nLab equivalence in homotopy type theory

Equivalences in type theory

Equivalences in type theory

Idea

In homotopy type theory, the notion of equivalence is an internalization of the notion of equivalence or homotopy equivalence.

These are sometimes called weak equivalences, but there is nothing weak about them (in particular, they always have homotopy inverses).

Definition

We work in intensional type theory with dependent sums, dependent products, and identity types.

Definition

For $f\colon A\to B$ a term of function type; we define new dependent types as follows:

$f \colon A \to B, b\colon B \vdash hfiber(f,b) \coloneqq \sum_{a\colon A} (f(a) = b)$

and the proposition that the homotopy fiber type is a (dependently) contractible type:

$f \colon A \to B \vdash isEquiv(f) \coloneqq \prod_{b\colon B} isContr(hfiber(f,b)) \,.$

We say $f$ is an equivalence if $isEquiv(f)$ is an inhabited type.

That is, a function is an equivalence if all of its homotopy fiber types are contractible types (in a way which depends continuously on the base point).

Definition

For $X, Y : Type$ two types, the type of equivalences from $X$ to $Y$ is the dependent sum

$Equiv(X,Y) = (X \stackrel{\simeq}{\to}Y) \coloneqq \sum_{f : (X \to Y)} isEquiv(f) \,.$

Three variations of this definition are, informally:

• $f\colon A\to B$ is an equivalence if there is a map $g\colon B\to A$ and homotopies $p\colon \prod_{a\colon A} (g(f(a)) = a)$ and $q\colon \prod_{b\colon B} (f(g(b)) = b)$ (a homotopy equivalence)

• $f\colon A\to B$ is an equivalence if there is the above data, together with a higher homotopy expressing one triangle identity for $f$ and $g$ (an adjoint equivalence).

• $f\colon A\to B$ is an equivalence if there are maps $g,h\colon B\to A$ and homotopies $p\colon \prod_{a\colon A} (g(f(a)) = a)$ and $q\colon \prod_{b\colon B} (f(h(b)) = b)$ (sometimes called a homotopy isomorphism).

By formalizing these, we obtain types $homotopyEquiv(f)$, $isAdjointEquiv(f)$, and $isHIso(f)$. All four of these types are co-inhabited: we have a function from any one of them to any of the others. Moreover, at least if we assume function extensionality, the types $isAdjointEquiv(f)$ and $isHIso(f)$ are themselves equivalent to $isEquiv(f)$, and all three are h-propositions.

This is not true for $homotopyEquiv(f)$, which is not in general an h-prop even with function extensionality. However, often the most convenient way to show that $f$ is an equivalence is by exhibiting a term in $homotopyEquiv(f)$ (although such a term could just as well be interpreted to lie in $isHIso(f)$ with $h\coloneqq g$).

As one-to-one correspondences

Let $\mathcal{U}$ be a universe and $A:\mathcal{U}$ and $B:\mathcal{U}$ be terms of the universe, and $R :A \times B \to \mathcal{U}$ be a correspondence between $A$ and $B$. We define the property of $R$ being one-to-one as follows:

$isOneToOne(R) \coloneqq \left(\prod_{a:A} \mathrm{isContr}\left(\sum_{b:B} R(a,b)\right)\right) \times \left(\prod_{b:B} \mathrm{isContr}\left(\sum_{a:A} R(a,b)\right)\right)$

We define the type of equivalences from $A$ to $B$ in $\mathcal{U}$ as

$(A \simeq_\mathcal{U} B) \equiv \sum_{R : (A \times B) \to \mathcal{U}} isOneToOne(R)$

Rules for isEquiv

In any dependent type theory with identity types, function types, fiber types, and isContr defined either through isProp or contraction types, all of which could be defined without dependent product types or dependent sum types, we can still define isEquiv by adding the formation, introduction, elimination, computation, and uniqueness rules for isEquiv

Formation rules for isEquiv types:

$\frac{\Gamma \vdash A \; \mathrm{type} \quad \Gamma \vdash B \; \mathrm{type} \quad \Gamma, f:A \to B, y:B \vdash \mathrm{isContr}(\mathrm{fiber}_{A, B}(f, y)) \; \mathrm{type}}{\Gamma \vdash \mathrm{isEquiv}_{A, B}(f) \; \mathrm{type}}$

Introduction rules for isEquiv types:

$\frac{\Gamma \vdash A \; \mathrm{type} \quad \Gamma \vdash B \; \mathrm{type} \quad \Gamma, f:A \to B, y:B \vdash b(y):\mathrm{isContr}(\mathrm{fiber}_{A, B}(f, y))}{\Gamma, f:A \to B \vdash \lambda x.b(x):\mathrm{isEquiv}_{A, B}(f)}$

Elimination rules for isEquiv types:

$\frac{\Gamma, f:A \to B \vdash p:\mathrm{isEquiv}_{A, B}(f) \quad \Gamma \vdash c:B}{\Gamma, f:A \to B \vdash p(a):\mathrm{isEquiv}_{A, B}(f)(c)}$

Computation rules for isEquiv types:

$\frac{\Gamma, f:A \to B, y:B \vdash b(y):\mathrm{isContr}(\mathrm{fiber}_{A, B}(f, y)) \quad \Gamma \vdash c:B}{\Gamma, f:A \to B \vdash \beta_\mathrm{isEquiv}:\lambda x.b(x)(c) =_{\mathrm{isContr}(\mathrm{fiber}_{A, B}(f, c))} b(c)}$

Uniqueness rules for isEquiv types:

$\frac{\Gamma, f:A \to B \vdash p:\mathrm{isEquiv}_{A, B}(f)}{\Gamma, f:A \to B \vdash \eta_\mathrm{isEquiv}:p =_{\mathrm{isEquiv}_{A, B}(f)} \lambda(x).p(x)}$

