# The flattening lemma for pushouts ```agda module synthetic-homotopy-theory.flattening-lemma-pushouts where ``` <details><summary>Imports</summary> ```agda open import foundation.action-on-identifications-functions open import foundation.commuting-cubes-of-maps open import foundation.commuting-squares-of-maps open import foundation.commuting-triangles-of-maps open import foundation.dependent-pair-types open import foundation.equality-dependent-pair-types open import foundation.equivalences open import foundation.function-extensionality open import foundation.function-types open import foundation.functoriality-dependent-function-types open import foundation.functoriality-dependent-pair-types open import foundation.homotopies open import foundation.identity-types open import foundation.span-diagrams open import foundation.transport-along-identifications open import foundation.universal-property-dependent-pair-types open import foundation.universe-levels open import synthetic-homotopy-theory.cocones-under-spans open import synthetic-homotopy-theory.dependent-cocones-under-spans open import synthetic-homotopy-theory.dependent-universal-property-pushouts open import synthetic-homotopy-theory.descent-data-pushouts open import synthetic-homotopy-theory.equivalences-descent-data-pushouts open import synthetic-homotopy-theory.universal-property-pushouts ``` </details> ## Idea The **flattening lemma** for [pushouts](synthetic-homotopy-theory.pushouts.md) states that pushouts commute with [dependent pair types](foundation.dependent-pair-types.md). More precisely, given a pushout square ```text g S -----> B | | f| | j ∨ ⌜ ∨ A -----> X i ``` with homotopy `H : i ∘ f ~ j ∘ g`, and for any type family `P` over `X`, the commuting square ```text Σ (s : S), P(if(s)) ---> Σ (s : S), P(jg(s)) ---> Σ (b : B), P(j(b)) | | | | ∨ ⌜ ∨ Σ (a : A), P(i(a)) -----------------------------> Σ (x : X), P(x) ``` is again a pushout square. ## Definitions ### The statement of the flattening lemma for pushouts ```agda module _ { l1 l2 l3 l4 l5 : Level} {S : UU l1} {A : UU l2} {B : UU l3} { X : UU l4} (P : X → UU l5) ( f : S → A) (g : S → B) (c : cocone f g X) where vertical-map-span-flattening-pushout : Σ S (P ∘ horizontal-map-cocone f g c ∘ f) → Σ A (P ∘ horizontal-map-cocone f g c) vertical-map-span-flattening-pushout = map-Σ-map-base f (P ∘ horizontal-map-cocone f g c) horizontal-map-span-flattening-pushout : Σ S (P ∘ horizontal-map-cocone f g c ∘ f) → Σ B (P ∘ vertical-map-cocone f g c) horizontal-map-span-flattening-pushout = map-Σ ( P ∘ vertical-map-cocone f g c) ( g) ( λ s → tr P (coherence-square-cocone f g c s)) horizontal-map-cocone-flattening-pushout : Σ A (P ∘ horizontal-map-cocone f g c) → Σ X P horizontal-map-cocone-flattening-pushout = map-Σ-map-base (horizontal-map-cocone f g c) P vertical-map-cocone-flattening-pushout : Σ B (P ∘ vertical-map-cocone f g c) → Σ X P vertical-map-cocone-flattening-pushout = map-Σ-map-base (vertical-map-cocone f g c) P coherence-square-cocone-flattening-pushout : coherence-square-maps ( horizontal-map-span-flattening-pushout) ( vertical-map-span-flattening-pushout) ( vertical-map-cocone-flattening-pushout) ( horizontal-map-cocone-flattening-pushout) coherence-square-cocone-flattening-pushout = coherence-square-maps-map-Σ-map-base P g f ( vertical-map-cocone f g c) ( horizontal-map-cocone f g c) ( coherence-square-cocone f g c) cocone-flattening-pushout : cocone ( vertical-map-span-flattening-pushout) ( horizontal-map-span-flattening-pushout) ( Σ X P) pr1 