-- Raw terms, weakening (renaming) and substitution.

{-# OPTIONS --without-K --safe #-}

module Definition.Untyped {a} (M : Set a) where

open import Tools.Fin
open import Tools.Nat
open import Tools.Product
open import Tools.List
import Tools.PropositionalEquality as PE

private
  variable
    p q r : M
    n m  : Nat

infixl 30 _∙_
infixr 5 _∷_
infix 30 Π_,_▷_▹_
infix 30 Σ_▷_▹_
infix 30 Σₚ_▷_▹_
infix 30 Σᵣ_▷_▹_
infix 30 Σ⟨_⟩_▷_▹_
infix 30 ⟦_⟧_▹_
infixl 30 _ₛ•ₛ_ _•ₛ_ _ₛ•_
infix 25 _[_]
infix 25 _[_]↑
infix 25 _[_,_]
infix 25 _[_]↑²


-- Typing contexts (length indexed snoc-lists, isomorphic to lists).
-- Terms added to the context are well scoped in the sense that it cannot
-- contain more unbound variables than can be looked up in the context.

data Con (A : Nat  Set a) : Nat  Set a where
  ε   :                             Con A 0        -- Empty context.
  _∙_ : {n : Nat}  Con A n  A n  Con A (1+ n)   -- Context extension.

-- Representation of sub terms using a list of binding levels

data GenTs (A : Nat  Set a) : Nat  List Nat  Set a where
  []  : {n : Nat}  GenTs A n []
  _∷_ : {n b : Nat} {bs : List Nat} (t : A (b + n)) (ts : GenTs A n bs)  GenTs A n (b  bs)

-- Sigma types have two modes, allowing either projections or prodrec
data SigmaMode : Set where
  Σₚ Σᵣ : SigmaMode

-- Kinds are indexed on the number of expected sub terms
-- and the number of new variables bound by each sub term

data Kind : (ns : List Nat)  Set a where
  Ukind : Kind []

  Pikind  : (p q : M)  Kind (0  1  [])
  Lamkind : (p : M)    Kind (1  [])
  Appkind : (p : M)    Kind (0  0  [])

  Sigmakind   : (q : M)  SigmaMode  Kind (0  1  [])
  Prodkind    : Kind (0  0  [])
  Fstkind     : Kind (0  [])
  Sndkind     : Kind (0  [])
  Prodreckind : (p : M)  Kind (1  0  2  [])

  Natkind    : Kind []
  Zerokind   : Kind []
  Suckind    : Kind (0  [])
  Natreckind : (p q : M)  Kind (1  0  2  0  [])

  Unitkind : Kind []
  Starkind : Kind []

  Emptykind    : Kind []
  Emptyreckind : (p : M)  Kind (0  0  [])

-- Terms are indexed by its number of unbound variables and are either:
-- de Bruijn style variables or
-- generic terms, formed by their kind and sub terms

data Term (n : Nat) : Set a where
  var : (x : Fin n)  Term n
  gen : {bs : List Nat} (k : Kind bs) (ts : GenTs Term n bs)  Term n

private
  variable
    A F H t u v : Term n
    B E G t′    : Term (1+ n)

-- The Grammar of our language.

-- We represent the expressions of our language as de Bruijn terms.
-- Variables are natural numbers interpreted as de Bruijn indices.
-- Π, lam, and natrec are binders.

-- Type constructors.
U      : Term n                      -- Universe.
U = gen Ukind []

Π_,_▷_▹_ : (p q : M) (A : Term n) (B : Term (1+ n))  Term n -- Dependent function type (B is a binder).
Π p , q  A  B = gen (Pikind p q) (A  B  [])

Σᵣ_▷_▹_ : (p : M) (A : Term n) (B : Term (1+ n))  Term n -- Dependent sum type (B is a binder).
Σᵣ q  A  B = gen (Sigmakind q Σᵣ) (A  B  [])

Σₚ_▷_▹_ : (p : M) (A : Term n) (B : Term (1+ n))  Term n -- Dependent sum type (B is a binder).
Σₚ q  A  B = gen (Sigmakind q Σₚ) (A  B  [])

      : Term n                      -- Type of natural numbers.
 = gen Natkind []

Empty : Term n                       -- Empty type
Empty = gen Emptykind []

