Synthetic topos theory

[0015] Sheaves on \(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)\)

The category \(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)\) is type-theoretically not nice because it is not locally cartesian closed. We consider embedding it into a logos (in a larger universe).

[0016] Definition

Let \(U\) be a universe and let \(V\) be a universe strictly greater than \(U\). We say a presheaf \(A:{(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U))}^{\mathord {\textnormal {\textsf {op}}}}\rightarrow V\) is a sheaf if it takes étale colimits in \(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)\) to limits in \(V\).

[0019] Definition

Let \(U\) be a universe and let \(V\) be a universe strictly greater than \(U\). We define \(\mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)\) to be the full subcategory of \(\mathord {\textnormal {\textsf {Fun}}}({(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U))}^{\mathord {\textnormal {\textsf {op}}}},V)\) spanned by the sheaves in the sense of [0016].

[001A] Exercise

Let \(U\) be a universe and let \(V\) be a universe strictly greater than \(U\). Then \(\mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)\) is a \(V\)-logos.

[001B] Proposition

Let \(U\) be a universe and let \(V\) be a universe strictly greater than \(U\). Then the Yoneda embedding \(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)\rightarrow \mathord {\textnormal {\textsf {Fun}}}({(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U))}^{\mathord {\textnormal {\textsf {op}}}},V)\) factors through \(\mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)\).

Proof

This is because representable presheaves preserve arbitrary limits.

[001C] Proposition

Let \(U\) be a universe and let \(V\) be a universe strictly greater than \(U\). Then the Yoneda embedding \(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)\rightarrow \mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)\) preserves finite limits and \(U\)-small products.

Proof

This is because the Yoneda embedding preserves arbitrary limits.

[001D] Proposition

Let \(U\) be a universe and let \(V\) be a universe strictly greater than \(U\). Then the Yoneda embedding \(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)\rightarrow \mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)\) takes étale colimits to colimits.

Proof

This follows from the Yoneda lemma and the definition of sheaves on \(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)\).

[002Z] Exercise

Let \(U\) be a universe and let \(V\) be a universe strictly greater than \(U\). Then the Yoneda embedding exhibits \(\mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)\) as the completion of \(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)\) under \(V\)-small colimits preserving étale colimits.

The object \(\mathbb {A}\) in \(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)\) classifies \(U\)-small sheaves by [000Z]. In \(\mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)\), we can construct objects that classify large sheaves.

[001E] Definition

Let \(U\) be a universe, let \(V\) be a universe greater than or equal to \(U\), and let \(W\) be a universe strictly greater than \(V\). The forgetful functor \(\mathord {\textnormal {\textsf {LexCocomp}}}(V,W)\rightarrow \mathord {\textnormal {\textsf {LexCocomp}}}(U,W)\) has a left adjoint. This left adjoint takes \(U\)-compact objects in \(\mathord {\textnormal {\textsf {LexCocomp}}}(U,W)\) to \(V\)-compact objects in \(\mathord {\textnormal {\textsf {LexCocomp}}}(V,W)\) and thus induces a functor \(C\mapsto \mathop {\Uparrow ^{V}}C:\mathord {\textnormal {\textsf {Logos}}}(U)\rightarrow \mathord {\textnormal {\textsf {Logos}}}(V)\). For a \(U\)-logos \(C\), the \(V\)-logos \(\mathop {\Uparrow ^{V}}C\) is called the \(V\)-enlargement of \(C\). Notice that the choice of \(W\) does not matter.

[001F] Exercise

Let \(U\) be a universe and let \(V\) be a universe greater than or equal to \(U\). For any \(U\)-logos \(C\), the unit \(C\rightarrow \mathop {\Uparrow ^{V}}C\), which is a morphism in \(\mathord {\textnormal {\textsf {LexCocomp}}}(U,W)\) for some universe \(W\) strictly greater than \(V\), is fully faithful. We thus regard \(C\) as a full subcategory of \(\mathop {\Uparrow ^{V}}C\).

