graded.hlean

/- Graded (left-) R-modules for a ring R. -/

-- Author: Floris van Doorn

import .left_module .direct_sum .submodule --..heq

open is_trunc algebra eq left_module pointed function equiv is_equiv prod group sigma sigma.ops nat
  trunc_index property
namespace left_module

definition graded [reducible] (str : Type) (I : Type) : Type := I → str
definition graded_module [reducible] (R : Ring) : Type → Type := graded (LeftModule R)

-- TODO: We can (probably) make I a type everywhere
variables {R : Ring} {I : AddGroup} {M M₁ M₂ M₃ : graded_module R I}

/-
  morphisms between graded modules.
  The definition is unconventional in two ways:
  (1) The degree is determined by an endofunction instead of a element of I, which is equal to adding
    i on the right. This is more flexible. For example, the
    composition of two graded module homomorphisms φ₂ and φ₁ with degrees i₂ and i₁ has type
    M₁ i → M₂ ((i + i₁) + i₂).
    However, a homomorphism with degree i₁ + i₂ must have type
    M₁ i → M₂ (i + (i₁ + i₂)),
    which means that we need to insert a transport. With endofunctions this is not a problem:
    λi, (i + i₁) + i₂
    is a perfectly fine degree of a map
  (2) Since we cannot eliminate all possible transports, we don't define a homomorphism as function
    M₁ i →lm M₂ (i + deg f)  or  M₁ i →lm M₂ (deg f i)
    but as a function taking a path as argument. Specifically, for every path
    deg f i = j
    we get a function M₁ i → M₂ j.
-/

definition graded_hom_of_deg (d : I ≃ I) (M₁ M₂ : graded_module R I) : Type :=
Π⦃i j : I⦄ (p : d i = j), M₁ i →lm M₂ j

definition gmd_constant [constructor] (d : I ≃ I) (M₁ M₂ : graded_module R I) : graded_hom_of_deg d M₁ M₂ :=
λi j p, lm_constant (M₁ i) (M₂ j)

definition gmd0 [constructor] {d : I ≃ I} {M₁ M₂ : graded_module R I} : graded_hom_of_deg d M₁ M₂ :=
gmd_constant d M₁ M₂

structure graded_hom (M₁ M₂ : graded_module R I) : Type :=
mk' ::  (d : I ≃ I)
        (deg_eq : Π(i : I), d i = i + d 0)
        (fn' : graded_hom_of_deg d M₁ M₂)


definition deg_eq_id (i : I) : erfl i = i + erfl 0 :=
!add_zero⁻¹

definition deg_eq_inv {d : I ≃ I} (pd : Π(i : I), d i = i + d 0) (i : I) : d⁻¹ᵉ i = i + d⁻¹ᵉ 0 :=
inv_eq_of_eq (!pd ⬝ !neg_add_cancel_right)⁻¹ ⬝
ap (λx, i + x) ((to_left_inv d _)⁻¹ ⬝ ap d⁻¹ᵉ (!pd ⬝ add.left_inv (d 0)))

definition deg_eq_con {d₁ d₂ : I ≃ I} (pd₁ : Π(i : I), d₁ i = i + d₁ 0) (pd₂ : Π(i : I), d₂ i = i + d₂ 0)
  (i : I) : (d₁ ⬝e d₂) i = i + (d₁ ⬝e d₂) 0 :=
ap d₂ !pd₁ ⬝ !pd₂ ⬝ !add.assoc ⬝ ap (λx, i + x) !pd₂⁻¹

notation M₁ ` →gm ` M₂ := graded_hom M₁ M₂

abbreviation deg [unfold 5] := @graded_hom.d
abbreviation deg_eq [unfold 5] := @graded_hom.deg_eq
postfix ` ↘`:max := graded_hom.fn' -- there is probably a better character for this? Maybe ↷?

definition graded_hom_fn [reducible] [unfold 5] [coercion] (f : M₁ →gm M₂) (i : I) : M₁ i →lm M₂ (deg f i) :=
f ↘ idp

definition graded_hom_fn_out [reducible] [unfold 5] (f : M₁ →gm M₂) (i : I) : M₁ ((deg f)⁻¹ᵉ i) →lm M₂ i :=
f ↘ (to_right_inv (deg f) i)

infix ` ← `:max := graded_hom_fn_out -- todo: change notation

-- definition graded_hom_fn_out_rec (f : M₁ →gm M₂)
--   (P : Π{i j} (p : deg f i = j) (m : M₁ i) (n : M₂ j), Type)
--   (H : Πi m, P (right_inv (deg f) i) m (f ← i m)) {i j : I}
--   (p : deg f i = j) (m : M₁ i) (n : M₂ j) : P p m (f ↘ p m) :=
-- begin
--   revert i j p m n, refine equiv_rect (deg f)⁻¹ᵉ _ _, intro i,
--   refine eq.rec_to (right_inv (deg f) i) _,
--   intro m n, exact H i m
-- end

-- definition graded_hom_fn_rec (f : M₁ →gm M₂)
--   {P : Π{i j} (p : deg f i = j) (m : M₁ i) (n : M₂ j), Type}
--   (H : Πi m, P idp m (f i m)) ⦃i j : I⦄
--   (p : deg f i = j) (m : M₁ i) : P p m (f ↘ p m) :=
-- begin
--   induction p, apply H
-- end

-- definition graded_hom_fn_out_rec (f : M₁ →gm M₂)
--   {P : Π{i j} (p : deg f i = j) (m : M₁ i) (n : M₂ j), Type}
--   (H : Πi m, P idp m (f i m)) ⦃i : I⦄ (m : M₁ ((deg f)⁻¹ᵉ i)) :
--   P (right_inv (deg f) i) m (f ← i m) :=
-- graded_hom_fn_rec f H (right_inv (deg f) i) m

-- definition graded_hom_fn_out_rec_simple (f : M₁ →gm M₂)
--   {P : Π{j} (n : M₂ j), Type}
--   (H : Πi m, P (f i m)) ⦃i : I⦄ (m : M₁ ((deg f)⁻¹ᵉ i)) :
--   P (f ← i m) :=
-- graded_hom_fn_out_rec f H m

definition graded_hom.mk [constructor] (d : I ≃ I) (pd : Π(i : I), d i = i + d 0)
  (fn : Πi, M₁ i →lm M₂ (d i)) : M₁ →gm M₂ :=
graded_hom.mk' d pd (λi j p, homomorphism_of_eq (ap M₂ p) ∘lm fn i)

definition graded_hom.mk_out [constructor] (d : I ≃ I) (pd : Π(i : I), d i = i + d 0)
  (fn : Πi, M₁ (d⁻¹ i) →lm M₂ i) : M₁ →gm M₂ :=
graded_hom.mk' d pd (λi j p, fn j ∘lm homomorphism_of_eq (ap M₁ (eq_inv_of_eq p)))

definition graded_hom.mk_out' [constructor] (d : I ≃ I) (pd : Π(i : I), d i = i + d 0)
  (fn : Πi, M₁ (d i) →lm M₂ i) : M₁ →gm M₂ :=
graded_hom.mk' d⁻¹ᵉ (deg_eq_inv pd) (λi j p, fn j ∘lm homomorphism_of_eq (ap M₁ (eq_inv_of_eq p)))

definition graded_hom.mk_out_in [constructor] (d₁ d₂ : I ≃ I)
  (pd₁ : Π(i : I), d₁ i = i + d₁ 0) (pd₂ : Π(i : I), d₂ i = i + d₂ 0)
  (fn : Πi, M₁ (d₁ i) →lm M₂ (d₂ i)) : M₁ →gm M₂ :=
graded_hom.mk' (d₁⁻¹ᵉ ⬝e d₂) (deg_eq_con (deg_eq_inv pd₁) pd₂)
  (λi j p, homomorphism_of_eq (ap M₂ p) ∘lm fn (d₁⁻¹ᵉ i) ∘lm homomorphism_of_eq (ap M₁ (to_right_inv d₁ i)⁻¹))

definition graded_hom_eq_transport (f : M₁ →gm M₂) {i j : I} (p : deg f i = j) (m : M₁ i) :
  f ↘ p m = transport M₂ p (f i m) :=
by induction p; reflexivity

definition graded_hom_mk_refl (d : I ≃ I) (pd : Π(i : I), d i = i + d 0)
  (fn : Πi, M₁ i →lm M₂ (d i)) {i : I} (m : M₁ i) : graded_hom.mk d pd fn i m = fn i m :=
by reflexivity

lemma graded_hom_mk_out'_destruct (d : I ≃ I) (pd : Π(i : I), d i = i + d 0)
  (fn : Πi, M₁ (d i) →lm M₂ i) {i : I} (m : M₁ (d i)) :
  graded_hom.mk_out' d pd fn ↘ (left_inv d i) m = fn i m :=
begin
  unfold [graded_hom.mk_out'],
  apply ap (λx, fn i (cast x m)),
  refine !ap_compose⁻¹ ⬝ ap02 _ _,
  apply is_set.elim --note: we can also prove this if I is not a set
end

lemma graded_hom_mk_out_destruct (d : I ≃ I) (pd : Π(i : I), d i = i + d 0)
  (fn : Πi, M₁ (d⁻¹ i) →lm M₂ i) {i : I} (m : M₁ (d⁻¹ i)) :
  graded_hom.mk_out d pd fn ↘ (right_inv d i) m = fn i m :=
begin
  rexact graded_hom_mk_out'_destruct d⁻¹ᵉ (deg_eq_inv pd) fn m
end

lemma graded_hom_mk_out_in_destruct (d₁ : I ≃ I) (d₂ : I ≃ I)
  (pd₁ : Π(i : I), d₁ i = i + d₁ 0) (pd₂ : Π(i : I), d₂ i = i + d₂ 0)
  (fn : Πi, M₁ (d₁ i) →lm M₂ (d₂ i)) {i : I} (m : M₁ (d₁ i)) :
  graded_hom.mk_out_in d₁ d₂ pd₁ pd₂ fn ↘ (ap d₂ (left_inv d₁ i)) m = fn i m :=
begin
  unfold [graded_hom.mk_out_in],
  rewrite [adj d₁, -ap_inv, - +ap_compose, ],
  refine cast_fn_cast_square fn _ _ !con.left_inv m
end

definition graded_hom_square (f : M₁ →gm M₂) {i₁ i₂ j₁ j₂ : I} (p : deg f i₁ = j₁) (q : deg f i₂ = j₂)
  (r : i₁ = i₂) (s : j₁ = j₂) :
  hsquare (f ↘ p) (f ↘ q) (homomorphism_of_eq (ap M₁ r)) (homomorphism_of_eq (ap M₂ s)) :=
begin
  induction p, induction q, induction r,
  have rfl = s, from !is_set.elim, induction this,
  exact homotopy.rfl
end

variable (I) -- for some reason Lean needs to know what I is when applying this lemma
definition graded_hom_eq_zero {f : M₁ →gm M₂} {i j k : I} {q : deg f i = j} {p : deg f i = k}
  (m : M₁ i) (r : f ↘ q m = 0) : f ↘ p m = 0 :=
have f ↘ p m = transport M₂ (q⁻¹ ⬝ p) (f ↘ q m), begin induction p, induction q, reflexivity end,
this ⬝ ap (transport M₂ (q⁻¹ ⬝ p)) r ⬝ tr_eq_of_pathover (apd (λi, 0) (q⁻¹ ⬝ p))