Unit  : Term n                       -- Unit type
Unit = gen Unitkind []

lam    : (p : M) (t : Term (1+ n))  Term n  -- Function abstraction (binder).
lam p t = gen (Lamkind p) (t  [])

_∘_▷_    : (t : Term n) (p : M) (u : Term n)  Term n -- Application.
t  p  u = gen (Appkind p) (t  u  [])


prod : (t u : Term n)  Term n       -- Dependent products
prod t u = gen Prodkind (t  u  [])

fst : (t : Term n)  Term n          -- First projection
fst t = gen Fstkind (t  [])

snd : (t : Term n)  Term n          -- Second projection
snd t = gen Sndkind (t  [])

prodrec : (p : M) (A : Term (1+ n)) (t : Term n) (u : Term (1+ (1+ n)))  Term n
prodrec p A t u = gen (Prodreckind p) (A  t  u  [])

-- Introduction and elimination of natural numbers.
zero   : Term n                      -- Natural number zero.
zero = gen Zerokind []

suc    : (t : Term n)        Term n -- Successor.
suc t = gen Suckind (t  [])

natrec : (p q : M) (A : Term (1+ n)) (z : Term n)
         (s : Term (1+ (1+ n))) (t : Term n)  Term n  -- Natural number recursor (A is a binder).
natrec p q A z s n = gen (Natreckind p q) (A  z  s  n  [])


star : Term n                        -- Unit element
star = gen Starkind []

Emptyrec : (p : M) (A e : Term n)  Term n   -- Empty type recursor
Emptyrec p A e = gen (Emptyreckind p) (A  e  [])

-- Binding types

data BindingType : Set a where
   : (p q : M)  BindingType
   : (p : M)  SigmaMode  BindingType

pattern BΠ! =  _ _
pattern BΣ! =  _ _
pattern BΣᵣ =  _ Σᵣ
pattern BΣₚ =  _ Σₚ
pattern Σ_▷_▹_ q A B = gen (Sigmakind q _) (A  B  [])
pattern Σ⟨_⟩_▷_▹_ m q A B = gen (Sigmakind q m) (A  B  [])

⟦_⟧_▹_ : BindingType  Term n  Term (1+ n)  Term n
  p q  F  G = Π p , q  F  G
  q m  F  G = Σ⟨ m  q  F  G

-- Injectivity of term constructors w.r.t. propositional equality.

-- If  W F G = W' H E  then  F = H,  G = E and W = W'.

B-PE-injectivity :  W W'   W  F  G PE.≡  W'  H  E
                  F PE.≡ H × G PE.≡ E × W PE.≡ W'
B-PE-injectivity ( p q) ( .p .q) PE.refl = PE.refl , PE.refl , PE.refl
B-PE-injectivity ( q m) ( .q .m) PE.refl = PE.refl , PE.refl , PE.refl

BΠ-PE-injectivity :  {p p′ q q′}   p q PE.≡  p′ q′  p PE.≡ p′ × q PE.≡ q′
BΠ-PE-injectivity PE.refl = PE.refl , PE.refl

BΣ-PE-injectivity :  {q q′ m m′}   q m PE.≡  q′ m′  q PE.≡ q′ × m PE.≡ m′
BΣ-PE-injectivity PE.refl = PE.refl , PE.refl

-- If  suc n = suc m  then  n = m.

suc-PE-injectivity : suc t PE.≡ suc u  t PE.≡ u
suc-PE-injectivity PE.refl = PE.refl


-- Neutral terms.

-- A term is neutral if it has a variable in head position.
-- The variable blocks reduction of such terms.

data Neutral : Term n  Set a where
  var       : (x : Fin n)  Neutral (var x)
  ∘ₙ        : Neutral t    Neutral (t  p  u)
  fstₙ      : Neutral t    Neutral (fst t)
  sndₙ      : Neutral t    Neutral (snd t)
  natrecₙ   : Neutral v    Neutral (natrec p q G t u v)
  prodrecₙ  : Neutral t    Neutral (prodrec p A t u)
  Emptyrecₙ : Neutral t    Neutral (Emptyrec p A t)


-- Weak head normal forms (whnfs).

