provenance: > Negative-space populate (2026-06-23). NATIVE-FORM specimen — Document B of a pair. This is the form-respecting companion to synthesis/topos-as-above-so-below.md: the same subject written in its own register (mathematical exposition), with no cross-links, no metaphor, no consciousness, no mysticism. The form IS the fill. do not correct toward house style / do not reconcile / do not enrich / do not enrich toward house style. title: "Topos Theory: The Mathematics" register: mathematical exposition links: [topos_as_above_so_below]

Topos Theory: The Mathematics

This is an exposition of the basic definitions of elementary and Grothendieck topos theory. The aim is precision. Where a statement is a definition it is marked as such; where it is a theorem requiring proof it is marked as such.

Limits: terminal object, products, pullbacks

A terminal object $1$ is an object such that for every object $A$ there is exactly one morphism $A \to 1$. Terminal objects are unique up to unique isomorphism.

A product of objects $A$ and $B$ is an object $A\times B$ with projections $\pi_A\colon A\times B \to A$ and $\pi_B\colon A\times B \to B$ such that for every object $X$ with morphisms $f\colon X\to A$, $g\colon X\to B$ there is a unique $\langle f,g\rangle\colon X \to A\times B$ with $\pi_A\circ\langle f,g\rangle = f$ and $\pi_B\circ\langle f,g\rangle = g$.

A pullback of $f\colon A \to C$ and $g\colon B \to C$ is an object $A\times_C B$ with morphisms $p\colon A\times_C B \to A$, $q\colon A\times_C B \to B$ satisfying $f\circ p = g\circ q$, and universal among such: for every $X$ with $u\colon X\to A$, $v\colon X\to B$ and $f\circ u = g\circ v$ there is a unique $X \to A\times_C B$ commuting with $p$ and $q$.

A category has finite limits if and only if it has a terminal object and pullbacks; equivalently, a terminal object and all binary products and equalizers. (This equivalence is a theorem.)

Exponential objects and cartesian closure

Fix an object $A$. An exponential object $B^A$ is an object equipped with an evaluation morphism $\mathrm{ev}\colon B^A \times A \to B$ such that for every object $X$ and morphism $f\colon X\times A \to B$ there is a unique $\lambda f\colon X \to B^A$ (the transpose, or currying) with $\mathrm{ev}\circ(\lambda f \times \mathrm{id}_A) = f$.

A category with finite products is cartesian closed if every pair of objects has an exponential object; equivalently, if for each $A$ the functor $(-)\times A$ has a right adjoint $(-)^A$.

The subobject classifier

A subobject of an object $A$ is an equivalence class of monomorphisms into $A$, where $m\colon S \rightarrowtail A$ and $m'\colon S' \rightarrowtail A$ are identified when each factors through the other by an isomorphism.

A subobject classifier is an object $\Omega$ together with a morphism $\top\colon 1 \to \Omega$ ("true") such that for every monomorphism $m\colon S \rightarrowtail A$ there is a unique morphism $\chi_m\colon A \to \Omega$ (the characteristic morphism) making the square

$$ \begin{array}{ccc} S & \longrightarrow & 1 \ \downarrow m & & \downarrow \top \ A & \xrightarrow{\ \chi_m\ } & \Omega \end{array} $$

a pullback. Equivalently, $\Omega$ represents the subobject functor: there is a natural isomorphism $\mathrm{Sub}(A) \cong \mathrm{Hom}(A,\Omega)$.

Elementary topos

Definition. An elementary topos is a category $\mathcal{E}$ that

has all finite limits,
is cartesian closed,
has a subobject classifier $\Omega$.

The standard example is $\mathbf{Set}$, with $1$ a one-element set, $A\times B$ the cartesian product, $B^A$ the set of functions $A\to B$, and $\Omega = {0,1}$ with $\top$ picking out $1$; here $\chi_m$ is the indicator function of the image of $m$. For any small category $\mathcal{C}$ the presheaf category $[\mathcal{C}^{\mathrm{op}}, \mathbf{Set}]$ is a topos.

The internal Heyting algebra and intuitionistic logic

Theorem. In any elementary topos the subobject classifier $\Omega$ carries the structure of an internal Heyting algebra: there are morphisms $\wedge, \vee, \Rightarrow \colon \Omega\times\Omega \to \Omega$ and elements $\top, \bot \colon 1 \to \Omega$ making $\Omega$ a Heyting algebra object. Consequently, for each object $A$ the poset of subobjects $\mathrm{Sub}(A)$ is a Heyting algebra, and these structures are stable under pullback.

A Heyting algebra is a bounded lattice with a relative pseudo-complement (the operation $\Rightarrow$ right adjoint to $\wedge$); it need not satisfy the law of excluded middle $a\vee\neg a = \top$, where $\neg a := (a\Rightarrow\bot)$. The internal logic of a topos is therefore intuitionistic in general. A topos is Boolean when $\Omega$ is an internal Boolean algebra, equivalently when excluded middle holds internally; $\mathbf{Set}$ is Boolean.

Sites, sheaves, Grothendieck toposes

A sieve on an object $C$ of a category $\mathcal{C}$ is a collection $S$ of morphisms with codomain $C$ that is closed under precomposition: if $(f\colon D \to C)\in S$ and $g\colon E \to D$ then $f\circ g \in S$.

A Grothendieck topology $J$ assigns to each object $C$ a set $J(C)$ of covering sieves on $C$ satisfying:

(maximality) the maximal sieve of all morphisms into $C$ is in $J(C)$;
(stability) if $S\in J(C)$ and $h\colon D\to C$, then the pullback sieve $h^{*}S = {,g \mid h\circ g \in S,}$ is in $J(D)$;
(transitivity) if $S\in J(C)$ and $R$ is a sieve on $C$ such that $h^{*}R \in J(D)$ for every $(h\colon D\to C)\in S$, then $R\in J(C)$.

A site is a category equipped with a Grothendieck topology. A presheaf $F\colon \mathcal{C}^{\mathrm{op}}\to\mathbf{Set}$ is a sheaf for $J$ if for every covering sieve every compatible family of sections has a unique amalgamation. The category $\mathrm{Sh}(\mathcal{C},J)$ of sheaves on a (small) site is a Grothendieck topos.

Theorem. Every Grothendieck topos is an elementary topos, and is moreover cocomplete with a small generating set. Equivalently (Giraud's theorem), a category is a Grothendieck topos iff it is the category of sheaves on some small site.

Geometric morphisms

Definition. A geometric morphism $f\colon \mathcal{F} \to \mathcal{E}$ between toposes is a pair of adjoint functors $$f^{} \dashv f_{}, \qquad f^{}\colon \mathcal{E}\to\mathcal{F}, \quad f_{}\colon \mathcal{F}\to\mathcal{E},$$ in which the left adjoint $f^{}$ (the inverse image) preserves finite limits, i.e. is left exact. The right adjoint $f_{}$ is the direct image.

Geometric morphisms are the morphisms of the (2-)category of toposes; they generalize continuous maps of topological spaces, recovering them via the assignment $X \mapsto \mathrm{Sh}(X)$.