variable {I}

definition graded_hom_change_image {f : M₁ →gm M₂} {i j k : I} {m : M₂ k} (p : deg f i = k)
  (q : deg f j = k) (h : image (f ↘ p) m) : image (f ↘ q) m :=
begin
  have Σ(r : i = j), ap (deg f) r = p ⬝ q⁻¹,
  from ⟨inj (deg f) (p ⬝ q⁻¹), !ap_inj'⟩,
  induction this with r s, induction r, induction q, esimp at s, induction s, exact h
end

definition graded_hom_codom_rec {f : M₁ →gm M₂} {j : I} {P : Π⦃i⦄, deg f i = j → Type}
  {i i' : I} (p : deg f i = j) (h : P p) (q : deg f i' = j) : P q :=
begin
  have Σ(r : i = i'), ap (deg f) r = p ⬝ q⁻¹,
  from ⟨inj (deg f) (p ⬝ q⁻¹), !ap_inj'⟩,
  induction this with r s, induction r, induction q, esimp at s, induction s, exact h
end

variables {f' : M₂ →gm M₃} {f g h : M₁ →gm M₂}

definition graded_hom_compose [constructor] (f' : M₂ →gm M₃) (f : M₁ →gm M₂) : M₁ →gm M₃ :=
graded_hom.mk' (deg f ⬝e deg f') (deg_eq_con (deg_eq f) (deg_eq f')) (λi j p, f' ↘ p ∘lm f i)

infixr ` ∘gm `:75 := graded_hom_compose

definition graded_hom_compose_fn (f' : M₂ →gm M₃) (f : M₁ →gm M₂) (i : I) (m : M₁ i) :
  (f' ∘gm f) i m = f' (deg f i) (f i m) :=
by reflexivity

definition graded_hom_compose_fn_ext (f' : M₂ →gm M₃) (f : M₁ →gm M₂) ⦃i j k : I⦄
  (p : deg f i = j) (q : deg f' j = k) (r : (deg f ⬝e deg f') i = k) (s : ap (deg f') p ⬝ q = r)
  (m : M₁ i) : ((f' ∘gm f) ↘ r) m = (f' ↘ q) (f ↘ p m) :=
by induction s; induction q; induction p; reflexivity

definition graded_hom_compose_fn_out (f' : M₂ →gm M₃) (f : M₁ →gm M₂) (i : I)
  (m : M₁ ((deg f ⬝e deg f')⁻¹ᵉ i)) : (f' ∘gm f) ← i m = f' ← i (f ← ((deg f')⁻¹ᵉ i) m) :=
graded_hom_compose_fn_ext f' f _ _ _ idp m

-- the following composition might be useful if you want tight control over the paths to which f and f' are applied
definition graded_hom_compose_ext [constructor] (f' : M₂ →gm M₃) (f : M₁ →gm M₂)
  (d : Π⦃i j⦄ (p : (deg f ⬝e deg f') i = j), I)
  (pf  : Π⦃i j⦄ (p : (deg f ⬝e deg f') i = j), deg f i = d p)
  (pf' : Π⦃i j⦄ (p : (deg f ⬝e deg f') i = j), deg f' (d p) = j) : M₁ →gm M₃ :=
graded_hom.mk' (deg f ⬝e deg f') (deg_eq_con (deg_eq f) (deg_eq f')) (λi j p, (f' ↘ (pf' p)) ∘lm (f ↘ (pf p)))

variable (M)
definition graded_hom_id [constructor] [refl] : M →gm M :=
graded_hom.mk erfl deg_eq_id (λi, lmid)
variable {M}
abbreviation gmid [constructor] := graded_hom_id M

/- reindexing a graded morphism along a group homomorphism.
  We could also reindex along an affine transformation, but don't prove that here
-/
definition graded_hom_reindex [constructor] {J : AddGroup} (e : J ≃g I) (f : M₁ →gm M₂) :
  (λy, M₁ (e y)) →gm (λy, M₂ (e y)) :=
graded_hom.mk' (group.equiv_of_isomorphism e ⬝e deg f ⬝e (group.equiv_of_isomorphism e)⁻¹ᵉ)
  begin intro i, exact ap e⁻¹ᵍ (deg_eq f (e i)) ⬝ respect_add e⁻¹ᵍ _ _ ⬝
    ap011 add (to_left_inv (group.equiv_of_isomorphism e) i)
              (ap (e⁻¹ᵍ ∘ deg f) (respect_zero e)⁻¹) end
  (λy₁ y₂ p, f ↘ (to_eq_of_inv_eq (group.equiv_of_isomorphism e) p))

definition gm_constant [constructor] (M₁ M₂ : graded_module R I) (d : I ≃ I) (pd : Π(i : I), d i = i + d 0)
   (pd : Π(i : I), d i = i + d 0) : M₁ →gm M₂ :=
graded_hom.mk' d pd (gmd_constant d M₁ M₂)

definition is_surjective_graded_hom_compose ⦃x z⦄
  (f' : M₂ →gm M₃) (f : M₁ →gm M₂) (p : deg f' (deg f x) = z)
  (H' : Π⦃y⦄ (q : deg f' y = z), is_surjective (f' ↘ q))
  (H : Π⦃y⦄ (q : deg f x = y), is_surjective (f ↘ q)) : is_surjective ((f' ∘gm f) ↘ p) :=
begin
  induction p,
  apply is_surjective_compose (f' (deg f x)) (f x),
  apply H', apply H
end

structure graded_iso (M₁ M₂ : graded_module R I) : Type :=
mk' :: (to_hom : M₁ →gm M₂)
       (is_equiv_to_hom : Π⦃i j⦄ (p : deg to_hom i = j), is_equiv (to_hom ↘ p))

infix ` ≃gm `:25 := graded_iso
attribute graded_iso.to_hom [coercion]
attribute graded_iso._trans_of_to_hom [unfold 5]

definition is_equiv_graded_iso [instance] [priority 1010] (φ : M₁ ≃gm M₂) (i : I) :
  is_equiv (φ i) :=
graded_iso.is_equiv_to_hom φ idp

definition isomorphism_of_graded_iso' [constructor] (φ : M₁ ≃gm M₂) {i j : I} (p : deg φ i = j) :
  M₁ i ≃lm M₂ j :=
isomorphism.mk (φ ↘ p) !graded_iso.is_equiv_to_hom

definition isomorphism_of_graded_iso [constructor] (φ : M₁ ≃gm M₂) (i : I) :
  M₁ i ≃lm M₂ (deg φ i) :=
isomorphism.mk (φ i) _

definition isomorphism_of_graded_iso_out [constructor] (φ : M₁ ≃gm M₂) (i : I) :
  M₁ ((deg φ)⁻¹ᵉ i) ≃lm M₂ i :=
isomorphism_of_graded_iso' φ (to_right_inv (deg φ) i)

protected definition graded_iso.mk [constructor] (d : I ≃ I)  (pd : Π(i : I), d i = i + d 0)
  (φ : Πi, M₁ i ≃lm M₂ (d i)) : M₁ ≃gm M₂ :=
begin
  apply graded_iso.mk' (graded_hom.mk d pd φ),
  intro i j p, induction p,
  exact to_is_equiv (equiv_of_isomorphism (φ i)),
end

protected definition graded_iso.mk_out [constructor] (d : I ≃ I)
   (pd : Π(i : I), d i = i + d 0) (φ : Πi, M₁ (d⁻¹ i) ≃lm M₂ i) :
  M₁ ≃gm M₂ :=
begin
  apply graded_iso.mk' (graded_hom.mk_out d pd φ),
  intro i j p, esimp,
  exact @is_equiv_compose _ _ _ _ _ !is_equiv_cast _,
end

definition graded_iso_of_eq [constructor] {M₁ M₂ : graded_module R I} (p : M₁ ~ M₂)
  : M₁ ≃gm M₂ :=
graded_iso.mk erfl deg_eq_id (λi, isomorphism_of_eq (p i))

-- definition to_gminv [constructor] (φ : M₁ ≃gm M₂) : M₂ →gm M₁ :=
-- graded_hom.mk_out (deg φ)⁻¹ᵉ
--   abstract begin
--     intro i, apply isomorphism.to_hom, symmetry,
--     apply isomorphism_of_graded_iso φ
--   end end

variable (M)
definition graded_iso.refl [refl] [constructor] : M ≃gm M :=
graded_iso.mk equiv.rfl deg_eq_id (λi, isomorphism.rfl)
variable {M}

definition graded_iso.rfl [refl] [constructor] : M ≃gm M := graded_iso.refl M

definition graded_iso.symm [symm] [constructor] (φ : M₁ ≃gm M₂) : M₂ ≃gm M₁ :=
graded_iso.mk_out (deg φ)⁻¹ᵉ (deg_eq_inv (deg_eq φ)) (λi, (isomorphism_of_graded_iso φ i)⁻¹ˡᵐ)

definition graded_iso.trans [trans] [constructor] (φ : M₁ ≃gm M₂) (ψ : M₂ ≃gm M₃) : M₁ ≃gm M₃ :=
graded_iso.mk (deg φ ⬝e deg ψ) (deg_eq_con (deg_eq φ) (deg_eq ψ))
  (λi, isomorphism_of_graded_iso φ i ⬝lm isomorphism_of_graded_iso ψ (deg φ i))

definition graded_iso.eq_trans [trans] [constructor]
   {M₁ M₂ M₃ : graded_module R I} (φ : M₁ ~ M₂) (ψ : M₂ ≃gm M₃) : M₁ ≃gm M₃ :=
proof graded_iso.trans (graded_iso_of_eq φ) ψ qed

definition graded_iso.trans_eq [trans] [constructor]
  {M₁ M₂ M₃ : graded_module R I} (φ : M₁ ≃gm M₂) (ψ : M₂ ~ M₃) : M₁ ≃gm M₃ :=
graded_iso.trans φ (graded_iso_of_eq ψ)

postfix `⁻¹ᵉᵍᵐ`:(max + 1) := graded_iso.symm
infixl ` ⬝egm `:75 := graded_iso.trans
infixl ` ⬝egmp `:75 := graded_iso.trans_eq
infixl ` ⬝epgm `:75 := graded_iso.eq_trans

definition graded_hom_of_eq [constructor] {M₁ M₂ : graded_module R I} (p : M₁ ~ M₂) : M₁ →gm M₂ :=
proof graded_iso_of_eq p qed

definition fooff {I : Set} (P : I → Type) {i j : I} (M : P i) (N : P j) := unit
notation M ` ==[`:50 P:0 `] `:0 N:50 := fooff P M N

definition graded_homotopy (f g : M₁ →gm M₂) : Type :=
Π⦃i j k⦄ (p : deg f i = j) (q : deg g i = k) (m : M₁ i), f ↘ p m ==[λ(i : Set_of_AddGroup I), M₂ i] g ↘ q m
-- mk' :: (hd : deg f ~ deg g)
--        (hfn : Π⦃i j : I⦄ (pf : deg f i = j) (pg : deg g i = j), f ↘ pf ~ g ↘ pg)