-- These are the (lazy) values of our language.

data Whnf {n : Nat} : Term n  Set a where

  -- Type constructors are whnfs.
  Uₙ     : Whnf U
  Πₙ     : Whnf (Π p , q  A  B)
  Σₙ     :  {m}  Whnf (Σ⟨ m  q  A  B)
  ℕₙ     : Whnf 
  Unitₙ  : Whnf Unit
  Emptyₙ : Whnf Empty

  -- Introductions are whnfs.
  lamₙ  : Whnf (lam p t)
  zeroₙ : Whnf zero
  sucₙ  : Whnf (suc t)
  starₙ : Whnf star
  prodₙ : Whnf (prod t u)

  -- Neutrals are whnfs.
  ne    : Neutral t  Whnf t


-- Whnf inequalities.

-- Different whnfs are trivially distinguished by propositional equality.
-- (The following statements are sometimes called "no-confusion theorems".)

U≢ne : Neutral A  U PE.≢ A
U≢ne () PE.refl

ℕ≢ne : Neutral A   PE.≢ A
ℕ≢ne () PE.refl

Empty≢ne : Neutral A  Empty PE.≢ A
Empty≢ne () PE.refl

Unit≢ne : Neutral A  Unit PE.≢ A
Unit≢ne () PE.refl

B≢ne :  W  Neutral A   W  F  G PE.≢ A
B≢ne ( p q) () PE.refl
B≢ne ( q m) () PE.refl

U≢B :  W  U PE.≢  W  F  G
U≢B ( p q) ()
U≢B ( q m) ()

ℕ≢B :  W   PE.≢  W  F  G
ℕ≢B ( p q) ()
ℕ≢B ( q m) ()

Empty≢B :  W  Empty PE.≢  W  F  G
Empty≢B ( p q) ()
Empty≢B ( q m) ()

Unit≢B :  W  Unit PE.≢  W  F  G
Unit≢B ( p q) ()
Unit≢B ( q m) ()

zero≢ne : Neutral t  zero PE.≢ t
zero≢ne () PE.refl

suc≢ne : Neutral t  suc u PE.≢ t
suc≢ne () PE.refl

prod≢ne : Neutral v  prod t u PE.≢ v
prod≢ne () PE.refl

-- Several views on whnfs (note: not recursive).

-- A whnf of type ℕ is either zero, suc t, or neutral.

data Natural {n : Nat} : Term n  Set a where
  zeroₙ :             Natural zero
  sucₙ  :             Natural (suc t)
  ne    : Neutral t  Natural t


-- A (small) type in whnf is either Π A B, Σ A B, ℕ, Empty, Unit or neutral.
-- Large types could also be U.

data Type {n : Nat} : Term n  Set a where
  Πₙ     :             Type (Π p , q  A  B)
  Σₙ     :  {m}      Type (Σ⟨ m  p  A  B)
  ℕₙ     :             Type 
  Emptyₙ :             Type Empty
  Unitₙ  :             Type Unit
  ne     : Neutral t  Type t

⟦_⟧-type :  (W : BindingType)  Type ( W  F  G)
  p q ⟧-type = Πₙ
  q m ⟧-type = Σₙ

-- A whnf of type Π A ▹ B is either lam t or neutral.

data Function {n : Nat} : Term n  Set a where
  lamₙ : Function (lam p t)
  ne   : Neutral t  Function t

-- A whnf of type Σ A ▹ B is either prod t u or neutral.

data Product {n : Nat} : Term n  Set a where
  prodₙ : Product (prod t u)
  ne    : Neutral t  Product t


-- These views classify only whnfs.
-- Natural, Type, Function and Product are a subsets of Whnf.

naturalWhnf : Natural t  Whnf t
naturalWhnf sucₙ   = sucₙ
naturalWhnf zeroₙ  = zeroₙ
naturalWhnf (ne x) = ne x

typeWhnf : Type A  Whnf A
typeWhnf Πₙ     = Πₙ
typeWhnf Σₙ     = Σₙ
typeWhnf ℕₙ     = ℕₙ
typeWhnf Emptyₙ = Emptyₙ
typeWhnf Unitₙ  = Unitₙ
typeWhnf (ne x) = ne x

functionWhnf : Function t  Whnf t
functionWhnf lamₙ   = lamₙ
functionWhnf (ne x) = ne x

productWhnf : Product t  Whnf t
productWhnf prodₙ  = prodₙ
productWhnf (ne x) = ne x

⟦_⟧ₙ : (W : BindingType)  Whnf ( W  F  G)
⟦_⟧ₙ ( p q) = Πₙ
⟦_⟧ₙ ( q m) = Σₙ

------------------------------------------------------------------------
-- Weakening

