4.4.8 Isbell Duality for Sets

Let $X$ be a set.

The Isbell function of $X$ is the map

\[ \mathsf{I}\colon \mathcal{P}(X)\to \mathsf{Sets}(X,\mathcal{P}(X)) \]

defined by

\[ \mathsf{I}(U)\mathrel {\smash {\overset {\mathclap {\scriptscriptstyle \text{def}}}=}}[\mspace {-3mu}[x\mapsto [U,\left\{ x\right\} ]_{X}]\mspace {-3mu}] \]

for each $U\in \mathcal{P}(X)$.

Recall from Remark 4.4.1.1.2 that we may view the powerset $\mathcal{P}(X)$ of a set $X$ as the decategorification of the category of presheaves $\mathsf{PSh}(\mathcal{C})$ of a category $\mathcal{C}$. Building upon this analogy, we want to mimic the definition of the Isbell $\mathsf{Spec}$ functor, which is given on objects by

\[ \mathsf{Spec}(\mathcal{F})\mathrel {\smash {\overset {\mathclap {\scriptscriptstyle \text{def}}}=}}\operatorname {\mathrm{Nat}}(\mathcal{F},h_{(-)}) \]

for each $\mathcal{F}\in \operatorname {\mathrm{Obj}}(\mathsf{PSh}(\mathcal{C}))$. To this end, we could define

\[ \mathsf{I}(U)\mathrel {\smash {\overset {\mathclap {\scriptscriptstyle \text{def}}}=}}[U,\chi _{(-)}]_{X}, \]

replacing:

  • The Yoneda embedding $X\mapsto h_{X}$ of $\mathcal{C}$ into $\mathsf{PSh}(\mathcal{C})$ with the characteristic embedding $x\mapsto \chi _{x}$ of $X$ into $\mathcal{P}(X)$ of Definition 4.5.4.1.1.

  • The internal Hom $\operatorname {\mathrm{Nat}}$ of $\mathsf{PSh}(\mathcal{C})$ with the internal Hom $[-,-]_{X}$ of $\mathcal{P}(X)$ of Proposition 4.4.7.1.1.

However, since $[U,\chi _{x}]_{X}$ is a subset of $U$ instead of a truth value, we get a function

\[ \mathsf{I}\colon \mathcal{P}(X)\to \mathsf{Sets}(X,\mathcal{P}(X)) \]

instead of a function

\[ \mathsf{I}\colon \mathcal{P}(X)\to \mathcal{P}(X). \]

This makes some of the properties involving $\mathsf{I}$ a bit more cumbersome to state, although we still have an analogue of Isbell duality in that $\mathsf{I}_{!}\circ \mathsf{I}$ evaluates to $\operatorname {\mathrm{id}}_{\mathcal{P}(X)}$ in the sense of Proposition 4.4.8.1.3.

The diagram

commutes, i.e. we have

\[ \mathsf{I}_{!}(\mathsf{I}(U))=[\mspace {-3mu}[x\mapsto [\mspace {-3mu}[y\mapsto U]\mspace {-3mu}]]\mspace {-3mu}] \]

for each $U\in \mathcal{P}(X)$.

We have

\begin{align*} \mathsf{I}_{!}(\mathsf{I}(U)) & \mathrel {\smash {\overset {\mathclap {\scriptscriptstyle \text{def}}}=}}\mathsf{I}_{!}([\mspace {-3mu}[x\mapsto U^{\textsf{c}}\cup \left\{ x\right\} ]\mspace {-3mu}])\\ & \mathrel {\smash {\overset {\mathclap {\scriptscriptstyle \text{def}}}=}}[\mspace {-3mu}[x\mapsto \mathsf{I}(U^{\textsf{c}}\cup \left\{ x\right\} )]\mspace {-3mu}]\\ & \mathrel {\smash {\overset {\mathclap {\scriptscriptstyle \text{def}}}=}}[\mspace {-3mu}[x\mapsto [\mspace {-3mu}[y\mapsto (U^{\textsf{c}}\cup \left\{ x\right\} )^{\textsf{c}}\cup \left\{ x\right\} ]\mspace {-3mu}]]\mspace {-3mu}]\\ & = [\mspace {-3mu}[x\mapsto [\mspace {-3mu}[y\mapsto (U\cap (X\setminus \left\{ x\right\} ))\cup \left\{ x\right\} ]\mspace {-3mu}]]\mspace {-3mu}]\\ & = [\mspace {-3mu}[x\mapsto [\mspace {-3mu}[y\mapsto (U\setminus \left\{ x\right\} )\cup \left\{ x\right\} ]\mspace {-3mu}]]\mspace {-3mu}]\\ & = [\mspace {-3mu}[x\mapsto [\mspace {-3mu}[y\mapsto U]\mspace {-3mu}]]\mspace {-3mu}], \end{align*}

where we have used Item 2 of Proposition 4.3.11.1.2 for the fourth equality above.


Noticed something off, or have any comments? Feel free to reach out!


You can also use the contact form below: