Embeddings
Content created by Egbert Rijke, Fredrik Bakke, Jonathan Prieto-Cubides, Elisabeth Stenholm, Eléonore Mangel, Daniel Gratzer, Tom de Jong and Victor Blanchi.
Created on 2022-01-26.
Last modified on 2024-09-17.
module foundation.embeddings where open import foundation-core.embeddings public
Imports
open import foundation.action-on-identifications-functions open import foundation.cones-over-cospan-diagrams open import foundation.dependent-pair-types open import foundation.equivalences open import foundation.functoriality-cartesian-product-types open import foundation.functoriality-dependent-pair-types open import foundation.fundamental-theorem-of-identity-types open import foundation.identity-types open import foundation.transport-along-identifications open import foundation.truncated-maps open import foundation.universe-levels open import foundation-core.cartesian-product-types open import foundation-core.commuting-squares-of-maps open import foundation-core.commuting-triangles-of-maps open import foundation-core.contractible-types open import foundation-core.fibers-of-maps open import foundation-core.function-types open import foundation-core.homotopies open import foundation-core.propositional-maps open import foundation-core.propositions open import foundation-core.pullbacks open import foundation-core.retractions open import foundation-core.sections open import foundation-core.torsorial-type-families open import foundation-core.truncation-levels
Properties
Being an embedding is a property
module _ {l1 l2 : Level} {A : UU l1} {B : UU l2} where is-property-is-emb : (f : A → B) → is-prop (is-emb f) is-property-is-emb f = is-prop-Π (λ x → is-prop-Π (λ y → is-property-is-equiv (ap f))) is-emb-Prop : (A → B) → Prop (l1 ⊔ l2) pr1 (is-emb-Prop f) = is-emb f pr2 (is-emb-Prop f) = is-property-is-emb f
Embeddings are closed under homotopies
module _ {l1 l2 : Level} {A : UU l1} {B : UU l2} where abstract is-emb-htpy : {f g : A → B} (H : f ~ g) → is-emb g → is-emb f is-emb-htpy {f} {g} H is-emb-g x y = is-equiv-top-is-equiv-left-square ( ap g) ( concat' (f x) (H y)) ( ap f) ( concat (H x) (g y)) ( nat-htpy H) ( is-equiv-concat (H x) (g y)) ( is-emb-g x y) ( is-equiv-concat' (f x) (H y)) is-emb-htpy-emb : {f : A → B} (e : A ↪ B) → f ~ map-emb e → is-emb f is-emb-htpy-emb e H = is-emb-htpy H (is-emb-map-emb e) module _ {l1 l2 : Level} {A : UU l1} {B : UU l2} where abstract is-emb-htpy' : {f g : A → B} (H : f ~ g) → is-emb f → is-emb g is-emb-htpy' H is-emb-f = is-emb-htpy (inv-htpy H) is-emb-f is-emb-htpy-emb' : (e : A ↪ B) {g : A → B} → map-emb e ~ g → is-emb g is-emb-htpy-emb' e H = is-emb-htpy' H (is-emb-map-emb e)
Any map between propositions is an embedding
is-emb-is-prop : {l1 l2 : Level} {A : UU l1} {B : UU l2} {f : A → B} → is-prop A → is-prop B → is-emb f is-emb-is-prop H K = is-emb-is-prop-map (is-trunc-map-is-trunc-domain-codomain neg-one-𝕋 H K)
Embeddings are closed under composition
module _ {l1 l2 l3 : Level} {A : UU l1} {B : UU l2} {C : UU l3} where is-emb-comp : (g : B → C) (h : A → B) → is-emb g → is-emb h → is-emb (g ∘ h) is-emb-comp g h is-emb-g is-emb-h x y = is-equiv-left-map-triangle ( ap (g ∘ h)) ( ap g) ( ap h) ( ap-comp g h) ( is-emb-h x y) ( is-emb-g (h x) (h y)) abstract is-emb-left-map-triangle : (f : A → C) (g : B → C) (h : A → B) (H : coherence-triangle-maps f g h) → is-emb g → is-emb h → is-emb f is-emb-left-map-triangle f g h H is-emb-g is-emb-h = is-emb-htpy H (is-emb-comp g h is-emb-g is-emb-h) comp-emb : (B ↪ C) → (A ↪ B) → (A ↪ C) pr1 (comp-emb (g , H) (f , K)) = g ∘ f pr2 (comp-emb (g , H) (f , K)) = is-emb-comp g f H K
The right factor of a composed embedding is an embedding
module _ {l1 l2 l3 : Level} {A : UU l1} {B : UU l2} {C : UU l3} where is-emb-right-factor : (g : B → C) (h : A → B) → is-emb g → is-emb (g ∘ h) → is-emb h is-emb-right-factor g h is-emb-g is-emb-gh x y = is-equiv-top-map-triangle ( ap (g ∘ h)) ( ap g) ( ap h) ( ap-comp g h) ( is-emb-g (h x) (h y)) ( is-emb-gh x y) abstract is-emb-top-map-triangle : (f : A → C) (g : B → C) (h : A → B) (H : coherence-triangle-maps f g h) → is-emb g → is-emb f → is-emb h is-emb-top-map-triangle f g h H is-emb-g is-emb-f x y = is-equiv-top-map-triangle ( ap (g ∘ h)) ( ap g) ( ap h) ( ap-comp g h) ( is-emb-g (h x) (h y)) ( is-emb-htpy (inv-htpy H) is-emb-f x y) abstract is-emb-triangle-is-equiv : (f : A → C) (g : B → C) (e : A → B) (H : coherence-triangle-maps f g e) → is-equiv e → is-emb g → is-emb f is-emb-triangle-is-equiv f g e H is-equiv-e is-emb-g = is-emb-left-map-triangle f g e H is-emb-g (is-emb-is-equiv is-equiv-e) module _ {l1 l2 l3 : Level} {A : UU l1} {B : UU l2} {C : UU l3} where abstract is-emb-triangle-is-equiv' : (f : A → C) (g : B → C) (e : A → B) (H : coherence-triangle-maps f g e) → is-equiv e → is-emb f → is-emb g is-emb-triangle-is-equiv' f g e H is-equiv-e is-emb-f = is-emb-triangle-is-equiv g f ( map-inv-is-equiv is-equiv-e) ( triangle-section f g e H ( pair ( map-inv-is-equiv is-equiv-e) ( is-section-map-inv-is-equiv is-equiv-e))) ( is-equiv-map-inv-is-equiv is-equiv-e) ( is-emb-f)
The map on total spaces induced by a family of embeddings is an embedding
module _ {l1 l2 l3 : Level} {A : UU l1} {B : A → UU l2} {C : A → UU l3} where is-emb-tot : {f : (x : A) → B x → C x} → ((x : A) → is-emb (f x)) → is-emb (tot f) is-emb-tot H = is-emb-is-prop-map (is-prop-map-tot (λ x → is-prop-map-is-emb (H x))) emb-tot : ((x : A) → B x ↪ C x) → Σ A B ↪ Σ A C pr1 (emb-tot f) = tot (λ x → map-emb (f x)) pr2 (emb-tot f) = is-emb-tot (λ x → is-emb-map-emb (f x))
The functoriality of dependent pair types preserves embeddings
module _ {l1 l2 l3 : Level} {A : UU l1} {B : UU l2} where abstract is-emb-map-Σ-map-base : {f : A → B} (C : B → UU l3) → is-emb f → is-emb (map-Σ-map-base f C) is-emb-map-Σ-map-base C H = is-emb-is-prop-map (is-prop-map-map-Σ-map-base C (is-prop-map-is-emb H)) emb-Σ-emb-base : (f : A ↪ B) (C : B → UU l3) → Σ A (λ a → C (map-emb f a)) ↪ Σ B C pr1 (emb-Σ-emb-base f C) = map-Σ-map-base (map-emb f) C pr2 (emb-Σ-emb-base f C) = is-emb-map-Σ-map-base C (is-emb-map-emb f) module _ {l1 l2 l3 l4 : Level} {A : UU l1} {B : UU l2} {C : A → UU l3} where is-emb-map-Σ : (D : B → UU l4) {f : A → B} {g : (x : A) → C x → D (f x)} → is-emb f → ((x : A) → is-emb (g x)) → is-emb (map-Σ D f g) is-emb-map-Σ D H K = is-emb-is-prop-map ( is-prop-map-map-Σ D ( is-prop-map-is-emb H) ( λ x → is-prop-map-is-emb (K x))) emb-Σ : (D : B → UU l4) (f : A ↪ B) (g : (x : A) → C x ↪ D (map-emb f x)) → Σ A C ↪ Σ B D pr1 (emb-Σ D f g) = map-Σ D (map-emb f) (λ x → map-emb (g x)) pr2 (emb-Σ D f g) = is-emb-map-Σ D (is-emb-map-emb f) (λ x → is-emb-map-emb (g x))
Equivalence on total spaces induced by embedding on the base types
We saw above that given an embedding f : A ↪ B
and a type family C
over B
we obtain an embedding
Σ A (C ∘ f) ↪ Σ B C.
