Would the category of directed sets be better behaved with the empty set included, or excluded

definitionorder-theory

In a topology book of mine, a directed set is defined as a nonempty set $D$ equipped with a relation $R$ that is transitive, reflexive, and for all elements $x$ and $y$ of $D$, there exists a $z$ in $D$ such that both $xRz$ and $yRz$.

Would the category of directed sets be better behaved with the empty set included, or excluded?

The empty set equipped with the empty relation satisfies all three conditions of the definitions, vacuously. I think this is a similar scenario to when some books require a magma to be nonempty, or a metric space to be nonempty, or a topological space to be nonempty. However, I believe allowing the empty set in those categories of things gives those same categories better properties, such as having initial and/or terminal objects.

Best Answer

No, a directed set cannot be empty.

This situation is not like the other ones you name. The point of this definition is that a directed set has a direction (that's why it's called that); it is "going somewhere." For example $(\mathbb{N}, \le)$ as a directed set is "going towards $\infty$." In topology directed sets are used to define nets, which are functions from a directed set to a topological space $X$, and these eneralize sequences (which are nets of shape $(\mathbb{N}, \le)$). They have a notion of convergence which generalizes convergent sequences, which relies on this idea that a directed set is "going somewhere."

The empty set is not going anywhere! It does not have a direction.

Related Question