lean4-htt/src/Init/Data/BitVec/Bitblast.lean
Siddharth 4415a81f35
chore: add Siddharth to authors list of BitVec (#5647)
Add @bollu to the authors list of the BitVec files :)
2024-10-08 08:01:17 +00:00

1033 lines
40 KiB
Text
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

/-
Copyright (c) 2024 Lean FRO, LLC. All rights reserved.
Released under Apache 2.0 license as described in the file LICENSE.
Authors: Harun Khan, Abdalrhman M Mohamed, Joe Hendrix, Siddharth Bhat
-/
prelude
import Init.Data.BitVec.Folds
import Init.Data.Nat.Mod
/-!
# Bitblasting of bitvectors
This module provides theorems for showing the equivalence between BitVec operations using
the `Fin 2^n` representation and Boolean vectors. It is still under development, but
intended to provide a path for converting SAT and SMT solver proofs about BitVectors
as vectors of bits into proofs about Lean `BitVec` values.
The module is named for the bit-blasting operation in an SMT solver that converts bitvector
expressions into expressions about individual bits in each vector.
### Example: How bitblasting works for multiplication
We explain how the lemmas here are used for bitblasting,
by using multiplication as a prototypical example.
Other bitblasters for other operations follow the same pattern.
To bitblast a multiplication of the form `x * y`,
we must unfold the above into a form that the SAT solver understands.
We assume that the solver already knows how to bitblast addition.
This is known to `bv_decide`, by exploiting the lemma `add_eq_adc`,
which says that `x + y : BitVec w` equals `(adc x y false).2`,
where `adc` builds an add-carry circuit in terms of the primitive operations
(bitwise and, bitwise or, bitwise xor) that bv_decide already understands.
In this way, we layer bitblasters on top of each other,
by reducing the multiplication bitblaster to an addition operation.
The core lemma is given by `getLsbD_mul`:
```lean
x y : BitVec w ⊢ (x * y).getLsbD i = (mulRec x y w).getLsbD i
```
Which says that the `i`th bit of `x * y` can be obtained by
evaluating the `i`th bit of `(mulRec x y w)`.
Once again, we assume that `bv_decide` knows how to implement `getLsbD`,
given that `mulRec` can be understood by `bv_decide`.
We write two lemmas to enable `bv_decide` to unfold `(mulRec x y w)`
into a complete circuit, **when `w` is a known constant**`.
This is given by two recurrence lemmas, `mulRec_zero_eq` and `mulRec_succ_eq`,
which are applied repeatedly when the width is `0` and when the width is `w' + 1`:
```lean
mulRec_zero_eq :
mulRec x y 0 =
if y.getLsbD 0 then x else 0
mulRec_succ_eq
mulRec x y (s + 1) =
mulRec x y s +
if y.getLsbD (s + 1) then (x <<< (s + 1)) else 0 := rfl
```
By repeatedly applying the lemmas `mulRec_zero_eq` and `mulRec_succ_eq`,
one obtains a circuit for multiplication.
Note that this circuit uses `BitVec.add`, `BitVec.getLsbD`, `BitVec.shiftLeft`.
Here, `BitVec.add` and `BitVec.shiftLeft` are (recursively) bitblasted by `bv_decide`,
using the lemmas `add_eq_adc` and `shiftLeft_eq_shiftLeftRec`,
and `BitVec.getLsbD` is a primitive that `bv_decide` knows how to reduce to SAT.
The two lemmas, `mulRec_zero_eq`, and `mulRec_succ_eq`,
are used in `Std.Tactic.BVDecide.BVExpr.bitblast.blastMul`
to prove the correctness of the circuit that is built by `bv_decide`.
```lean
def blastMul (aig : AIG BVBit) (input : AIG.BinaryRefVec aig w) : AIG.RefVecEntry BVBit w
theorem denote_blastMul (aig : AIG BVBit) (lhs rhs : BitVec w) (assign : Assignment) :
...
⟦(blastMul aig input).aig, (blastMul aig input).vec.get idx hidx, assign.toAIGAssignment⟧
=
(lhs * rhs).getLsbD idx
```
The definition and theorem above are internal to `bv_decide`,
and use `mulRec_{zero,succ}_eq` to prove that the circuit built by `bv_decide`
computes the correct value for multiplication.
To zoom out, therefore, we follow two steps:
First, we prove bitvector lemmas to unfold a high-level operation (such as multiplication)
into already bitblastable operations (such as addition and left shift).
We then use these lemmas to prove the correctness of the circuit that `bv_decide` builds.
We use this workflow to implement bitblasting for all SMT-LIB2 operations.
## Main results
* `x + y : BitVec w` is `(adc x y false).2`.
## Future work
All other operations are to be PR'ed later and are already proved in
https://github.com/mhk119/lean-smt/blob/bitvec/Smt/Data/Bitwise.lean.