[001G] Exercise

Let \(U\) be a universe and let \(V\) be a universe greater than or equal to \(U\). Then the functor \(C\mapsto \mathop {\Uparrow ^{V}}C:\mathord {\textnormal {\textsf {Logos}}}(U)\rightarrow \mathord {\textnormal {\textsf {Logos}}}(V)\) preserves étale morphisms. More precisely, for any \(U\)-logos \(C\) and any object \(A:C\), the embedding \(C/A\rightarrow \mathop {\Uparrow ^{V}}C/A\) induces an equivalence \(\mathop {\Uparrow ^{V}}(C/A)\simeq \mathop {\Uparrow ^{V}}C/A\).

[001H] Proposition

Let \(U\) be a universe and let \(V\) be a universe greater than or equal to \(U\). Then the functor \(C\mapsto \mathop {\Uparrow ^{V}}C:\mathord {\textnormal {\textsf {Logos}}}(U)\rightarrow \mathord {\textnormal {\textsf {Logos}}}(V)\) takes étale limits in \(\mathord {\textnormal {\textsf {Logos}}}(U)\) to étale limits in \(\mathord {\textnormal {\textsf {Logos}}}(V)\).

Proof

Let \(C\) be a \(U\)-logos, let \(I\) be a \(U\)-small category, and let \(A:I\rightarrow C\) be a functor. Let \(B:C\) denote the colimit of \(A\). By [0011], the étale limit of the diagram induced by \(A\) is \(C/B\). Then apply [001G].

[001I] Definition

Let \(U\) be a universe, let \(V\) be a universe greater than or equal to \(U\), and let \(W\) be a universe strictly greater than \(V\). We define presheaves

\(\mathbb {A}^{(V)},\mathbb {A}_{\bullet }^{(V)}:{(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U))}^{\mathord {\textnormal {\textsf {op}}}}\rightarrow W\)
by
\(\mathbb {A}^{(V)}(X)=\mathord {\textnormal {\textsf {Obj}}}(\mathop {\Uparrow ^{V}}(\mathord {\textnormal {\textsf {Sh}}}(X)))\) and \(\mathbb {A}_{\bullet }^{(V)}(X)=\mathord {\textnormal {\textsf {Obj}}}(\mathop {\Uparrow ^{V}}(\mathord {\textnormal {\textsf {Sh}}}(X))\backslash \mathord {\textnormal {\textsf {1}}})\).
We also define a morphism \(\mathord {\textnormal {\textsf {p}}}^{(V)}:\mathbb {A}_{\bullet }^{(V)}\rightarrow \mathbb {A}^{(V)}\) by the codomain projection.

[001J] Proposition

Let \(U\) be a universe, let \(V\) be a universe greater than or equal to \(U\), and let \(W\) be a universe strictly greater than \(W\). Then the presheaves \(\mathbb {A}^{(V)},\mathbb {A}_{\bullet }^{(V)}:{(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U))}^{\mathord {\textnormal {\textsf {op}}}}\rightarrow W\) are sheaves.

Proof

\(\mathbb {A}^{(V)}\) is a sheaf by [001H]. \(\mathbb {A}_{\bullet }^{(V)}\) is a sheaf because coslicing commutes with limits.

[001K] Exercise

Let \(U\) be a universe, let \(V\) be a universe greater than or equal to \(U\), and let \(W\) be a universe strictly greater than \(V\). Then the morphism \(\mathord {\textnormal {\textsf {p}}}^{(V)}:\mathbb {A}_{\bullet }^{(V)}\rightarrow \mathbb {A}^{(V)}\) in \(\mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),W)\) is univalent.

We can restore part of the \(2\)-categorical structure of \(\mathord {\textnormal {\textsf {Topos}}}(U)\).

[0053] Definition

Let \(U\) be a universe. We define a functor \(\mathord {\textnormal {\textsf {Q}}}:\mathord {\textnormal {\textsf {Cat}}}(U)\rightarrow \mathord {\textnormal {\textsf {Topos}}}(U)\) by \(\mathord {\textnormal {\textsf {Q}}}(C)=C\bullet \mathord {\textnormal {\textsf {1}}}\).