infix ` ~gm `:50 := graded_homotopy

-- definition graded_homotopy.mk2 (hd : deg f ~ deg g) (hfn : Πi m, f i m =[hd i] g i m) : f ~gm g :=
-- graded_homotopy.mk' hd
--   begin
--     intro i j pf pg m, induction (is_set.elim (hd i ⬝ pg) pf), induction pg, esimp,
--     exact graded_hom_eq_transport f (hd i) m ⬝ tr_eq_of_pathover (hfn i m),
--   end

definition graded_homotopy.mk (h : Πi m, f i m ==[λ(i : Set_of_AddGroup I), M₂ i] g i m) : f ~gm g :=
begin
  intros i j k p q m, induction q, induction p, constructor --exact h i m
end

-- definition graded_hom_compose_out {d₁ d₂ : I ≃ I} (f₂ : Πi, M₂ i →lm M₃ (d₂ i))
--   (f₁ : Πi, M₁ (d₁⁻¹ i) →lm M₂ i) : graded_hom.mk d₂ f₂ ∘gm graded_hom.mk_out d₁ f₁ ~gm
--   graded_hom.mk_out_in d₁⁻¹ᵉ d₂ _ :=
-- _

-- definition graded_hom_out_in_compose_out {d₁ d₂ d₃ : I ≃ I} (f₂ : Πi, M₂ (d₂ i) →lm M₃ (d₃ i))
--   (f₁ : Πi, M₁ (d₁⁻¹ i) →lm M₂ i) : graded_hom.mk_out_in d₂ d₃ f₂ ∘gm graded_hom.mk_out d₁ f₁ ~gm
--   graded_hom.mk_out_in (d₂ ⬝e d₁⁻¹ᵉ) d₃ (λi, f₂ i ∘lm (f₁ (d₂ i))) :=
-- begin
--   apply graded_homotopy.mk, intro i m, exact sorry
-- end

-- definition graded_hom_out_in_rfl {d₁ d₂ : I ≃ I} (f : Πi, M₁ i →lm M₂ (d₂ i))
--   (p : Πi, d₁ i = i) :
--   graded_hom.mk_out_in d₁ d₂ (λi, sorry) ~gm graded_hom.mk d₂ f :=
-- begin
--   apply graded_homotopy.mk, intro i m, exact sorry
-- end

-- definition graded_homotopy.trans (h₁ : f ~gm g) (h₂ : g ~gm h) : f ~gm h :=
-- begin
--   exact sorry
-- end

-- postfix `⁻¹ᵍᵐ`:(max + 1) := graded_iso.symm
--infixl ` ⬝gm `:75 := graded_homotopy.trans
-- infixl ` ⬝gmp `:75 := graded_iso.trans_eq
-- infixl ` ⬝pgm `:75 := graded_iso.eq_trans


-- definition graded_homotopy_of_deg (d : I ≃ I) (f g : graded_hom_of_deg d M₁ M₂) : Type :=
-- Π⦃i j : I⦄ (p : d i = j), f p ~ g p

-- notation f ` ~[`:50 d:0 `] `:0 g:50 := graded_homotopy_of_deg d f g

-- variables {d : I ≃ I} {f₁ f₂ : graded_hom_of_deg d M₁ M₂}

-- definition graded_homotopy_of_deg.mk [constructor] (h : Πi, f₁ (idpath (d i)) ~ f₂ (idpath (d i))) :
--   f₁ ~[d] f₂ :=
-- begin
--   intro i j p, induction p, exact h i
-- end

-- definition graded_homotopy.mk_out [constructor] {M₁ M₂ : graded_module R I} (d : I ≃ I)
--   (fn : Πi, M₁ (d⁻¹ i) →lm M₂ i) : M₁ →gm M₂ :=
-- graded_hom.mk' d (λi j p, fn j ∘lm homomorphism_of_eq (ap M₁ (eq_inv_of_eq p)))
-- definition is_gconstant (f : M₁ →gm M₂) : Type :=
-- f↘ ~[deg f] gmd0

definition compose_constant (f' : M₂ →gm M₃) (f : M₁ →gm M₂) : Type :=
Π⦃i j k : I⦄ (p : deg f i = j) (q : deg f' j = k) (m : M₁ i), f' ↘ q (f ↘ p m) = 0

definition compose_constant.mk (h : Πi m, f' (deg f i) (f i m) = 0) : compose_constant f' f :=
by intros; induction p; induction q; exact h i m

definition compose_constant.elim (h : compose_constant f' f) (i : I) (m : M₁ i) : f' (deg f i) (f i m) = 0 :=
h idp idp m

definition is_gconstant (f : M₁ →gm M₂) : Type :=
Π⦃i j : I⦄ (p : deg f i = j) (m : M₁ i), f ↘ p m = 0

definition is_gconstant.mk (h : Πi m, f i m = 0) : is_gconstant f :=
by intros; induction p; exact h i m

definition is_gconstant.elim (h : is_gconstant f) (i : I) (m : M₁ i) : f i m = 0 :=
h idp m

