7. Constructions¶
7.1. Set and class¶
In Section 3.2 we saw that we could start with a category \(\mathcal C\) and obtain a (possibly different) category \(\mathcal C^{\mathrm{op}}\). There are many such constructions which produce new examples of categories.
7.1.1. Small and locally small category¶
As mentioned briefly in Section 2.2, there are set-theoretic difficulties that can arise when one attempts to handle some categories “all at once”. Namely, the objects and morphisms of a category often form proper classes. In order to prevent these issues in some of our constructions we introduce the following terminology.
A category is called small if both its objects and morphisms form sets.
A category \(\mathcal C\) is locally small if for every pair \(A\), \(B\) of objects in \(\mathcal C\) the collection of morphisms from \(A\) to \(B\) is a set. In this case, the set of morphisms from \(A\) to \(B\), called a “hom set,” is denoted by \(\mathrm{Hom}_{\mathcal C}(A, B)\).
Let’s pause to consider an example of a small category.
The category Ord \(_{\mathrm{fin}}\) of finite ordinals (also called the simplex category \(\Delta\)) has
- objects. \(\underline{n} = \{0, 1, \dots, n-1\}\) for each \(n \in \mathbb N\);
- morphisms. \(f \colon \underline{n} \to \underline{m}\) monotone functions (\(i\leq j \; \Rightarrow f i \leq f j\)).
Note that all morphisms of \(\Delta\) are generated by (i.e., can be written as finite compositions of) face maps and degeneracy maps.
Given a positive natural number \(n > 0\) and a natural number \(i\) such that \(0\leq i < n\), the face map \(\delta^n_i \colon \underline{n-1} \to \underline{n}\) is define as follows:
The degeneracy map \(\sigma^n_i \colon \underline{n+1} \to \underline{n}\) is define as follows:
The intuition is that the elements of \(\Delta\) are \(n\)-simplices, i.e., \(n\)-dimensional triangles.
An example of a large category is the category of small categories, denoted \(\mathbf{Cat}\).
Cat: | The (large) category of small categories has
|
---|
Naturally, the usual functor composition is morphism composition and the identity functors are the identity morphisms.
7.2. Functor category¶
Given categories \(\mathcal C\) and \(\mathcal D\), the functor category from \(\mathcal C\) to \(\mathcal D\) has
- objects. functors from \(F \colon \mathcal C \to \mathcal D\);
- morphisms. natural transformations \(\alpha \colon F \Rightarrow G\).
Some common notations for this category are \(\mathrm{Fun}(\mathcal C, \mathcal D)\) and \(\mathcal D^{\mathcal C}\) and \((\mathcal C, \mathcal D)\).
To define composition of morphisms in \(\mathcal D^{\mathcal C}\), suppose \(F, G, H\) are functors from \(\mathcal C\) to \(\mathcal D\), and suppose \(\alpha \colon F \Rightarrow G\) and \(\beta \colon G \Rightarrow H\) are natural transformations.
The composition of \(\alpha\) and \(\beta\) is the natural transformation \(\beta\alpha\colon F \Rightarrow H\) with components \((\beta \alpha)_A = \beta_A \alpha_A\).
A special case is the category \(\mathbf{Set}^{\mathcal C}\) of functors from a category \(\mathcal C\) into the category of sets.
If \(\mathcal C\) is a category, then \(\mathbf{Set}^{\mathcal C}\) is the category with
- objects. functors \(F \colon \mathcal C \to \mathbf{Set}\);
- morphisms. natural trasformations \(\alpha \colon F \Rightarrow D\).
7.2.1. Evaluation functor¶
With \(\mathbf{Set}^{\mathcal C}\) in hand, let us now revisit the evaluation functor introduced above. This is the functor \(Ev\colon \mathcal C \times \mathbf{Set}^{\mathcal C}\to \mathbf{Set}\), which takes each pair \((A, F) \in \mathcal C_0 \times (\mathbf{Set}^{\mathcal C})_0\) of objects to the set \(Ev(A, F) = FA\), and takes each pair \((g, \mu) \in \mathcal C_1 \times (\mathbf{Set}^{\mathcal C})_1\) of morphisms to a function on sets, namely
where \(g\in \mathcal C(A, A')\) and \(\mu \colon F \Rightarrow F'\).
7.3. Morphism category¶
We can create categories whose objects are the morphisms from a given category. Three standard cases are described below, but before describing them we recall some notational conventions appearing in these examples.
If \(\mathcal C\) is a category, then \(\mathcal C_0\) denotes the collection of objects of \(\mathcal C\).
If \(A, B \in \mathcal C_0\) then \(\mathcal C(A,B)\) denotes the collection of morphisms from \(A\) to \(B\) in \(\mathcal C\).
Arrow: | Given a category \(\mathcal C\), the arrow category \(\mathcal C^\rightarrow\) has
|
---|---|
Slice: | Given a category \(\mathcal C\) and an object \(C\) from \(\mathcal C\), the slice category \(\mathcal C/C\) has
|
Comma: | Given categories \(\mathcal C\) and \(\mathcal D\) and functors \(F \colon \mathcal C \to \mathcal D\) and \(G\colon \mathcal C' \to \mathcal D\) (with a common codomain), the comma category \((F \downarrow G)\) has
|
7.4. Presheaf¶
Another important special case of a functor category is the category of contravariant functors with values in sets. This is denoted by \(\mathbf{Set}^{\mathcal C^{\mathrm{op}}}\) and is called the category of presheaves on \(\mathcal C\).
In particular, the objects of \(\mathbf{Set}^{\Delta^{\mathrm{op}}}\) are called simplicial sets.
A simplicial set \(X \colon \Delta^{\mathrm{op}} \to \mathbf{Set}\) is a geometric object constructed by glueing together simplices, i.e., \(n\)-dimensional triangles.
\(X\underline{n}\) is the set of simplices in \(X\) of dimension \(n\).
The images of the face maps are used to specify how the simplices are glued together.
\(X\delta^n_i : X\underline{n} \to X\underline{(n-1)}\) specifies which \((n-1)\)-simplices are the \(i\)-th face of the \(n\)-simplices.
7.5. Solutions¶
Solution (to do).
Solution (to do).
Solution (to do).
Solution (to do).
Solution (to do).
Solution (to do).
Solution (to do).
Solution (to do).
The claim, “the identity map \(\mathrm{id}_{B^A × A}\) is a perfectly good substitute for the composition \((λ π_1 × \mathrm{id}_A) ∘ h\)” is false.
To see this, take an element \((f, a) ∈ B^A × A\) and trace around the commuting diagram in the definitions of product and exponential object. We arrive at \(f(a) = eval^A_B(f, a) = π_1 (h(f,a))\) and \(a = π_2 (h(f, a))\), from which it follows that \(h(f,a) = (f(a), a)\).
The function \(λ π_1: B → B^A\) takes \(b ∈ B\) to the constant map \(λ x . b\). Therefore,
where \(λ x . f(a)\) denotes the constant function that takes every \(x ∈ A\) to \(f(a)\).
If \((λ π_1 × \mathrm{id}_A) ∘ h\) were really the identity map, \(\mathrm{id}_{B^A × A}\), then we should have ended up with \((f, a)\) instead of \((λ x . f(a), a)\).