# Limits and Colimits, Part 2 (Definitions)

/*are*limits and colimits?" As we saw then, there are

*two*main ways that mathematicians construct

*new*objects from a collection of given objects: 1) take a "sub-collection," contingent on some condition or 2) "glue" things together. The first construction is usually a limit, the second is usually a colimit. Of course, this might've left the reader wondering, "Okay... but

*what*are we taking the (co)limit

*of*?" The answer?

*A diagram*. And as we saw a couple of weeks ago, a diagram is really a functor.

We are now ready to give the formal definitions (along with more intuition). First, here's a bit of setup.

## The Setup

*constant functor*- let's also call it $X$ - from $\mathsf{I}$ to $\mathsf{C}$. This functor sends every object to $X$ and every morphism to the identity of $X$.

*Therefore*, given any functor (ahem,

*diagram*) $F:\mathsf{I}\to\mathsf{C}$, we can make sense of a natural transformation between $X$ and $F$. Such a natural transformation consists of a collection of morphisms between $X$ and the objects in the diagram $F$. Moreover, these morphisms must

*commute*with all the morphisms that appear in diagram. If the arrows point

*from*$X$

*to*the diagram $F$, then the setup is called a

*cone over $F$*, as we previously discussed here. If, on the other hand, the arrows point

*from*the diagram $F$

*to*$X$, then it's called a

*cone under $F$*(or sometimes a

*cocone*).

*object*AND a

*collection of arrows*to or from it. Now here's the punchline:

**
The limit of a diagram $F$ is a special cone over $F$. **

The colimit of $F$ is a special cone under $F$.

Let's take a look at the formal definitions. I'll give a lite version first, followed by the full version.

## Definitions (Lite Version)

**Definition (limit)**: The limit of a diagram $F$ is the "shallowest" cone over $F$.

*many*cones -- many objects with maps pointing down to the diagram $F$ (depicted as a blob) -- over $F$, but the

**limit**is the cone that is as close as possible to the diagram $F$. Perhaps this is why "limit" is a good choice of terminology. You might imagine all the cones over $F$ as cascading down to the

*limit*.

If we let gravity pull all the arrows down, then we obtain the dual notion: a colimit.

**Definition (colimit)**: The colimit of a diagram $F$ is the "shallowest" cone under $F$.

*many*cones under $F$, but the colimit is the one that's closest to the diagram. It's the shallowest.

Okay, this is all very handwavy and not very informative. To capture the mathematics behind "shallowest," we'll use a universal property. I'll comment on intuition below.

## Definition: Limit (Full Version)

*slowly*to minimize the shedding of tears.

**Definition (1)**: The **limit** of a diagram $F:\mathsf{I}\to\mathsf{C}$ the universal cone over $F$.

*Let's unwind this a bit...*

**Definition (2):**The

**limit**of a diagram $F:\mathsf{I}\to\mathsf{C}$ is an object $\text{lim }F$ in $\mathsf{C}$ together with a natural transformation $\eta:\text{lim }F\Rightarrow F$ with the following property: for any object $X$ and for any natural transformation $\alpha\colon X\Rightarrow F$, there is a unique morphism $f\colon X\to \text{lim }F$ such that $\alpha=\eta\circ f$.

*Let's unwind this a bit more...
*

**Definition (3):**The

**limit**of a diagram $F:\mathsf{I}\to\mathsf{C}$ is an object $\text{lim }F$ in $\mathsf{C}$ together with morphisms $\eta_A:\text{lim }F\to A$, for each $A$ in the diagram, satisfying $\eta_B=\phi_{AB}\circ \eta_A$ for every morphism $\phi_{AB}\colon A\to B$ in the diagram. Morever, these maps have the following property: for any object $X$ and for any collection of morphisms $\alpha_A:X\to A$ satisfying $\alpha_B=\phi_{AB}\circ \alpha_A$, there exists a unique morphism $f\colon X\to\text{lim }F$ such that $$\alpha_A=\eta_A\circ f \qquad\text{for all objects $A$ in the diagram.}$$

## Definition: Colimit (Full Version)

**Definition (1)**: The **colimit** of a diagram $F:\mathsf{I}\to\mathsf{C}$ the universal cone under $F$.

*Let's unwind this a bit...*

**Definition (2):**The

**colimit**of a diagram $F:\mathsf{I}\to\mathsf{C}$ is an object $\text{colim }F$ in $\mathsf{C}$ together with a natural transformation $\epsilon:F\Rightarrow \text{colim }F$ with the following property: for any object $X$ and for any natural transformation $\beta\colon F\Rightarrow X$, there is a unique morphism $g\colon \text{colim }F\to F$ such that $\beta= g\circ \epsilon$.

*Let's unwind this a bit more...
*

**Definition (3):**The

**colimit**of a diagram $F:\mathsf{I}\to\mathsf{C}$ is an object $\text{colim }F$ in $\mathsf{C}$ together with morphisms $\epsilon_A:A\to\text{colim }F$, for each $A$ in the diagram, satisfying $\epsilon_A=\epsilon_B\circ \phi_{AB}$ for every morphism $\phi_{AB}\colon A\to B$ in the diagram. Morever, these maps have the following property: for any object $X$ and for any collection of morphisms $\beta_A: A\to X$ satisfying $\beta_A=\beta_B\circ \phi_{AB}$, there exists a unique morphism $g\colon \text{colim }F\to X$ such that $$\beta_A= g \circ \epsilon_A\qquad\text{for all objects $A$ in the diagram.}$$

## a little intuition + a little exercise

*special role*than an object - or in our case, a cone - plays. I like that analogy, and it's exactly what's going on with limits. (Similar sentiments hold for colimits.) Let me elaborate:

Out of

*all*the cones over a diagram $F$, there is exactly

*one*that plays the role of limit, namely the pair $(\text{lim}F,\eta)$. Of course you

*might*come across another cone $(X,\alpha)$ that plays a

*very*similar role. Perhaps $\alpha$ behaves very similarly to $\eta.$ BUT -- and this is the punchline -- this behavior is no coincidence! The natural transformation $\alpha$

*"behaves"*like $\eta$

*because*it is

**built up**from $\eta$! More precisely, it has $\eta$ as a factor: $\alpha=\eta\circ f$ for some unique morphism $f$!

By way of analogy, think of the role that the number 2 plays among the integers. Out of all the integers, we might say that 2 is the quintessential candidate for "an integer which possesses the quality of 'two-ness,'" that is, of being *even*. Of course, there are *other* integers $a$ that play a similar role. In particular, if $a$ is an even integer, then it also possesses the quality of "two-ness." But this is no coinicidence! An even integer is even *because* it is **built up** from 2! More precisely, it has 2 as a factor: $a=2k$, for some unique integer $k$!

These two equations $a=2k$ and $\alpha=\eta\circ f$ are analogous. In fact, they're more than analogous....

### ** EXERCISE **

**
Show that $2$ is the limit of a particular diagram in $2\mathbb{Z}$.
**

*and*a natural transformation $\eta$, and show that $2$ and $\eta$ satisfy the universal property above.

*Hint: Despite all the fancy jargon, it's really simple!*(Next question: does this diagram have a colimit? If so, what is it?)

*very familiar*names, depending on the shape of the indexing category $\mathsf{I}$!

Until then!