Skip to content

Commit

Permalink
feat: toInt_abs
Browse files Browse the repository at this point in the history
We implement `toInt_abs`.
A subtle wrinkle is to note that `abs (intMin w) = intMin w`,
which complicates our proof.
  • Loading branch information
bollu committed Oct 21, 2024
1 parent b814be6 commit d3cf869
Show file tree
Hide file tree
Showing 3 changed files with 214 additions and 11 deletions.
194 changes: 183 additions & 11 deletions src/Init/Data/BitVec/Lemmas.lean
Original file line number Diff line number Diff line change
Expand Up @@ -206,6 +206,7 @@ theorem eq_of_getMsbD_eq {x y : BitVec w}
theorem of_length_zero {x : BitVec 0} : x = 0#0 := by ext; simp

theorem toNat_zero_length (x : BitVec 0) : x.toNat = 0 := by simp [of_length_zero]
theorem toInt_zero_length (x : BitVec 0) : x.toInt = 0 := by simp [of_length_zero]
theorem getLsbD_zero_length (x : BitVec 0) : x.getLsbD i = false := by simp
theorem getMsbD_zero_length (x : BitVec 0) : x.getMsbD i = false := by simp
theorem msb_zero_length (x : BitVec 0) : x.msb = false := by simp [BitVec.msb, of_length_zero]
Expand Down Expand Up @@ -353,7 +354,16 @@ theorem msb_eq_getLsbD_last (x : BitVec w) :
@[bv_toNat] theorem getLsbD_succ_last (x : BitVec (w + 1)) :
x.getLsbD w = decide (2 ^ w ≤ x.toNat) := getLsbD_last x

@[bv_toNat] theorem msb_eq_decide (x : BitVec w) : BitVec.msb x = decide (2 ^ (w-1) ≤ x.toNat) := by

/-- An alternative to `msb_eq_decide`-/
@[bv_toNat] theorem msb_eq_decide_le_mul_two (x : BitVec w) :
BitVec.msb x = decide (2 ^ w ≤ 2 * x.toNat) := by
rw [x.msb_eq_getLsbD_last, x.getLsbD_last]
simp
rcases w with rfl | w <;> simp <;> omega

@[bv_toNat, deprecated msb_eq_decide_le_mul_two (since := "21-10-2024") ]
theorem msb_eq_decide (x : BitVec w) : BitVec.msb x = decide (2 ^ (w-1) ≤ x.toNat) := by
simp [msb_eq_getLsbD_last, getLsbD_last]

theorem toNat_ge_of_msb_true {x : BitVec n} (p : BitVec.msb x = true) : x.toNat ≥ 2^(n-1) := by
Expand Down Expand Up @@ -2070,16 +2080,6 @@ theorem smod_zero {x : BitVec n} : x.smod 0#n = x := by
· simp
· by_cases h : x = 0#n <;> simp [h]

/-! ### abs -/

@[simp, bv_toNat]
theorem toNat_abs {x : BitVec w} : x.abs.toNat = if x.msb then 2^w - x.toNat else x.toNat := by
simp only [BitVec.abs, neg_eq]
by_cases h : x.msb = true
· simp only [h, ↓reduceIte, toNat_neg]
have : 2 * x.toNat ≥ 2 ^ w := BitVec.msb_eq_true_iff_two_mul_ge.mp h
rw [Nat.mod_eq_of_lt (by omega)]
· simp [h]

/-! ### mul -/

Expand Down Expand Up @@ -2643,6 +2643,23 @@ theorem toInt_neg_of_ne_intMin {x : BitVec w} (rs : x ≠ intMin w) :
have := @Nat.two_pow_pred_mul_two w (by omega)
split <;> split <;> omega


/-- The msb of `intMin w` is `true` for all `w > 0` -/
@[simp]
theorem msb_intMin : (intMin w).msb = decide (w > 0) := by
rw [intMin]
rw [msb_eq_decide]
simp
rcases w with rfl | w
· rfl
· simp
have : 0 < 2^w := Nat.pow_pos (by decide)
have : 2^w < 2^(w + 1) := by
rw [Nat.pow_succ]
omega
rw [Nat.mod_eq_of_lt (by omega)]
simp

/-! ### intMax -/

/-- The bitvector of width `w` that has the largest value when interpreted as an integer. -/
Expand Down Expand Up @@ -2674,6 +2691,161 @@ theorem getLsbD_intMax (w : Nat) : (intMax w).getLsbD i = decide (i + 1 < w) :=
· rw [Nat.sub_add_cancel (Nat.two_pow_pos (w - 1)), Nat.two_pow_pred_mod_two_pow (by omega)]