This embedding can be upgraded to an equivalence if we furthermore know that the
support of C
is contained in the image of f
. More precisely, if we are given
a section ((b , c) : Σ B C) → fiber f b
, then it follows that
Σ A (C ∘ f) ≃ Σ B C.
module _ {l1 l2 l3 : Level} {A : UU l1} {B : UU l2} {C : B → UU l3} (f : A ↪ B) (H : ((b , c) : Σ B C) → fiber (map-emb f) b) where map-inv-Σ-emb-base : Σ B C → Σ A (C ∘ map-emb f) pr1 (map-inv-Σ-emb-base u) = pr1 (H u) pr2 (map-inv-Σ-emb-base u) = inv-tr C (pr2 (H u)) (pr2 u) is-section-map-inv-Σ-emb-base : is-section (map-Σ-map-base (map-emb f) C) map-inv-Σ-emb-base is-section-map-inv-Σ-emb-base (b , c) = ap ( λ s → (pr1 s , inv-tr C (pr2 s) c)) ( eq-is-contr (is-torsorial-Id' b)) is-retraction-map-inv-Σ-emb-base : is-retraction (map-Σ-map-base (map-emb f) C) map-inv-Σ-emb-base is-retraction-map-inv-Σ-emb-base (a , c) = ap ( λ s → (pr1 s , inv-tr C (pr2 s) c)) ( eq-is-prop (is-prop-map-is-emb (pr2 f) (map-emb f a))) equiv-Σ-emb-base : Σ A (C ∘ map-emb f) ≃ Σ B C pr1 equiv-Σ-emb-base = map-Σ-map-base (map-emb f) C pr2 equiv-Σ-emb-base = is-equiv-is-invertible map-inv-Σ-emb-base is-section-map-inv-Σ-emb-base is-retraction-map-inv-Σ-emb-base
The product of two embeddings is an embedding
module _ {l1 l2 l3 l4 : Level} {A : UU l1} {B : UU l2} {C : UU l3} {D : UU l4} where emb-product : (A ↪ C) → (B ↪ D) → ((A × B) ↪ (C × D)) emb-product f g = emb-Σ (λ _ → D) f (λ _ → g) is-emb-map-product : {f : A → C} {g : B → D} → is-emb f → is-emb g → (is-emb (map-product f g)) is-emb-map-product {f} {g} is-emb-f is-emb-g = is-emb-map-emb (emb-product (f , is-emb-f) (g , is-emb-g))
If the action on identifications has a section, then f
is an embedding
module _ {l1 l2 : Level} {A : UU l1} {B : UU l2} (f : A → B) where abstract is-emb-section-ap : ((x y : A) → section (ap f {x} {y})) → is-emb f is-emb-section-ap section-ap-f x = fundamental-theorem-id-section x (λ y → ap f) (section-ap-f x)
If there is an equivalence (f x = f y) ≃ (x = y)
that sends refl
to refl
, then f is an embedding
module _ {l1 l2 : Level} {A : UU l1} {B : UU l2} (f : A → B) where abstract is-emb-equiv-refl-to-refl : (e : (x y : A) → (f x = f y) ≃ (x = y)) → ((x : A) → map-equiv (e x x) refl = refl) → is-emb f is-emb-equiv-refl-to-refl e p x y = is-equiv-htpy-equiv ( inv-equiv (e x y)) ( λ where refl → inv (is-retraction-map-inv-equiv (e x x) refl) ∙ ap (map-equiv (inv-equiv (e x x))) (p x))
Embeddings are closed under pullback
module _ {l1 l2 l3 l4 : Level} {A : UU l1} {B : UU l2} {C : UU l3} {X : UU l4} (f : A → X) (g : B → X) (c : cone f g C) where abstract is-emb-vertical-map-cone-is-pullback : is-pullback f g c → is-emb g → is-emb (vertical-map-cone f g c) is-emb-vertical-map-cone-is-pullback pb is-emb-g = is-emb-is-prop-map ( is-trunc-vertical-map-is-pullback neg-one-𝕋 f g c pb ( is-prop-map-is-emb is-emb-g)) abstract is-emb-horizontal-map-cone-is-pullback : is-pullback f g c → is-emb f → is-emb (horizontal-map-cone f g c) is-emb-horizontal-map-cone-is-pullback pb is-emb-f = is-emb-is-prop-map ( is-trunc-horizontal-map-is-pullback neg-one-𝕋 f g c pb ( is-prop-map-is-emb is-emb-f))
