 '14 at 10:07
Hm, yes, I was pointed out there is this notion of being a ``cohopfian" object that would correspond to $\mathcal{C}$-finiteness... it seems this has been investigated in categories of rings, modules, groups and topological spaces. – José Siqueira Aug 19 '14 at 10:33

This is an enlarged comment. See section $8.2$ of $[1]$.

Transitivity Property: Let $B$ be $\mathscr{C}$-infinite, and let $B\hookrightarrow A$. Then $A$ is $\mathscr{C}$-infinite. This is not satisfied for all $\mathscr{C}$; e.g., let $\mathscr{C}$ be the category of presheaves on $(\mathbb{N},\leq)$ (Counterexample $8.2.4$, $[1]$).

The following theorem gives a sufficient condition (there might be many more) for the transitivity property to hold.

Theorem 1 (Proposition $8.2.3$, $[1]$): Let $A$ be an object of a topos $\mathscr{C}$ such that there is a complemented subobject $B$ of $A$ that is $\mathscr{C}$-infinite. Then $A$ is itself $\mathscr{C}$-infinite.

Proof (taken from $[1]$): Let $C\hookrightarrow B$ be a subobject of $B$ not containing the global element $b:\mathbf{1}\to B$ and an isomorphism $C\to B$. As $\models b\in B$, it follows that $\models\neg(b\in \overline{B})$; hence $\models\neg(b\in S\cup\overline{B})$. Thus $S\coprod\overline{B}$ is a proper subobject of $A$. There is an isomorphism $A\to S\coprod \overline{B}$. Hence $A$ is $\mathscr{C}$-infinite. $\square$


$[1]$ Francis Borceux, Handbook of Categorical Algebra: Volume $3$, Sheaf Theory. $522$ pp. Cambridge University Press. Published $1994$.

share|cite|improve this answer
I think the OP is asking about what conditions on a category we can impose to force the transitivity property to hold, not what conditions on objects we can impose. – Sam Hopkins Aug 18 '14 at 2:27
@SamHopkins, if you impose a condition on all objects of a category, then it'll become a condition imposed on the category. – Michal R. Przybylek Aug 18 '14 at 12:28

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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