-/
set_option linter.missingDocs true
open Nat Bool
namespace Bool
/-- At least two out of three booleans are true. -/
abbrev atLeastTwo (a b c : Bool) : Bool := a && b || a && c || b && c
@[simp] theorem atLeastTwo_false_left : atLeastTwo false b c = (b && c) := by simp [atLeastTwo]
@[simp] theorem atLeastTwo_false_mid : atLeastTwo a false c = (a && c) := by simp [atLeastTwo]
@[simp] theorem atLeastTwo_false_right : atLeastTwo a b false = (a && b) := by simp [atLeastTwo]
@[simp] theorem atLeastTwo_true_left : atLeastTwo true b c = (b || c) := by cases b <;> cases c <;> simp [atLeastTwo]
@[simp] theorem atLeastTwo_true_mid : atLeastTwo a true c = (a || c) := by cases a <;> cases c <;> simp [atLeastTwo]
@[simp] theorem atLeastTwo_true_right : atLeastTwo a b true = (a || b) := by cases a <;> cases b <;> simp [atLeastTwo]
end Bool
/-! ### Preliminaries -/
namespace BitVec
private theorem testBit_limit {x i : Nat} (x_lt_succ : x < 2^(i+1)) :
testBit x i = decide (x ≥ 2^i) := by
cases xi : testBit x i with
| true =>
simp [testBit_implies_ge xi]
| false =>
simp
cases Nat.lt_or_ge x (2^i) with
| inl x_lt =>
exact x_lt
| inr x_ge =>
have ⟨j, ⟨j_ge, jp⟩⟩ := ge_two_pow_implies_high_bit_true x_ge
cases Nat.lt_or_eq_of_le j_ge with
| inr x_eq =>
simp [x_eq, jp] at xi
| inl x_lt =>
exfalso
apply Nat.lt_irrefl
calc x < 2^(i+1) := x_lt_succ
_ ≤ 2 ^ j := Nat.pow_le_pow_of_le_right Nat.zero_lt_two x_lt
_ ≤ x := testBit_implies_ge jp
private theorem mod_two_pow_succ (x i : Nat) :
x % 2^(i+1) = 2^i*(x.testBit i).toNat + x % (2 ^ i):= by
rw [Nat.mod_pow_succ, Nat.add_comm, Nat.toNat_testBit]
private theorem mod_two_pow_add_mod_two_pow_add_bool_lt_two_pow_succ
(x y i : Nat) (c : Bool) : x % 2^i + (y % 2^i + c.toNat) < 2^(i+1) := by
have : c.toNat ≤ 1 := Bool.toNat_le c
rw [Nat.pow_succ]
omega
/-! ### Addition -/
/-- carry i x y c returns true if the `i` carry bit is true when computing `x + y + c`. -/
def carry (i : Nat) (x y : BitVec w) (c : Bool) : Bool :=
decide (x.toNat % 2^i + y.toNat % 2^i + c.toNat ≥ 2^i)
@[simp] theorem carry_zero : carry 0 x y c = c := by
cases c <;> simp [carry, mod_one]
theorem carry_succ (i : Nat) (x y : BitVec w) (c : Bool) :
carry (i+1) x y c = atLeastTwo (x.getLsbD i) (y.getLsbD i) (carry i x y c) := by
simp only [carry, mod_two_pow_succ, atLeastTwo, getLsbD]
simp only [Nat.pow_succ']
have sum_bnd : x.toNat%2^i + (y.toNat%2^i + c.toNat) < 2*2^i := by
simp only [← Nat.pow_succ']
exact mod_two_pow_add_mod_two_pow_add_bool_lt_two_pow_succ ..
cases x.toNat.testBit i <;> cases y.toNat.testBit i <;> (simp; omega)
/--
If `x &&& y = 0`, then the carry bit `(x + y + 0)` is always `false` for any index `i`.
Intuitively, this is because a carry is only produced when at least two of `x`, `y`, and the
previous carry are true. However, since `x &&& y = 0`, at most one of `x, y` can be true,
and thus we never have a previous carry, which means that the sum cannot produce a carry.
-/
theorem carry_of_and_eq_zero {x y : BitVec w} (h : x &&& y = 0#w) : carry i x y false = false := by
induction i with
| zero => simp
| succ i ih =>
replace h := congrArg (·.getLsbD i) h
simp_all [carry_succ]
/-- The final carry bit when computing `x + y + c` is `true` iff `x.toNat + y.toNat + c.toNat ≥ 2^w`. -/
theorem carry_width {x y : BitVec w} :
carry w x y c = decide (x.toNat + y.toNat + c.toNat ≥ 2^w) := by
simp [carry]
/--
If `x &&& y = 0`, then addition does not overflow, and thus `(x + y).toNat = x.toNat + y.toNat`.
-/
theorem toNat_add_of_and_eq_zero {x y : BitVec w} (h : x &&& y = 0#w) :
(x + y).toNat = x.toNat + y.toNat := by
rw [toNat_add]
apply Nat.mod_eq_of_lt
suffices ¬ decide (x.toNat + y.toNat + false.toNat ≥ 2^w) by
simp only [decide_eq_true_eq] at this
omega
rw [← carry_width]
simp [not_eq_true, carry_of_and_eq_zero h]
/-- Carry function for bitwise addition. -/
def adcb (x y c : Bool) : Bool × Bool := (atLeastTwo x y c, x ^^ (y ^^ c))
/-- Bitwise addition implemented via a ripple carry adder. -/
def adc (x y : BitVec w) : