I am reading about category theory. So a category has a collection of objects and a collection of morphisms. Denote the set of morphisms from $A$ to $B$ by $Mor(A,B)$ so $f:A\rightarrow B$ has $f \in Mor(A,B)$. I haven't found it stated anywhere but do we take $Mor(A,B)$ to always be a non-empty set?

I wonder because I wanted to try to understand the small categories like $\textbf{2}$.

I was wondering if I take the collection of objects to be two vertices $a$ and $b$ and the morphisms to just be the identity morphism for $a$ and the identity morphism for $b$ so $Mor(a,b)=Mor(b,a)=\varnothing$ and $Mor(a,a)=\{1_a\}$ and $Mor(b,b)=\{1_b\}$, if this would form a category or if I would need to throw in a morphism $a\rightarrow b$ to form the category.

  • 3
    $\begingroup$ Category theorists have no animosity toward the empty set. In particular, the set of morphisms between two objects may be empty. $\endgroup$ – Zhen Lin Jan 8 '14 at 5:42
  • 2
    $\begingroup$ As you will soon discover (see @Aaron answer below, for example), "the" category $\mathbf 2$ has different meanings for different authors. You mean the category with no arrows between different objects while, for example, MacLane means the category with 2 objects and exactly one arrow joining the first to the second object. $\endgroup$ – magma Jan 8 '14 at 9:29

Every set gives rise to a category. In particular, let $S$ be a set. Then we may form a a category which we may call $disc(S)$ which is called the discrete category of $S$. The objects of $S$ are the elements, and we set $$Mor(s,s')=\emptyset\mbox{ if } s\neq s',$$ and $$Mor(s,s)=\{Id_s\}.$$ A category satisfying the above is called a discrete category. Likewise given any (small) discrete category, we may recover the underlying set by taking the set of objects.

Moreover, their is a one to one correspondence between the functors between two discrete categories and the set of functions between the two underlying sets.


A category need not have any morphisms other than the identities. Such a category is called discrete.


There are two different categories sometimes known as $\mathbb{2}$.

The first option you describe — with only identity morphisms, so $\newcommand{Hom}{\mathrm{Hom}}\Hom(a,b) = \Hom(b,a) = \phi$ — would be less ambiguously called the discrete category on the set 2. Generally, one can construct the discrete category on any set $X$, by taking $X$ as the objects, and just identities as the morphisms.

The second option you describe — with $\Hom(b,a) = \phi$ but with a single morphism $a \to b$ thrown in — is less ambiguously known as the arrow category, the walking arrow, or the poset category on $\{0 < 1\}$. Generally, any poset or pre-order gives a category, with objects the elements of the pre-order, and with $\Hom(a,b) \cong 1$ if $a \leq b$ and $\Hom(a,b) = 0$ otherwise. Indeed, pre-orders can be seen in this way as a class of categories: categories such that each $\Hom(a,b)$ contains at most one arrow.

In any case: both of the walking arrow and the discrete category on 2 are perfectly good categories, and indeed arise quite often in practice.


While there is no problem with a $\operatorname{Hom}(A,B)=\emptyset$, when $A\neq B$, there are non-identity morphisms in the category $\mathbf 2$. Assuming we are using the same notation, we have three objects, $[0], [1], [2]$, and $\operatorname{Hom}_{\mathbf 2}([i],[j])$ is empty if $i>j$ and has exactly one morphism otherwise (which is the identity when $i=j$).

More generally, any partially ordered set gives rise to a category, and this gives a particularly nice collection of small categories to study.

  • $\begingroup$ your category $\mathbf 2$ is different from the category mentioned by the OP. The category described by the OP is simply the discrete category with 2 objects, while yours is the linear order with three objects (corresponding to the ordinal number 3). $\endgroup$ – magma Jan 8 '14 at 9:21
  • $\begingroup$ @magma When I was reading the question, somehow I didn't make the connection between the category $\mathbf 2$ and the final paragraph. My choice of naming convention here comes from looking at the category as categorical representation of the geometric $2$-simplex. Usually, when I see discrete categories, they are explicitly labeled as such. $\endgroup$ – Aaron Jan 8 '14 at 20:11

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.