a82f0d9413
This changes how Nat typeclass checks in offset terms from syntactic equality to definitional equality with "instances" transparency. This may have a negative performance penalty in `isOffset?`, but it should be small in common cases since the relevant instances are small terms. This closes #3836
127 lines
4 KiB
Text
127 lines
4 KiB
Text
variable (a b : Nat)
|
|
|
|
/- subtract diff tests -/
|
|
|
|
#check_simp (1000 : Nat) - 400 ~> 600
|
|
|
|
#check_simp (a + 1000) - 1000 ~> a
|
|
#check_simp (a + 400) - 1000 ~> a - 600
|
|
#check_simp (a + 1000) - 400 ~> a + 600
|
|
|
|
#check_simp 1000 - (a + 400) ~> 600 - a
|
|
#check_simp 400 - (a + 1000) ~> 0
|
|
|
|
#check_simp (a + 1000) - (b + 1000) ~> a - b
|
|
#check_simp (a + 1000) - (b + 400) ~> a + 600 - b
|
|
#check_simp (a + 400) - (b + 1000) ~> a - (b + 600)
|
|
|
|
/- eq tests -/
|
|
|
|
#check_simp (1234567 : Nat) = 123456 ~> False
|
|
#check_simp (1234567 : Nat) = 1234567 ~> True
|
|
#check_simp (123456 : Nat) = 1234567 ~> False
|
|
|
|
#check_simp (a + 1000) = 1000 ~> a = 0
|
|
#check_simp (a + 1000) = 400 ~> False
|
|
#check_simp (a + 400) = 1000 ~> a = 600
|
|
|
|
#check_simp 1000 = (a + 1000) ~> a = 0
|
|
#check_simp 400 = (a + 1000) ~> False
|
|
#check_simp 1000 = (a + 400) ~> a = 600
|
|
|
|
#check_simp (a + 1000) = (b + 1000) ~> a = b
|
|
#check_simp (a + 1000) = (b + 400) ~> a + 600 = b
|
|
#check_simp (a + 400) = (b + 1000) ~> a = b + 600
|
|
#check_simp (Nat.add a 400) = (Add.add b 1000) ~> a = b + 600
|
|
#check_simp (Nat.add a 400) = b.succ.succ ~> a + 398 = b
|
|
|
|
/- ne -/
|
|
|
|
#check_simp 1000 ≠ (a + 1000) ~> a ≠ 0
|
|
#check_simp (1234567 : Nat) ≠ 123456 ~> True
|
|
#check_simp (a + 400) ≠ (b + 1000) ~> a ≠ b + 600
|
|
|
|
/- leq -/
|
|
|
|
#check_simp (1234567 : Nat) ≤ 123456 ~> False
|
|
#check_simp (1234567 : Nat) ≤ 1234567 ~> True
|
|
#check_simp (123456 : Nat) ≤ 1234567 ~> True
|
|
|
|
#check_simp (a + 1000) ≤ 1000 ~> a = 0
|
|
#check_simp (a + 1000) ≤ 400 ~> False
|
|
#check_simp (a + 400) ≤ 1000 ~> a ≤ 600
|
|
|
|
#check_simp 1000 ≤ (a + 1000) ~> True
|
|
#check_simp 400 ≤ (a + 1000) ~> True
|
|
#check_simp 1000 ≤ (a + 400) ~> 600 ≤ a
|
|
|
|
#check_simp (a + 1000) ≤ (b + 1000) ~> a ≤ b
|
|
#check_simp (a + 1000) ≤ (b + 400) ~> a + 600 ≤ b
|
|
#check_simp (a + 400) ≤ (b + 1000) ~> a ≤ b + 600
|
|
#check_simp (Nat.add a 400) ≤ (Add.add b 1000) ~> a ≤ b + 600
|
|
#check_simp (Nat.add a 400) ≤ b.succ.succ ~> a + 398 ≤ b
|
|
|
|
/- ge (just make sure le rules apply) -/
|
|
|
|
#check_simp (123456 : Nat) ≥ 1234567 ~> False
|
|
#check_simp (a + 400) ≥ 1000 ~> a ≥ 600
|
|
#check_simp 1000 ≥ (a + 1000) ~> a = 0
|
|
#check_simp (a + 1000) ≥ (b + 400) ~> a + 600 ≥ b
|
|
|
|
/- beq tests -/
|
|
|
|
#check_simp (1234567 : Nat) == 123456 ~> false
|
|
#check_simp (1234567 : Nat) == 1234567 ~> true
|
|
#check_simp (123456 : Nat) == 1234567 ~> false
|
|
|
|
#check_simp (a + 1000) == 1000 ~> a == 0
|
|
#check_simp (a + 1000) == 400 ~> false
|
|
#check_simp (a + 400) == 1000 ~> a == 600
|
|
|
|
#check_simp 1000 == (a + 1000) ~> a == 0
|
|
#check_simp 400 == (a + 1000) ~> false
|
|
#check_simp 1000 == (a + 400) ~> a == 600
|
|
|
|
#check_simp (a + 1000) == (b + 1000) ~> a == b
|
|
#check_simp (a + 1000) == (b + 400) ~> a + 600 == b
|
|
#check_simp (a + 400) == (b + 1000) ~> a == b + 600
|
|
|
|
/- bne tests -/
|
|
|
|
#check_simp (1234567 : Nat) != 123456 ~> true
|
|
|
|
#check_simp (a + 1000) != 1000 ~> a != 0
|
|
#check_simp (a + 1000) != 400 ~> true
|
|
#check_simp (a + 400) != 1000 ~> a != 600
|
|
|
|
#check_simp 1000 != (a + 1000) ~> a != 0
|
|
#check_simp 400 != (a + 1000) ~> true
|
|
#check_simp 1000 != (a + 400) ~> a != 600
|
|
|
|
#check_simp (a + 1000) != (b + 1000) ~> a != b
|
|
#check_simp (a + 1000) != (b + 400) ~> a + 600 != b
|
|
#check_simp (a + 400) != (b + 1000) ~> a != b + 600
|
|
|
|
/-! Alterate instance tests
|
|
|
|
These check that the simplification rules will matching
|
|
offets still trigger even when the expression for the
|
|
index is definition equal but not syntactically equal
|
|
to the defualt instance.
|
|
|
|
This can be relevant in Mathlib when rewriting using
|
|
theorems involving algebraic hierarchy classes.
|
|
-/
|
|
|
|
class AddCommMagma (G : Type u) extends Add G where
|
|
add_comm : ∀(x y : G), x + y = y + x
|
|
|
|
instance instAddExtNat : AddCommMagma Nat where
|
|
add_comm := Nat.add_comm
|
|
|
|
#check_tactic @Add.add _ instAddExtNat.toAdd a 1 = 4 ~> a = 3 by simp only [Nat.succ.injEq]
|
|
#check_tactic @HAdd.hAdd _ _ _ (@instHAdd _ instAddExtNat.toAdd) a 1 = 4 ~> a = 3 by simp only [Nat.succ.injEq]
|
|
|
|
#check_tactic @Add.add _ instAddNat a 1 = 4 ~> a = 3 by simp
|
|
#check_tactic @Add.add _ instAddExtNat.toAdd a 1 = 4 ~> a = 3 by simp
|
|
#check_tactic @HAdd.hAdd _ _ _ (@instHAdd _ instAddExtNat.toAdd) a 1 = 4 ~> a = 3 by simp
|