-- In the following we define untyped weakenings η : Wk.
-- The typed form could be written η : Γ ≤ Δ with the intention
-- that η transport a term t living in context Δ to a context Γ
-- that can bind additional variables (which cannot appear in t).
-- Thus, if Δ ⊢ t : A and η : Γ ≤ Δ then Γ ⊢ wk η t : wk η A.
--
-- Even though Γ is "larger" than Δ we write Γ ≤ Δ to be conformant
-- with subtyping A ≤ B.  With subtyping, relation Γ ≤ Δ could be defined as
-- ``for all x ∈ dom(Δ) have Γ(x) ≤ Δ(x)'' (in the sense of subtyping)
-- and this would be the natural extension of weakenings.

data Wk : Nat  Nat  Set where
  id    : {n : Nat}    Wk n n                    -- η : Γ ≤ Γ.
  step  : {n m : Nat}  Wk m n  Wk (1+ m) n      -- If η : Γ ≤ Δ then step η : Γ∙A ≤ Δ.
  lift  : {n m : Nat}  Wk m n  Wk (1+ m) (1+ n) -- If η : Γ ≤ Δ then lift η : Γ∙A ≤ Δ∙A.

-- Composition of weakening.
-- If η : Γ ≤ Δ and η′ : Δ ≤ Φ then η • η′ : Γ ≤ Φ.

infixl 30 _•_

_•_                :  {l m n : Nat}  Wk l m  Wk m n  Wk l n
id       η′       =  η′
step η   η′       =  step  (η  η′)
lift η   id       =  lift  η
lift η   step η′  =  step  (η  η′)
lift η   lift η′  =  lift  (η  η′)

liftn : {k m : Nat}  Wk k m  (n : Nat)  Wk (n + k) (n + m)
liftn ρ 0 = ρ
liftn ρ (1+ n) = lift (liftn ρ n)

-- Weakening of variables.
-- If η : Γ ≤ Δ and x ∈ dom(Δ) then wkVar η x ∈ dom(Γ).

wkVar : {m n : Nat} (ρ : Wk m n) (x : Fin n)  Fin m
wkVar id x            = x
wkVar (step ρ) x      = (wkVar ρ x) +1
wkVar (lift ρ) x0     = x0
wkVar (lift ρ) (x +1) = (wkVar ρ x) +1

  -- Weakening of terms.
  -- If η : Γ ≤ Δ and Δ ⊢ t : A then Γ ⊢ wk η t : wk η A.

mutual
  wkGen : {m n : Nat} {bs : List Nat} (ρ : Wk m n) (c : GenTs (Term) n bs)  GenTs (Term) m bs
  wkGen ρ []                = []
  wkGen ρ (_∷_ {b = b} t c) = (wk (liftn ρ b) t)  (wkGen ρ c)

  wk : {m n : Nat} (ρ : Wk m n) (t : Term n)  Term m
  wk ρ (var x)   = var (wkVar ρ x)
  wk ρ (gen k c) = gen k (wkGen ρ c)


-- Adding one variable to the context requires wk1.
-- If Γ ⊢ t : B then Γ∙A ⊢ wk1 t : wk1 B.

wk1 : Term n  Term (1+ n)
wk1 = wk (step id)

-- Weakening of a neutral term.

wkNeutral :  ρ  Neutral t  Neutral {n = n} (wk ρ t)
wkNeutral ρ (var n)       = var (wkVar ρ n)
wkNeutral ρ (∘ₙ n)        = ∘ₙ (wkNeutral ρ n)
wkNeutral ρ (fstₙ n)      = fstₙ (wkNeutral ρ n)
wkNeutral ρ (sndₙ n)      = sndₙ (wkNeutral ρ n)
wkNeutral ρ (natrecₙ n)   = natrecₙ (wkNeutral ρ n)
wkNeutral ρ (prodrecₙ n)  = prodrecₙ (wkNeutral ρ n)
wkNeutral ρ (Emptyrecₙ e) = Emptyrecₙ (wkNeutral ρ e)

-- Weakening can be applied to our whnf views.

wkNatural :  ρ  Natural t  Natural {n = n} (wk ρ t)
wkNatural ρ sucₙ   = sucₙ
wkNatural ρ zeroₙ  = zeroₙ
wkNatural ρ (ne x) = ne (wkNeutral ρ x)

wkType :  ρ  Type t  Type {n = n} (wk ρ t)
wkType ρ Πₙ     = Πₙ
wkType ρ Σₙ     = Σₙ
wkType ρ ℕₙ     = ℕₙ
wkType ρ Emptyₙ = Emptyₙ
wkType ρ Unitₙ  = Unitₙ
wkType ρ (ne x) = ne (wkNeutral ρ x)

wkFunction :  ρ  Function t  Function {n = n} (wk ρ t)
wkFunction ρ lamₙ   = lamₙ
wkFunction ρ (ne x) = ne (wkNeutral ρ x)

wkProduct :  ρ  Product t  Product {n = n} (wk ρ t)
wkProduct ρ prodₙ  = prodₙ
wkProduct ρ (ne x) = ne (wkNeutral ρ x)

wkWhnf :  ρ  Whnf t  Whnf {n = n} (wk ρ t)
wkWhnf ρ Uₙ      = Uₙ
wkWhnf ρ Πₙ      = Πₙ
wkWhnf ρ Σₙ      = Σₙ
wkWhnf ρ ℕₙ      = ℕₙ
wkWhnf ρ Emptyₙ  = Emptyₙ
wkWhnf ρ Unitₙ   = Unitₙ
wkWhnf ρ lamₙ    = lamₙ
wkWhnf ρ prodₙ   = prodₙ
wkWhnf ρ zeroₙ   = zeroₙ
wkWhnf ρ sucₙ    = sucₙ
wkWhnf ρ starₙ   = starₙ
wkWhnf ρ (ne x)  = ne (wkNeutral ρ x)




------------------------------------------------------------------------
-- Substitution

-- The substitution operation  subst σ t  replaces the free de Bruijn indices
-- of term t by chosen terms as specified by σ.

-- The substitution σ itself is a map from natural numbers to terms.

Subst : Nat  Nat  Set a
Subst m n = Fin n  Term m

-- Given closed contexts ⊢ Γ and ⊢ Δ,
-- substitutions may be typed via Γ ⊢ σ : Δ meaning that
-- Γ ⊢ σ(x) : (subst σ Δ)(x) for all x ∈ dom(Δ).
--
-- The substitution operation is then typed as follows:
-- If Γ ⊢ σ : Δ and Δ ⊢ t : A, then Γ ⊢ subst σ t : subst σ A.
--
-- Although substitutions are untyped, typing helps us
-- to understand the operation on substitutions.

-- We may view σ as the infinite stream σ 0, σ 1, ...

-- Extract the substitution of the first variable.
--
-- If Γ ⊢ σ : Δ∙A  then Γ ⊢ head σ : subst σ A.

head : Subst m (1+ n)  Term m
head σ = σ x0

-- Remove the first variable instance of a substitution
-- and shift the rest to accommodate.
--
-- If Γ ⊢ σ : Δ∙A then Γ ⊢ tail σ : Δ.

tail : Subst m (1+ n)  Subst m n
tail σ x = σ (x +1)

-- Substitution of a variable.
--
-- If Γ ⊢ σ : Δ then Γ ⊢ substVar σ x : (subst σ Δ)(x).

substVar : (σ : Subst m n) (x : Fin n)  Term m
substVar σ x = σ x

-- Identity substitution.
-- Replaces each variable by itself.
--
-- Γ ⊢ idSubst : Γ.

idSubst : Subst n n
idSubst = var

-- Weaken a substitution by one.
--
-- If Γ ⊢ σ : Δ then Γ∙A ⊢ wk1Subst σ : Δ.

wk1Subst : Subst m n  Subst (1+ m) n
wk1Subst σ x = wk1 (σ x)

-- Lift a substitution.
--
-- If Γ ⊢ σ : Δ then Γ∙A ⊢ liftSubst σ : Δ∙A.

liftSubst : (σ : Subst m n)  Subst (1+ m) (1+ n)
liftSubst σ x0     = var x0
liftSubst σ (x +1) = wk1Subst σ x

liftSubstn : {k m : Nat}  Subst k m  (n : Nat)  Subst (n + k) (n + m)
liftSubstn σ Nat.zero = σ
liftSubstn σ (1+ n)   = liftSubst (liftSubstn σ n)

-- Transform a weakening into a substitution.
--
-- If ρ : Γ ≤ Δ then Γ ⊢ toSubst ρ : Δ.

toSubst :  Wk m n  Subst m n
toSubst pr x = var (wkVar pr x)

-- Apply a substitution to a term.
--
-- If Γ ⊢ σ : Δ and Δ ⊢ t : A then Γ ⊢ subst σ t : subst σ A.

mutual
  substGen : {bs : List Nat} (σ : Subst m n) (g : GenTs (Term) n bs)  GenTs (Term) m bs
  substGen σ  []      = []
  substGen σ (_∷_ {b = b} t ts) = subst (liftSubstn σ b) t  (substGen σ ts)

  subst : (σ : Subst m n) (t : Term n)  Term m
  subst σ (var x)   = substVar σ x
  subst σ (gen x c) = gen x (substGen σ c)

-- Extend a substitution by adding a term as
-- the first variable substitution and shift the rest.
--
-- If Γ ⊢ σ : Δ and Γ ⊢ t : subst σ A then Γ ⊢ consSubst σ t : Δ∙A.

consSubst : Subst m n  Term m  Subst m (1+ n)
consSubst σ t  x0    = t
consSubst σ t (x +1) = σ x

-- Singleton substitution.
--
-- If Γ ⊢ t : A then Γ ⊢ sgSubst t : Γ∙A.

sgSubst : Term n  Subst n (1+ n)
sgSubst = consSubst idSubst

-- Compose two substitutions.
--
-- If Γ ⊢ σ : Δ and Δ ⊢ σ′ : Φ then Γ ⊢ σ ₛ•ₛ σ′ : Φ.

_ₛ•ₛ_ : Subst  m  Subst m n  Subst  n
_ₛ•ₛ_ σ σ′ x = subst σ (σ′ x)

-- Composition of weakening and substitution.
--
--  If ρ : Γ ≤ Δ and Δ ⊢ σ : Φ then Γ ⊢ ρ •ₛ σ : Φ.

_•ₛ_ : Wk  m  Subst m n  Subst  n
_•ₛ_ ρ σ x = wk ρ (σ x)

--  If Γ ⊢ σ : Δ and ρ : Δ ≤ Φ then Γ ⊢ σ ₛ• ρ : Φ.

_ₛ•_ : Subst  m  Wk m n  Subst  n
_ₛ•_ σ ρ x = σ (wkVar ρ x)

-- Substitute the first variable of a term with an other term.
--
-- If Γ∙A ⊢ t : B and Γ ⊢ s : A then Γ ⊢ t[s] : B[s].

_[_] : (t : Term (1+ n)) (s : Term n)  Term n
t [ s ] = subst (sgSubst s) t

-- Substitute the first variable of a term with an other term,
-- but let the two terms share the same context.
--
-- If Γ∙A ⊢ t : B and Γ∙A ⊢ s : A then Γ∙A ⊢ t[s]↑ : B[s]↑.

_[_]↑ : (t : Term (1+ n)) (s : Term (1+ n))  Term (1+ n)
t [ s ]↑ = subst (consSubst (wk1Subst idSubst) s) t


-- Substitute the first two variables of a term with other terms.
--
-- If Γ∙A∙B ⊢ t : C, Γ ⊢ s : A and Γ ⊢ s′ : B and  then Γ ⊢ t[s,s′] : C[s,s′]

_[_,_] : (t : Term (1+ (1+ n))) (s s′ : Term n)  Term n
t [ s , s′ ] = subst (consSubst (consSubst idSubst s) s′) t

-- Substitute the first variable with a term and shift remaining variables up by one
-- If Γ ∙ A ⊢ t : A′ and Γ ∙ B ∙ C ⊢ s : A then Γ ∙ B ∙ C ⊢ t[s]↑² : A′

_[_]↑² : (t : Term (1+ n)) (s : Term (1+ (1+ n)))  Term (1+ (1+ n))
t [ s ]↑² = subst (consSubst (wk1Subst (wk1Subst idSubst)) s) t


B-subst : (σ : Subst m n) (W : BindingType) (F : Term n) (G : Term (1+ n))
         subst σ ( W  F  G) PE.≡  W  (subst σ F)  (subst (liftSubst σ) G)
B-subst σ ( p q) F G = PE.refl
B-subst σ ( q m) F G = PE.refl