cocone-flattening-pushout = horizontal-map-cocone-flattening-pushout pr1 (pr2 cocone-flattening-pushout) = vertical-map-cocone-flattening-pushout pr2 (pr2 cocone-flattening-pushout) = coherence-square-cocone-flattening-pushout flattening-lemma-pushout-statement : UUω flattening-lemma-pushout-statement = universal-property-pushout f g c → universal-property-pushout ( vertical-map-span-flattening-pushout) ( horizontal-map-span-flattening-pushout) ( cocone-flattening-pushout) module _ {l1 l2 l3 l4 l5 : Level} {𝒮 : span-diagram l1 l2 l3} {X : UU l4} (c : cocone-span-diagram 𝒮 X) (P : X → UU l5) where span-diagram-flattening-pushout : span-diagram (l1 ⊔ l5) (l2 ⊔ l5) (l3 ⊔ l5) span-diagram-flattening-pushout = make-span-diagram ( vertical-map-span-flattening-pushout P _ _ c) ( horizontal-map-span-flattening-pushout P _ _ c) ``` ### The statement of the flattening lemma for pushouts, phrased using descent data The above statement of the flattening lemma works with a provided type family over the pushout. We can instead accept a definition of this family via descent data for the pushout. ```agda module _ {l1 l2 l3 l4 : Level} {𝒮 : span-diagram l1 l2 l3} (P : descent-data-pushout 𝒮 l4) where vertical-map-span-flattening-descent-data-pushout : Σ ( spanning-type-span-diagram 𝒮) ( λ s → pr1 P (left-map-span-diagram 𝒮 s)) → Σ ( domain-span-diagram 𝒮) (pr1 P) vertical-map-span-flattening-descent-data-pushout = map-Σ-map-base ( left-map-span-diagram 𝒮) ( pr1 P) horizontal-map-span-flattening-descent-data-pushout : Σ ( spanning-type-span-diagram 𝒮) ( λ s → pr1 P (left-map-span-diagram 𝒮 s)) → Σ ( codomain-span-diagram 𝒮) (pr1 (pr2 P)) horizontal-map-span-flattening-descent-data-pushout = map-Σ ( pr1 (pr2 P)) ( right-map-span-diagram 𝒮) ( λ s → map-equiv (pr2 (pr2 P) s)) span-diagram-flattening-descent-data-pushout : span-diagram (l1 ⊔ l4) (l2 ⊔ l4) (l3 ⊔ l4) span-diagram-flattening-descent-data-pushout = make-span-diagram ( vertical-map-span-flattening-descent-data-pushout) ( horizontal-map-span-flattening-descent-data-pushout) module _ { l1 l2 l3 l4 l5 : Level} {S : UU l1} {A : UU l2} {B : UU l3} {X : UU l4} ( f : S → A) (g : S → B) (c : cocone f g X) ( P : descent-data-pushout (make-span-diagram f g) l5) ( Q : X → UU l5) ( e : equiv-descent-data-pushout P (descent-data-family-cocone-span-diagram c Q)) where horizontal-map-cocone-flattening-descent-data-pushout : Σ A (pr1 P) → Σ X Q horizontal-map-cocone-flattening-descent-data-pushout = map-Σ Q ( horizontal-map-cocone f g c) ( λ a → map-equiv (pr1 e a)) vertical-map-cocone-flattening-descent-data-pushout : Σ B (pr1 (pr2 P)) → Σ X Q vertical-map-cocone-flattening-descent-data-pushout = map-Σ Q ( vertical-map-cocone f g c) ( λ b → map-equiv (pr1 (pr2 e) b)) coherence-square-cocone-flattening-descent-data-pushout : coherence-square-maps ( horizontal-map-span-flattening-descent-data-pushout P) ( vertical-map-span-flattening-descent-data-pushout P) ( vertical-map-cocone-flattening-descent-data-pushout) ( horizontal-map-cocone-flattening-descent-data-pushout) coherence-square-cocone-flattening-descent-data-pushout = htpy-map-Σ Q ( coherence-square-cocone f g c) ( λ s → map-equiv (pr1 e (f s))) ( λ s → inv-htpy (pr2 (pr2 e) s)) cocone-flattening-descent-data-pushout : cocone ( vertical-map-span-flattening-descent-data-pushout P) ( horizontal-map-span-flattening-descent-data-pushout P) ( Σ X Q) pr1 cocone-flattening-descent-data-pushout = horizontal-map-cocone-flattening-descent-data-pushout pr1 (pr2 