Let's call this category $\mathbf{Pt}$ (if there's a standard name for it then I don't know it). I have been stuck for a few days on this problem. There is a terminal object $(\{\bullet\},\bullet)$ and products can be taken to be $(A,a)\times (B,b)=(A\times B, (a,b))$, so one might expect $\mathbf{Pt}$ to be cartesian closed. However, I strongly suspect that it's not. Intuitively, it seems to me that the functor $U:\mathbf{Pt}\to\mathbf{Set}$ sending objects $(A,a)$ to the set $A$ and sending arrows to themselves (i.e. the functor which "forgets" about the specified points) might preserve exponentials. I've solved the problem if this were true, but I don't know how to go about showing it (I don't even know if it's actually correct, honestly).

Any hints would be great.

  • 2
    $\begingroup$ It’s not. You can verify that products don’t distribute over coproducts. $\endgroup$ – Qiaochu Yuan Jul 18 '18 at 18:23
  • $\begingroup$ @QiaochuYuan Interesting. I never thought to look at it like that. I'll take a look at it. Thank you. $\endgroup$ – D. Brogan Jul 18 '18 at 18:25
  • $\begingroup$ Another possible approach might start with: the forgetful functor $U$ is represented by $2$, the pointed set $(\{ 0, 1 \}, 0)$. So, if the category were cartesian closed, the exponential would have to satisfy $U([A, B]) \simeq \operatorname{Hom}_{\mathbf{Pt}}(2, [A, B]) \simeq \operatorname{Hom}_{\mathbf{Pt}}(2 \times A, B)$. An element of this would be equivalent to two functions $A \to B$, one preserving base points and the other not necessarily preserving base points. $\endgroup$ – Daniel Schepler Jul 18 '18 at 19:34
  • $\begingroup$ A standard notation is $1/Set$. $\endgroup$ – Berci Jul 18 '18 at 21:37

One way to define a cartesian closed category is that it's a category with finite products where every product functor $(-) \times X$ has a right adjoint, namely the exponential functor $(-)^X$. A necessary condition for this to be the case is that $(-) \times X$ preserves colimits; in other words, that products distribute over colimits, and in particular coproducts, when they exist.

But products in pointed sets don't distribute over coproducts. For finite pointed sets $A, B, C$ the coproduct satisfies $|A \sqcup B| = |A| + |B| - 1$ and you can use this to show that $(A \sqcup B) \times C$ and $(A \times C) \sqcup (B \times C)$ have different cardinalities in general.

Alternatively, you can use the fact that pointed sets have a zero object. Here's a fun exercise: the only cartesian closed category with a zero object is the terminal category.

What is true is that pointed sets have a closed monoidal structure. The closed structure is given by defining the pointed hom-object $[A, B]$ to consist of all functions $A \to B$, with basepoint given by the function which has constant value the basepoint of $B$. The monoidal structure is given by defining $A \otimes B$ to be the smash product, which is the quotient of $A \times B$ given by identifying all pairs $(a, b)$ such that either $a$ or $b$ is the basepoint down to a single basepoint.

  • $\begingroup$ Whoops, thanks. $\endgroup$ – Qiaochu Yuan Jul 18 '18 at 19:07

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.