/-! ### abs -/

theorem abs_def {x : BitVec w} : x.abs = if x.msb then .neg x else x := rfl

/-
Similar to toInt_eq_toNat_cond, but rewrites in terms of power of two manipulations,
instead of ugly `2 * x < 2^n`.
-/
@[simp]
theorem toInt_eq_toNat_if (x : BitVec n) (hn : n > 0) :
x.toInt =
if x.toNat < 2^(n - 1) then
(x.toNat : Int)
else
(x.toNat : Int) - (2^n : Nat) := by
rcases n with _ | n <;> try contradiction
simp only [gt_iff_lt, Nat.zero_lt_succ, Nat.add_one_sub_one] at *
rw [BitVec.toInt_eq_toNat_cond]
simp only [Nat.pow_succ, Int.natCast_mul, Int.Nat.cast_ofNat_Int]
by_cases hx : x.toNat < 2 ^ n
· simp [show 2 * x.toNat < 2 ^ n * 2 by omega, hx]
· simp [show ¬ 2 * x.toNat < 2^ n * 2 by omega, hx]

/-
Next thing we want to know: bounds on the value of `x.toInt`
-/
theorem toInt_bounds_of_msb_eq_false {x : BitVec n} (hmsb : x.msb = false) :
0 ≤ x.toInt ∧ 2 * x.toInt < 2^n := by
have := x.msb_eq_decide_le_mul_two
rw [hmsb] at this
simp only [false_eq_decide_iff, Nat.not_le] at this
rw [BitVec.toInt_eq_toNat_cond]
simp [this]
apply And.intro
· omega
· norm_cast

theorem toInt_bounds_of_msb_eq_true {x : BitVec n} (hmsb : x.msb = true) :
-2^n ≤ x.toInt ∧ x.toInt < 0 := by
have := x.msb_eq_decide_le_mul_two
rw [hmsb] at this
simp only [true_eq_decide_iff] at this
rw [BitVec.toInt_eq_toNat_cond]
simp [show ¬ 2 * x.toNat < 2 ^ n by omega]
apply And.intro
· norm_cast
omega
· omega

/--
info: 'BitVec.toInt_bounds_of_msb_eq_true' depends on axioms: [propext, Classical.choice, Quot.sound]
-/
#guard_msgs in #print axioms BitVec.toInt_bounds_of_msb_eq_true

theorem toInt_intMin_eq_cases (n : Nat) : (BitVec.intMin n).toInt =
if n = 0 then 0 else - 2^(n - 1) := by
simp [BitVec.toInt_intMin]
rcases n with rfl | n
· simp
· simp
norm_cast
have : 2^n > 0 := by exact Nat.two_pow_pos n
have : 2^n < 2^(n + 1) := by
simp [Nat.pow_succ]
omega
rw [Nat.mod_eq_of_lt (by omega)]

/-- info: 'BitVec.toInt_intMin_eq_cases' depends on axioms: [propext, Quot.sound] -/
#guard_msgs in #print axioms BitVec.toInt_intMin_eq_cases

-- ### TOINT OF NEG
/--
Define the value of (BitVec.neg.toInt) as a case split
on whether `x` is intMin or not, and showing that when this
exception does not occur, the defn obeys what mathematics says it should
-/
theorem toInt_neg_eq_cases {x : BitVec n} :
(-x).toInt =
if x = intMin n
then x.toInt
else - x.toInt := by
by_cases hx : x = intMin n
· simp [hx]
· simp [hx]
rw [toInt_neg_of_ne_intMin hx]

/-- info: 'BitVec.toInt_neg_eq_cases' depends on axioms: [propext, Quot.sound] -/
#guard_msgs in #print axioms BitVec.toInt_neg_eq_cases

-- @[simp]
-- theorem Int.abs_neg (x : Int) : (-x).abs = x.abs := by
-- have hx : (-x < 0) ∨ (x = 0) ∨ (-x > 0) := by omega
-- rcases hx with hx | hx | hx
-- · rw [Int.abs_eq_neg (x := -x) hx, Int.neg_neg, Int.abs_eq_self (x := x) (by omega)]
-- · simp [hx]
-- · rw [Int.abs_eq_self (x := -x) (by omega), Int.abs_eq_neg (x := x) (by omega)]
--
--
-- /-- info: 'Int.abs_neg' depends on axioms: [propext, Quot.sound] -/
-- #guard_msgs in #print axioms Int.abs_neg


theorem abs_cases' (x : BitVec w) : x.abs =
if x.msb = true then
if x = BitVec.intMin w then (BitVec.intMin w) else -x
else x := by
· rw [BitVec.abs_def]
by_cases hx : x.msb = true <;> by_cases hx' : x = BitVec.intMin w <;> simp [hx, hx']