cocone-flattening-descent-data-pushout) = vertical-map-cocone-flattening-descent-data-pushout pr2 (pr2 cocone-flattening-descent-data-pushout) = coherence-square-cocone-flattening-descent-data-pushout flattening-lemma-descent-data-pushout-statement : UUω flattening-lemma-descent-data-pushout-statement = universal-property-pushout f g c → universal-property-pushout ( vertical-map-span-flattening-descent-data-pushout P) ( horizontal-map-span-flattening-descent-data-pushout P) ( cocone-flattening-descent-data-pushout) ``` ## Properties ### Proof of the flattening lemma for pushouts The proof uses the theorem that maps from sigma types are equivalent to dependent maps over the index type, for which we can invoke the dependent universal property of the indexing pushout. ```agda module _ { l1 l2 l3 l4 l5 : Level} {S : UU l1} {A : UU l2} {B : UU l3} { X : UU l4} (P : X → UU l5) ( f : S → A) (g : S → B) (c : cocone f g X) where cocone-map-flattening-pushout : { l : Level} (Y : UU l) → ( Σ X P → Y) → cocone ( vertical-map-span-flattening-pushout P f g c) ( horizontal-map-span-flattening-pushout P f g c) ( Y) cocone-map-flattening-pushout Y = cocone-map ( vertical-map-span-flattening-pushout P f g c) ( horizontal-map-span-flattening-pushout P f g c) ( cocone-flattening-pushout P f g c) comparison-dependent-cocone-ind-Σ-cocone : { l : Level} (Y : UU l) → Σ ( (a : A) → P (horizontal-map-cocone f g c a) → Y) ( λ k → Σ ( (b : B) → P (vertical-map-cocone f g c b) → Y) ( λ l → ( s : S) (t : P (horizontal-map-cocone f g c (f s))) → ( k (f s) t) = ( l (g s) (tr P (coherence-square-cocone f g c s) t)))) ≃ dependent-cocone f g c (λ x → P x → Y) comparison-dependent-cocone-ind-Σ-cocone Y = equiv-tot ( λ k → equiv-tot ( λ l → equiv-Π-equiv-family ( equiv-htpy-dependent-function-dependent-identification-function-type ( Y) ( coherence-square-cocone f g c) ( k ∘ f) ( l ∘ g)))) triangle-comparison-dependent-cocone-ind-Σ-cocone : { l : Level} (Y : UU l) → coherence-triangle-maps ( dependent-cocone-map f g c (λ x → P x → Y)) ( map-equiv (comparison-dependent-cocone-ind-Σ-cocone Y)) ( map-equiv equiv-ev-pair³ ∘ cocone-map-flattening-pushout Y ∘ ind-Σ) triangle-comparison-dependent-cocone-ind-Σ-cocone Y h = eq-pair-eq-fiber ( eq-pair-eq-fiber ( eq-htpy ( inv-htpy ( compute-equiv-htpy-dependent-function-dependent-identification-function-type ( Y) ( coherence-square-cocone f g c) ( h))))) abstract flattening-lemma-pushout : flattening-lemma-pushout-statement P f g c flattening-lemma-pushout up-c Y = is-equiv-left-factor ( cocone-map-flattening-pushout Y) ( ind-Σ) ( is-equiv-right-factor ( map-equiv equiv-ev-pair³) ( cocone-map-flattening-pushout Y ∘ ind-Σ) ( is-equiv-map-equiv equiv-ev-pair³) ( is-equiv-top-map-triangle ( dependent-cocone-map f g c (λ x → P x → Y)) ( map-equiv (comparison-dependent-cocone-ind-Σ-cocone Y)) ( ( map-equiv equiv-ev-pair³) ∘ ( cocone-map-flattening-pushout Y) ∘ ( ind-Σ)) ( triangle-comparison-dependent-cocone-ind-Σ-cocone Y) ( is-equiv-map-equiv (comparison-dependent-cocone-ind-Σ-cocone Y)) ( dependent-universal-property-universal-property-pushout _ _ _ up-c ( λ x → P x → Y)))) ( is-equiv-ind-Σ) ``` ### Proof of the descent data statement of the flattening lemma The proof is carried out by constructing a commuting cube, which has equivalences for vertical maps, the `cocone-flattening-pushout` square for the bottom, and the `cocone-flattening-descent-data-pushout` square for the top. The bottom is a pushout by the above flattening lemma, which implies that the top is also a pushout. The other parts of the cube are defined naturally, and come from the following map of spans: ```text Σ (a : A) (PA a) <------- Σ (s : S) (PA (f s)) -----> Σ (b : B) (PB b) | | | | | | ∨ ∨ ∨ Σ (a : A) (P (i a)) <---- Σ (s : S) (P (i (f s))) ---> Σ (b : B) (P (j b)) ``` where the vertical maps are equivalences given fiberwise by the equivalence of descent data. ```agda module _ { l1 l2 l3 l4 l5 : Level} {S : UU l1} {A : UU l2} {B : UU l3} {X : UU l4} ( f : S → A) (g : S → B) (c : cocone f g X) ( P : descent-data-pushout (make-span-diagram f g) l5) ( Q : X → UU l5) ( e : equiv-descent-data-pushout P (descent-data-family-cocone-span-diagram c Q)) where coherence-cube-flattening-lemma-descent-data-pushout : coherence-cube-maps ( vertical-map-span-flattening-pushout Q f g c) ( horizontal-map-span-flattening-pushout Q f g c) ( horizontal-map-cocone-flattening-pushout Q f g c) ( vertical-map-cocone-flattening-pushout Q f g c) ( vertical-map-span-flattening-descent-data-pushout P) ( horizontal-map-span-flattening-descent-data-pushout P) ( horizontal-map-cocone-flattening-descent-data-pushout f g c P Q e) ( vertical-map-cocone-flattening-descent-data-pushout f g c P Q e) ( tot (λ s → map-equiv (pr1 e (f s)))) ( tot (λ a → map-equiv (pr1 e a))) ( tot (λ b → map-equiv (pr1 (pr2 e) b))) ( id) ( coherence-square-cocone-flattening-descent-data-pushout f g c P Q e) ( refl-htpy) ( htpy-map-Σ ( Q ∘ vertical-map-cocone f g c) ( refl-htpy) ( λ s → tr Q (coherence-square-cocone f g c s) ∘ (map-equiv (pr1 e (f s)))) ( λ s → inv-htpy (pr2 (pr2 e) s))) ( refl-htpy) ( refl-htpy) ( coherence-square-cocone-flattening-pushout Q f g c) coherence-cube-flattening-lemma-descent-data-pushout (s , t) = ( ap-id ( coherence-square-cocone-flattening-descent-data-pushout f g c P Q e ( s , t))) ∙ ( triangle-eq-pair-Σ Q ( coherence-square-cocone f g c s) ( inv (pr2 (pr2 e) s t))) ∙ ( ap ( eq-pair-Σ (coherence-square-cocone f g c s) refl ∙_) ( inv ( ( right-unit) ∙ ( compute-ap-map-Σ-map-base-eq-pair-Σ ( vertical-map-cocone f g c) ( Q) ( refl) ( inv (pr2 (pr2 e) s t)))))) abstract flattening-lemma-descent-data-pushout : flattening-lemma-descent-data-pushout-statement f g c P Q e flattening-lemma-descent-data-pushout up-c = universal-property-pushout-top-universal-property-pushout-bottom-cube-is-equiv ( vertical-map-span-flattening-pushout Q f g c) ( horizontal-map-span-flattening-pushout Q f g c) ( horizontal-map-cocone-flattening-pushout Q f g c) ( vertical-map-cocone-flattening-pushout Q f g c) ( vertical-map-span-flattening-descent-data-pushout P) ( horizontal-map-span-flattening-descent-data-pushout P) ( horizontal-map-cocone-flattening-descent-data-pushout f g c P Q e) ( vertical-map-cocone-flattening-descent-data-pushout f g c P Q e) ( tot (λ s → map-equiv (pr1 e (f s)))) ( tot (λ a → map-equiv (pr1 e a))) ( tot (λ b → map-equiv (pr1 (pr2 e) b))) ( id) ( coherence-square-cocone-flattening-descent-data-pushout f g c P Q e) ( refl-htpy) ( htpy-map-Σ ( Q ∘ vertical-map-cocone f g c) ( refl-htpy) ( λ s → tr Q (coherence-square-cocone f g c s) ∘ (map-equiv (pr1 e (f s)))) ( λ s → inv-htpy (pr2 (pr2 e) s))) ( refl-htpy) ( refl-htpy) ( coherence-square-cocone-flattening-pushout Q f g c) ( coherence-cube-flattening-lemma-descent-data-pushout) ( is-equiv-tot-is-fiberwise-equiv ( λ s → is-equiv-map-equiv (pr1 e (f s)))) ( is-equiv-tot-is-fiberwise-equiv ( λ a → is-equiv-map-equiv (pr1 e a))) ( is-equiv-tot-is-fiberwise-equiv ( λ b → is-equiv-map-equiv (pr1 (pr2 e) b))) ( is-equiv-id) ( flattening-lemma-pushout Q f g c up-c) ```