/- direct sum of graded R-modules -/

variables {J : Set} (N : graded_module R J)
definition dirsum' : AddAbGroup :=
group.dirsum (λj, AddAbGroup_of_LeftModule (N j))
variable {N}
definition dirsum_smul [constructor] (r : R) : dirsum' N →a dirsum' N :=
dirsum_functor (λi, smul_homomorphism (N i) r)

definition dirsum_smul_right_distrib (r s : R) (n : dirsum' N) :
  dirsum_smul (r + s) n = dirsum_smul r n + dirsum_smul s n :=
begin
  refine dirsum_functor_homotopy _ _ _ n ⬝ !dirsum_functor_mul⁻¹,
  intro i ni, exact to_smul_right_distrib r s ni
end

definition dirsum_mul_smul' (r s : R) (n : dirsum' N) :
  dirsum_smul (r * s) n = (dirsum_smul r ∘a dirsum_smul s) n :=
begin
  refine dirsum_functor_homotopy _ _ _ n ⬝ (dirsum_functor_compose _ _ n)⁻¹ᵖ,
  intro i ni, exact to_mul_smul r s ni
end

definition dirsum_mul_smul (r s : R) (n : dirsum' N) :
  dirsum_smul (r * s) n = dirsum_smul r (dirsum_smul s n) :=
proof dirsum_mul_smul' r s n qed

definition dirsum_one_smul (n : dirsum' N) : dirsum_smul 1 n = n :=
begin
  refine dirsum_functor_homotopy _ _ _ n ⬝ !dirsum_functor_gid,
  intro i ni, exact to_one_smul ni
end

definition dirsum : LeftModule R :=
LeftModule_of_AddAbGroup (dirsum' N) (λr n, dirsum_smul r n)
  proof (λr, homomorphism.addstruct (dirsum_smul r)) qed
  proof dirsum_smul_right_distrib qed
  proof dirsum_mul_smul qed
  proof dirsum_one_smul qed

/- graded variants of left-module constructions -/

definition graded_submodule [constructor] (S : Πi, property (M i)) [Π i, is_submodule (M i) (S i)] :
  graded_module R I :=
λi, submodule (S i)

definition graded_submodule_incl [constructor] (S : Πi, property (M i)) [H : Π i, is_submodule (M i) (S i)] :
  graded_submodule S →gm M :=
have Π i, is_submodule (M (to_fun erfl i)) (S i), from H,
graded_hom.mk erfl deg_eq_id (λi, submodule_incl (S i))

definition graded_hom_lift [constructor] (S : Πi, property (M₂ i)) [Π i, is_submodule (M₂ i) (S i)]
  (φ : M₁ →gm M₂)
  (h : Π(i : I) (m : M₁ i), φ i m ∈ S (deg φ i)) : M₁ →gm graded_submodule S :=
graded_hom.mk (deg φ) (deg_eq φ) (λi, hom_lift (φ i) (h i))

definition graded_submodule_functor [constructor]
  {S : Πi, property (M₁ i)} [Π i, is_submodule (M₁ i) (S i)]
  {T : Πi, property (M₂ i)} [Π i, is_submodule (M₂ i) (T i)]
  (φ : M₁ →gm M₂)
  (h : Π(i : I) (m : M₁ i), S i m → T (deg φ i) (φ i m)) :
  graded_submodule S →gm graded_submodule T :=
graded_hom.mk (deg φ) (deg_eq φ) (λi, submodule_functor (φ i) (h i))

definition graded_image (f : M₁ →gm M₂) : graded_module R I :=
λi, image_module (f ← i)

lemma graded_image_lift_lemma (f : M₁ →gm M₂) {i j: I} (p : deg f i = j) (m : M₁ i) :
  image (f ← j) (f ↘ p m) :=
graded_hom_change_image p (right_inv (deg f) j) (image.mk m idp)

definition graded_image_lift [constructor] (f : M₁ →gm M₂) : M₁ →gm graded_image f :=
graded_hom.mk' (deg f) (deg_eq f) (λi j p, hom_lift (f ↘ p) (graded_image_lift_lemma f p))

definition graded_image_lift_destruct (f : M₁ →gm M₂) {i : I}
  (m : M₁ ((deg f)⁻¹ᵉ i)) : graded_image_lift f ← i m = image_lift (f ← i) m :=
subtype_eq idp

definition graded_image.rec {f : M₁ →gm M₂} {i : I} {P : graded_image f (deg f i) → Type}
  [h : Πx, is_prop (P x)] (H : Πm, P (graded_image_lift f i m)) : Πm, P m :=
begin
  assert H₂ : Πi' (p : deg f i' = deg f i) (m : M₁ i'),
    P ⟨f ↘ p m, graded_hom_change_image p _ (image.mk m idp)⟩,
  { refine eq.rec_equiv_symm (deg f) _, intro m,
    refine transport P _ (H m), apply subtype_eq, reflexivity },
  refine @total_image.rec _ _ _ _ h _, intro m,
  refine transport P _ (H₂ _ (right_inv (deg f) (deg f i)) m),
  apply subtype_eq, reflexivity
end

definition image_graded_image_lift {f : M₁ →gm M₂} {i j : I} (p : deg f i = j)
  (m : graded_image f j)
  (h : image (f ↘ p) m.1) : image (graded_image_lift f ↘ p) m :=
begin
  induction p,
  revert m h, refine total_image.rec _, intro m h,
  induction h with n q, refine image.mk n (subtype_eq q)
end

lemma is_surjective_graded_image_lift ⦃x y⦄ (f : M₁ →gm M₂)
  (p : deg f x = y) : is_surjective (graded_image_lift f ↘ p) :=
begin
  intro m, apply image_graded_image_lift, exact graded_hom_change_image (right_inv (deg f) y) _ m.2
end

definition graded_image_elim_helper {f : M₁ →gm M₂} (g : M₁ →gm M₃)
  (h : Π⦃i m⦄, f i m = 0 → g i m = 0) (i : I) : graded_image f (deg f i) →lm M₃ (deg g i) :=
begin
  apply image_elim (g ↘ (ap (deg g) (to_left_inv (deg f) i))),
  intro m p,
  refine graded_hom_eq_zero I m (h _),
  exact graded_hom_eq_zero I m p
end

definition graded_image_elim [constructor] {f : M₁ →gm M₂} (g : M₁ →gm M₃)
  (h : Π⦃i m⦄, f i m = 0 → g i m = 0) :
  graded_image f →gm M₃ :=
graded_hom.mk_out_in (deg f) (deg g) (deg_eq f) (deg_eq g) (graded_image_elim_helper g h)

lemma graded_image_elim_destruct {f : M₁ →gm M₂} {g : M₁ →gm M₃}
  (h : Π⦃i m⦄, f i m = 0 → g i m = 0) {i j k : I}
  (p' : deg f i = j) (p : deg g ((deg f)⁻¹ᵉ j) = k)
  (q : deg g i = k) (r : ap (deg g) (to_left_inv (deg f) i) ⬝ q = ap ((deg f)⁻¹ᵉ ⬝e deg g) p' ⬝ p)
  (m : M₁ i) : graded_image_elim g h ↘ p (graded_image_lift f ↘ p' m) =
  g ↘ q m :=
begin
  revert i j p' k p q r m,
  refine equiv_rect (deg f ⬝e (deg f)⁻¹ᵉ) _ _,
  intro i, refine eq.rec_grading _ (deg f) (right_inv (deg f) (deg f i)) _,
  intro k p q r m,
  assert r' : q = p,
  { refine cancel_left _ (r ⬝ whisker_right _ _), refine !ap_compose ⬝ ap02 (deg g) _,
    exact !adj_inv⁻¹ },
  induction r', clear r,
  revert k q m, refine eq.rec_to (ap (deg g) (to_left_inv (deg f) i)) _, intro m,
  refine graded_hom_mk_out_in_destruct (deg f) (deg g) (deg_eq f) (deg_eq g)
           (graded_image_elim_helper g h) (graded_image_lift f ← (deg f i) m) ⬝ _,
  refine ap (image_elim _ _) !graded_image_lift_destruct ⬝ _, reflexivity
end

/- alternative (easier) definition of graded_image with "wrong" grading -/

-- definition graded_image' (f : M₁ →gm M₂) : graded_module R I :=
-- λi, image_module (f i)

-- definition graded_image'_lift [constructor] (f : M₁ →gm M₂) : M₁ →gm graded_image' f :=
-- graded_hom.mk erfl (λi, image_lift (f i))

-- definition graded_image'_elim [constructor] {f : M₁ →gm M₂} (g : M₁ →gm M₃)
--   (h : Π⦃i m⦄, f i m = 0 → g i m = 0) :
--   graded_image' f →gm M₃ :=
-- begin
--   apply graded_hom.mk (deg g),
--   intro i,
--   apply image_elim (g i),
--   intro m p, exact h p
-- end

-- theorem graded_image'_elim_compute {f : M₁ →gm M₂} {g : M₁ →gm M₃}
--   (h : Π⦃i m⦄, f i m = 0 → g i m = 0) :
--   graded_image'_elim g h ∘gm graded_image'_lift f ~gm g :=
-- begin
--   apply graded_homotopy.mk,
--   intro i m, exact sorry --reflexivity
-- end

-- theorem graded_image_elim_compute {f : M₁ →gm M₂} {g : M₁ →gm M₃}
--   (h : Π⦃i m⦄, f i m = 0 → g i m = 0) :
--   graded_image_elim g h ∘gm graded_image_lift f ~gm g :=
-- begin
--   refine _ ⬝gm graded_image'_elim_compute h,
--   esimp, exact sorry
--   -- refine graded_hom_out_in_compose_out _ _ ⬝gm _, exact sorry
--   -- -- apply graded_homotopy.mk,
--   -- -- intro i m,
-- end

-- variables {α β : I ≃ I}
-- definition gen_image (f : M₁ →gm M₂) (p : Πi, deg f (α i) = β i) : graded_module R I :=
-- λi, image_module (f ↘ (p i))

-- definition gen_image_lift [constructor] (f : M₁ →gm M₂) (p : Πi, deg f (α i) = β i) : M₁ →gm gen_image f p :=
-- graded_hom.mk_out α⁻¹ᵉ (λi, image_lift (f ↘ (p i)))

-- definition gen_image_elim [constructor] {f : M₁ →gm M₂} (p : Πi, deg f (α i) = β i) (g : M₁ →gm M₃)
--   (h : Π⦃i m⦄, f i m = 0 → g i m = 0) :
--   gen_image f p →gm M₃ :=
-- begin
--   apply graded_hom.mk_out_in α⁻¹ᵉ (deg g),
--   intro i,
--   apply image_elim (g ↘ (ap (deg g) (to_right_inv α i))),
--   intro m p,
--   refine graded_hom_eq_zero m (h _),
--   exact graded_hom_eq_zero m p
-- end

-- theorem gen_image_elim_compute {f : M₁ →gm M₂} {p : deg f ∘ α ~ β} {g : M₁ →gm M₃}
--   (h : Π⦃i m⦄, f i m = 0 → g i m = 0) :
--   gen_image_elim p g h ∘gm gen_image_lift f p ~gm g :=
-- begin
--   -- induction β with β βe, esimp at *, induction p using homotopy.rec_on_idp,
--   assert q : β ⬝e (deg f)⁻¹ᵉ = α,
--   { apply equiv_eq, intro i, apply inv_eq_of_eq, exact (p i)⁻¹ },
--   induction q,
--   -- unfold [gen_image_elim, gen_image_lift],

--   -- induction (is_prop.elim (λi, to_right_inv (deg f) (β i)) p),
--   -- apply graded_homotopy.mk,
--   -- intro i m, reflexivity
--   exact sorry
-- end

definition graded_kernel (f : M₁ →gm M₂) : graded_module R I :=
λi, kernel_module (f i)

definition graded_quotient (S : Πi, property (M i)) [Π i, is_submodule (M i) (S i)] : graded_module R I :=
λi, quotient_module (S i)

definition graded_quotient_map [constructor] (S : Πi, property (M i)) [Π i, is_submodule (M i) (S i)] :
  M →gm graded_quotient S :=
graded_hom.mk erfl deg_eq_id (λi, quotient_map (S i))

definition graded_quotient_elim [constructor]
  (S : Πi, property (M i)) [Π i, is_submodule (M i) (S i)]
  (φ : M →gm M₂)
  (H : Πi ⦃m⦄, S i m → φ i m = 0) : graded_quotient S →gm M₂ :=
graded_hom.mk (deg φ) (deg_eq φ) (λi, quotient_elim (φ i) (H i))

definition graded_homology (g : M₂ →gm M₃) (f : M₁ →gm M₂) : graded_module R I :=
graded_quotient (λ i, homology_quotient_property (g i) (f ← i))

-- the two reasonable definitions of graded_homology are definitionally equal
example (g : M₂ →gm M₃) (f : M₁ →gm M₂) :
  (λi, homology (g i) (f ← i)) = graded_homology g f := idp

definition graded_homology.mk (g : M₂ →gm M₃) (f : M₁ →gm M₂) {i : I} (m : M₂ i) (h : g i m = 0) :
  graded_homology g f i :=
homology.mk _ m h

definition graded_homology_intro [constructor] (g : M₂ →gm M₃) (f : M₁ →gm M₂) :
  graded_kernel g →gm graded_homology g f :=
@graded_quotient_map _ _ _ (λ i, homology_quotient_property (g i) (f ← i)) _

definition graded_homology_elim {g : M₂ →gm M₃} {f : M₁ →gm M₂} (h : M₂ →gm M)
  (H : compose_constant h f) : graded_homology g f →gm M :=
graded_hom.mk (deg h) (deg_eq h) (λi, homology_elim (h i) (H _ _))

definition graded_homology_isomorphism (g : M₂ →gm M₃) (f : M₁ →gm M₂) (x : I) :
  graded_homology g f (deg f x) ≃lm homology (g (deg f x)) (f x) :=
begin
  refine homology_isomorphism_homology (isomorphism_of_eq (ap M₁ (left_inv (deg f) x)))
    isomorphism.rfl isomorphism.rfl homotopy.rfl _,
  exact graded_hom_square f (to_right_inv (deg f) (deg f x)) idp (to_left_inv (deg f) x) idp
end

definition graded_homology_isomorphism_kernel_module
  (g : M₂ →gm M₃) (f : M₁ →gm M₂) (x : I)
  (H : Πm, image (f ← x) m → m = 0) : graded_homology g f x ≃lm graded_kernel g x :=
begin
  apply quotient_module_isomorphism, intro m h, apply subtype_eq, apply H, exact h
end

definition image_of_graded_homology_intro_eq_zero {g : M₂ →gm M₃} {f : M₁ →gm M₂}
  ⦃i j : I⦄ (p : deg f i = j) (m : graded_kernel g j) (H : graded_homology_intro g f j m = 0) :
  image (f ↘ p) m.1 :=
begin
  induction p, exact graded_hom_change_image _ _
    (@rel_of_quotient_map_eq_zero _ _ _ _ m H)
end

definition is_exact_gmod (f : M₁ →gm M₂) (f' : M₂ →gm M₃) : Type :=
Π⦃i j k⦄ (p : deg f i = j) (q : deg f' j = k), is_exact_mod (f ↘ p) (f' ↘ q)

definition is_exact_gmod.mk {f : M₁ →gm M₂} {f' : M₂ →gm M₃}
  (h₁ : Π⦃i⦄ (m : M₁ i), f' (deg f i) (f i m) = 0)
  (h₂ : Π⦃i⦄ (m : M₂ (deg f i)), f' (deg f i) m = 0 → image (f i) m) : is_exact_gmod f f' :=
begin intro i j k p q; induction p; induction q; split, apply h₁, apply h₂ end

definition gmod_im_in_ker (h : is_exact_gmod f f') : compose_constant f' f :=
λi j k p q, is_exact.im_in_ker (h p q)

definition gmod_ker_in_im (h : is_exact_gmod f f') ⦃i : I⦄ (m : M₂ i) (p : f' i m = 0) :
  image (f ← i) m :=
is_exact.ker_in_im (h (right_inv (deg f) i) idp) m p

definition is_exact_gmod_reindex [constructor] {J : AddGroup} (e : J ≃g I) (h : is_exact_gmod f f') :
  is_exact_gmod (graded_hom_reindex e f) (graded_hom_reindex e f') :=
λi j k p q, h (eq_of_inv_eq p) (eq_of_inv_eq q)

definition deg_commute {I : AddAbGroup} {M₁ M₂ M₃ M₄ : graded_module R I} (f : M₁ →gm M₂)
  (g : M₃ →gm M₄) : hsquare (deg f) (deg f) (deg g) (deg g) :=
begin
  intro i,
  refine ap (deg g) (deg_eq f i) ⬝ deg_eq g _ ⬝ _ ⬝ (ap (deg f) (deg_eq g i) ⬝ deg_eq f _)⁻¹,
  exact !add.assoc ⬝ ap (λx, i + x) !add.comm ⬝ !add.assoc⁻¹
end


end left_module