In a commuting square of which the sides are embeddings, the top map is an embedding if and only if the bottom map is an embedding
module _ {l1 l2 l3 l4 : Level} {A : UU l1} {B : UU l2} {C : UU l3} {D : UU l4} (top : A → C) (left : A → B) (right : C → D) (bottom : B → D) (H : coherence-square-maps top left right bottom) where is-emb-top-is-emb-bottom-is-equiv-coherence-square-maps : is-equiv left → is-equiv right → is-emb bottom → is-emb top is-emb-top-is-emb-bottom-is-equiv-coherence-square-maps K L M = is-emb-right-factor ( right) ( top) ( is-emb-is-equiv L) ( is-emb-htpy' ( H) ( is-emb-comp bottom left M (is-emb-is-equiv K))) module _ {l1 l2 l3 l4 : Level} {A : UU l1} {B : UU l2} {C : UU l3} {D : UU l4} (top : A → C) (left : A → B) (right : C → D) (bottom : B → D) (H : coherence-square-maps top left right bottom) where is-emb-bottom-is-emb-top-is-equiv-coherence-square-maps : is-equiv left → is-equiv right → is-emb top → is-emb bottom is-emb-bottom-is-emb-top-is-equiv-coherence-square-maps K L M = is-emb-top-is-emb-bottom-is-equiv-coherence-square-maps ( bottom) ( map-inv-is-equiv K) ( map-inv-is-equiv L) ( top) ( vertical-inv-equiv-coherence-square-maps ( top) ( left , K) ( right , L) ( bottom) ( H)) ( is-equiv-map-inv-is-equiv K) ( is-equiv-map-inv-is-equiv L) ( M)
A map is an embedding if and only if it has contractible fibers at values
module _ {l1 l2 : Level} {A : UU l1} {B : UU l2} (f : A → B) where is-emb-is-contr-fibers-values' : ((a : A) → is-contr (fiber' f (f a))) → is-emb f is-emb-is-contr-fibers-values' c a = fundamental-theorem-id (c a) (λ x → ap f {a} {x}) is-emb-is-contr-fibers-values : ((a : A) → is-contr (fiber f (f a))) → is-emb f is-emb-is-contr-fibers-values c = is-emb-is-contr-fibers-values' ( λ a → is-contr-equiv' ( fiber f (f a)) ( equiv-fiber f (f a)) ( c a)) is-contr-fibers-values-is-emb' : is-emb f → ((a : A) → is-contr (fiber' f (f a))) is-contr-fibers-values-is-emb' e a = fundamental-theorem-id' (λ x → ap f {a} {x}) (e a) is-contr-fibers-values-is-emb : is-emb f → ((a : A) → is-contr (fiber f (f a))) is-contr-fibers-values-is-emb e a = is-contr-equiv ( fiber' f (f a)) ( equiv-fiber f (f a)) ( is-contr-fibers-values-is-emb' e a)
Recent changes
- 2024-09-17. Fredrik Bakke. Some closure properties of decidable maps and embeddings (#1184).
- 2024-03-14. Egbert Rijke. Move torsoriality of the identity type to
foundation-core.torsorial-type-families
(#1065). - 2024-03-02. Fredrik Bakke. Factor out standard pullbacks (#1042).
- 2024-02-19. Fredrik Bakke. Additions for coherently invertible maps (#1024).
- 2024-02-06. Fredrik Bakke. Rename
(co)prod
to(co)product
(#1017).