Rules for equivalence types

We work in a dependent type theory with identity types, function types, and some set of rules for isEquiv defined above which does not require dependent product types or dependent sum types. The type of equivalences $A \simeq B$ is given by the following rules:

Formation rules for equivalence types:

$\frac{\Gamma \vdash A \; \mathrm{type} \quad \Gamma \vdash B \; \mathrm{type} \quad \Gamma, f:A \to B \vdash \mathrm{isEquiv}(f) \; \mathrm{type}}{\Gamma \vdash A \simeq B \; \mathrm{type}}$

Introduction rules for equivalence types:

$\frac{\Gamma \vdash A \; \mathrm{type} \quad \Gamma \vdash B \; \mathrm{type} \quad \Gamma, f:A \to B \vdash p(f):\mathrm{isEquiv}(f) \quad \Gamma \vdash g:A \to B \quad \Gamma \vdash p:\mathrm{isEquiv}[g/f]}{\Gamma \vdash (g, p):A \simeq B}$

Elimination rules for equivalence types:

$\frac{\Gamma \vdash h:A \simeq B}{\Gamma \vdash \pi_1(h):A \to B} \qquad \frac{\Gamma \vdash h:A \simeq B}{\Gamma \vdash \pi_2(h):\mathrm{isEquiv}(\pi_1(h))}$

Computation rules for equivalence types:

$\frac{\Gamma, f:A \to B \vdash p(f):\mathrm{isEquiv}(f) \quad \Gamma \vdash g:A \to B}{\Gamma \vdash \beta_{\simeq 1}:\pi_1(g, p) =_{A \to B} g} \qquad \frac{\Gamma, f:A \to B \vdash p:\mathrm{isEquiv} \quad \Gamma \vdash g:A \to B}{\Gamma \vdash \beta_{\simeq 2}:\pi_2(g, p) =_{\mathrm{isEquiv}(\pi_1(g, p))} p}$

Uniqueness rules for equivalence types:

$\frac{\Gamma \vdash h:A \simeq B}{\Gamma \vdash \eta_\simeq:h =_{A \simeq B} (\pi_1(h), \pi_2(h))}$

Semantics

We discuss the categorical semantics of equivalences in homotopy type theory.

Let $\mathcal{C}$ be a locally cartesian closed category which is a model category, in which the (acyclic cofibration, fibration) weak factorization system has stable path objects, and acyclic cofibrations are preserved by pullback along fibrations between fibrant objects. (We ignore questions of coherence, which are not important for this discussion.) For instance $\mathcal{C}$ could be a type-theoretic model category.

Of $isEquiv(-)$

Proposition

For $A, B$ two cofibrant-fibrant objects in $\mathcal{C}$, a morphism $f\colon A\to B$ is a weak equivalence or equivalently a homotopy equivalence in $\mathcal{C}$ precisely when the interpretation of $isEquiv(f)$ has a global point $* \to isEquiv(f)$.

Proof

For $f\colon A\to B$, the categorical semantics of the dependent type

$b\colon B \;\vdash\; hfiber(f,b)\colon Type$

is by the rules for the interpretation of identity types and substitution the mapping path space construction $P f$, given by the pullback

$\array{ [b : B \vdash hfiber(f,b)] &\coloneqq & P f &\to& A \\ && \downarrow && \downarrow^{\mathrlap{f}} \\ && B^I &\to& B \\ && \downarrow \\ && B }$

which, by the factorization lemma, is one way to factor $f$ as an acyclic cofibration followed by a fibration

$f : A \stackrel{\simeq}{\to} P f \to B \,.$

By definition and the semantics of contractible types, therefore, if $A$ and $B$ are cofibrant, then $isEquiv(f)$ has a global element

$* \to \prod_{b} isContr(hfiber(f,b))$

precisely when in this factorization, the fibration $P f \to B$ is an acyclic fibration. (See for instance (Shulman, page 49) for more details.)

But by the 2-out-of-3 property, this is equivalent to $f$ being a weak equivalence — and hence a homotopy equivalence, since it is a map between fibrant-cofibrant objects.

Remark

In the above we fixed one function $f : A \to X$. But the type $isEquiv$ is actually a dependent type

$f : A \to B \vdash isEquiv(f)$

on the type of all functions. To obtain the categorical semantics of this general dependent $isEquiv$-construction, first notice that the interpretation of

$f : A \to B,\; a : A,\; b : B \;\vdash\; (f(a) = b) \colon Type$

is by the rules for interpretation of identity types, evaluation and substitution the left vertical morphism in the pullback diagram

$\array{ Q &\to& B^I \\ \downarrow && \downarrow \\ [A,B] \times A \times B &\stackrel{(eval, id_B)}{\to}& B \times B } \,,$

where $eval : [A, B] \times A \to B$ is the evaluation map for the internal hom. This means that the interpretation of further dependent sum yielding $hfib$

$f : A \to B,\; b : B \; \vdash \; \left( \sum_{a : A } (f(a) = b) \right) \colon Type$

is the composite left vertical morphism in

$\array{ Q &\to& B^I \\ \downarrow && \downarrow \\ [A,B] \times A \times B &\stackrel{(eval, id_B)}{\to}& B \times B \\ \downarrow^{\mathrlap{p_{1,3}}} \\ [A,B] \times B }$

(…)

References

An introduction to equivalence in homotopy type theory is in

and basic ideas are also indicated from slide 60 of part 2, slide 49 of part 3 of

Coq code for homotopy equivalences is at

For equivalences as one-to-one correspondences in homotopy type theory, see

• Mike Shulman, Towards a Third-Generation HOTT Part 1 (slides, video)

Last revised on October 12, 2022 at 23:55:39. See the history of this page for a list of all contributions to it.