[0054] Exercise

Let \(U\) be a universe.

  • -\((\mathord {\textnormal {\textsf {Q}}}(\mathord {\textnormal {\textsf {s}}}^{1}),\mathord {\textnormal {\textsf {Q}}}(\mathord {\textnormal {\textsf {s}}}^{0})):\mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack 2\rbrack )\rightarrow \mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack 1\rbrack )\times \mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack 1\rbrack )\) determines a partial order in \(\mathord {\textnormal {\textsf {Topos}}}(U)\). We refer to this partially ordered object in \(\mathord {\textnormal {\textsf {Topos}}}(U)\) as \(\mathbb {I}\).
  • -\(\mathord {\textnormal {\textsf {Q}}}(\mathord {\textnormal {\textsf {d}}}^{1}):\mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack 0\rbrack )\rightarrow \mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack 1\rbrack )\) determines the least element of \(\mathbb {I}\).
  • -\(\mathord {\textnormal {\textsf {Q}}}(\mathord {\textnormal {\textsf {d}}}^{0}):\mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack 0\rbrack )\rightarrow \mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack 1\rbrack )\) determines the greatest element of \(\mathbb {I}\).
  • -\(\mathord {\textnormal {\textsf {0}}}\simeq \mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack 0\rbrack )\mathbin {{}_{\mathord {\textnormal {\textsf {Q}}}(\mathord {\textnormal {\textsf {d}}}^{1})}\mathord {\times _{\mathord {\textnormal {\textsf {Q}}}(\mathord {\textnormal {\textsf {d}}}^{0})}}}\mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack 0\rbrack )\)
  • -Let \(n\ge 3\) be an integer. Then the morphism
    \((\mathord {\textnormal {\textsf {Q}}}(\mathord {\textnormal {\textsf {s}}}^{n-1}),{(\mathord {\textnormal {\textsf {Q}}}(\mathord {\textnormal {\textsf {s}}}^{0}))}^{n-2}):\mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack n\rbrack )\rightarrow \mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack n-1\rbrack )\mathbin {{}_{{(\mathord {\textnormal {\textsf {Q}}}(\mathord {\textnormal {\textsf {s}}}^{0}))}^{n-2}}\mathord {\times _{\mathord {\textnormal {\textsf {Q}}}(\mathord {\textnormal {\textsf {s}}}^{1})}}}\mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack 2\rbrack )\)
    is an equivalence. (This means that \(\mathord {\textnormal {\textsf {Q}}}(\Delta \lbrack n\rbrack )\) is the object of increasing sequences in \(\mathbb {I}\) of length \(n\)).

One can also show that \(\mathbb {I}\) is totally ordered in \(\mathord {\textnormal {\textsf {Topos}}}(U)\), but that property would not be preserved by the embedding \(\mathord {\textnormal {\textsf {Topos}}}(U)\rightarrow \mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)\).

[0055] Definition

Let \(U\) be a universe, let \(C\) be a \(U\)-logos, let \(I\) be a partially ordered object in \(C\), and let \(n\) be a natural number. We define \(\Delta _{I}\lbrack n\rbrack \) to be the object in \(C\) that classifies (non-strictly) increasing sequences in \(I\) of length \(n\). When \(I\) has least and greatest elements, face maps and degeneracy maps between \(\Delta _{I}\lbrack n\rbrack \)'s can also be defined.

[0032] Definition

Let \(U\) be a universe, let \(C\) be a \(U\)-logos, and let \(I\) be a partially ordered object in \(C\) with least and greatest elements. We say an object in \(C\) is \(I\)-categorically fibrant if it is internally local for the following morphisms.

  • -\((\mathord {\textnormal {\textsf {d}}}^{2},\mathord {\textnormal {\textsf {d}}}^{0}):\Delta _{I}\lbrack 1\rbrack \mathbin {{}_{\mathord {\textnormal {\textsf {d}}}^{0}}\mathord {+_{\mathord {\textnormal {\textsf {d}}}^{1}}}}\Delta _{I}\lbrack 1\rbrack \rightarrow \Delta _{I}\lbrack 2\rbrack \)
  • -\(\Delta _{I}\lbrack 3\rbrack \mathbin {{}_{(\mathord {\textnormal {\textsf {d}}}^{3}\circ \mathord {\textnormal {\textsf {d}}}^{1},\mathord {\textnormal {\textsf {d}}}^{0}\circ \mathord {\textnormal {\textsf {d}}}^{1})}\mathord {+_{\mathord {\textnormal {\textsf {s}}}^{0}+\mathord {\textnormal {\textsf {s}}}^{0}}}}(\Delta _{I}\lbrack 0\rbrack +\Delta _{I}\lbrack 0\rbrack )\rightarrow \Delta _{I}\lbrack 0\rbrack \)
  • -\(((\mathord {\textnormal {\textsf {s}}}^{0},\mathord {\textnormal {\textsf {s}}}^{1}),(\mathord {\textnormal {\textsf {s}}}^{1},\mathord {\textnormal {\textsf {s}}}^{0})):\Delta _{I}\lbrack 2\rbrack \mathbin {{}_{\mathord {\textnormal {\textsf {d}}}^{1}}\mathord {+_{\mathord {\textnormal {\textsf {d}}}^{1}}}}\Delta _{I}\lbrack 2\rbrack \rightarrow \Delta _{I}\lbrack 1\rbrack \times \Delta _{I}\lbrack 1\rbrack \)

The locality for the first two morphisms in [0032] is the Segal condition and the Rezk condition, respectively. We add the third morphism because \(I\) is not assumed to be totally ordered. Local objects for the third morphism “believe that \(I\) is totally ordered”.

[005A] Definition

Let \(U\) be a universe, let \(C\) be a \(U\)-logos, and let \(I\) be a partially ordered object in \(C\) with least and greatest elements. We say a morphism in \(C\) is a \(I\)-left fibration if it is internally right orthogonal to \(\mathord {\textnormal {\textsf {d}}}^{1}:\Delta _{I}\lbrack 0\rbrack \rightarrow \Delta _{I}\lbrack 1\rbrack \).

[0034] Exercise

Let \(U\) be a universe, let \(V\) be a universe greater than or equal to \(U\), and let \(W\) be a universe strictly greater than \(V\). Then \(\mathbb {A}^{(V)}:\mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),W)\) is \(\mathbb {I}\)-categorically fibrant.

[0035] Exercise

Let \(U\) be a universe, let \(V\) be a universe greater than or equal to \(U\), and let \(W\) be a universe strictly greater than \(V\). Then the morphism \(\mathord {\textnormal {\textsf {p}}}^{(V)}:\mathbb {A}_{\bullet }^{(V)}\rightarrow \mathbb {A}^{(V)}\) is a \(\mathbb {I}\)-left fibration.

\(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)\) itself appears in \(\mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)\) as a universe.

[0050] Notation

Let \(C\) be a lex category. We write \(\rho (C):\mathord {\textnormal {\textsf {R}}}_{\bullet }(C)\rightarrow \mathord {\textnormal {\textsf {R}}}(C)\) for the universal representable map of presheaves on \(C\); see Definition 5.2 of [Nguyen--Uemura--2022-0000].

[0051] Exercise

Let \(U\) be a universe and let \(V\) be a universe strictly greater than \(U\). Then \(\mathord {\textnormal {\textsf {R}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U))\) and \(\mathord {\textnormal {\textsf {R}}}_{\bullet }(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U))\) are in \(\mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)\).

[0056] Exercise

Let \(U\) be a universe and let \(V\) be a universe strictly greater than \(U\). Then \(\mathord {\textnormal {\textsf {R}}}_{\bullet }(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U)):\mathord {\textnormal {\textsf {Sh}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U),V)/\mathord {\textnormal {\textsf {R}}}(\mathord {\textnormal {\textsf {Topos}}}^{(1)}(U))\) is \(\mathbb {I}\)-categorically fibrant.