/-- info: 'BitVec.abs_cases'' depends on axioms: [propext, Quot.sound] -/
#guard_msgs in #print axioms BitVec.abs_cases'



theorem toInt_intMin_eq_twoPow (hn : 0 < n) : (intMin n).toInt = -2^(n - 1) := by
-- Delete our toInt_intMin from simp set.
rw [BitVec.toInt_intMin_eq_cases]
simp [show ¬ n = 0 by omega]

/-- info: 'BitVec.toInt_intMin_eq_twoPow' depends on axioms: [propext, Quot.sound] -/
#guard_msgs in #print axioms BitVec.toInt_intMin_eq_twoPow

theorem toInt_abs (x : BitVec w) :
x.abs.toInt = if x = (intMin w) then if w = 0 then 0 else - 2^(w - 1) else x.toInt.abs := by
rcases w with rfl | w
· simp [toInt_zero_length]
· simp only [gt_iff_lt, Nat.zero_lt_succ, Nat.add_one_ne_zero, ↓reduceIte]
rw [BitVec.abs_cases']
by_cases hx : x = intMin (w + 1)
· simp only [hx, reduceIte]
have := BitVec.msb_intMin (w := w + 1)
rw [this]
simp only [gt_iff_lt, Nat.zero_lt_succ, decide_True, ↓reduceIte]
rw [BitVec.toInt_intMin_eq_cases]
simp
· simp only [hx, reduceIte]
rcases hmsb : x.msb
· simp only [Bool.false_eq_true, ↓reduceIte]
have := BitVec.toInt_bounds_of_msb_eq_false hmsb
rw [Int.abs_eq_self]
omega
· simp only [reduceIte]
have hxbounds := BitVec.toInt_bounds_of_msb_eq_true hmsb
rw [BitVec.toInt_neg_eq_cases]
have hxneq : x.toInt ≠ (intMin (w + 1)).toInt := by
rw [BitVec.toInt_ne]
exact hx
rw [BitVec.toInt_intMin_eq_twoPow (by omega)] at hxneq
-- TODO: remove toInt_eq_toNat_cond from simp set.
simp only [hx, reduceIte]
rw [Int.abs_eq_neg (by omega)]

/-- info: 'BitVec.toInt_abs' depends on axioms: [propext, Classical.choice, Quot.sound] -/
#guard_msgs in #print axioms BitVec.toInt_abs

/-! ### Non-overflow theorems -/

/-- If `x.toNat * y.toNat < 2^w`, then the multiplication `(x * y)` does not overflow. -/
Expand Down
7 changes: 7 additions & 0 deletions src/Init/Data/Int/Basic.lean
Original file line number Diff line number Diff line change
Expand Up @@ -333,6 +333,13 @@ instance : Min Int := minOfLe

instance : Max Int := maxOfLe

/--
Return the absolute value of an integer.
-/
def abs : Int → Int
| ofNat n => .ofNat n
| negSucc n => .ofNat n.succ

end Int

/--
Expand Down
24 changes: 24 additions & 0 deletions src/Init/Data/Int/Lemmas.lean
Original file line number Diff line number Diff line change
Expand Up @@ -531,4 +531,28 @@ theorem natCast_one : ((1 : Nat) : Int) = (1 : Int) := rfl
@[simp] theorem natCast_mul (a b : Nat) : ((a * b : Nat) : Int) = (a : Int) * (b : Int) := by
simp

/-! abs lemmas -/

@[simp]
theorem abs_eq_self {x : Int} (h : x ≥ 0) : x.abs = x := by
cases x
case ofNat h =>
rfl
case negSucc h =>
contradiction

@[simp]
theorem Int.abs_zero : Int.abs 0 = 0 := rfl

@[simp]
theorem abs_eq_neg {x : Int} (h : x < 0) : x.abs = -x := by
cases x
case ofNat h =>
contradiction
case negSucc n =>
rfl

@[simp]
theorem ofNat_abs (x : Nat) : (x : Int).abs = (x : Int) := rfl

end Int

0 comments on commit d3cf869

Please sign in to comment.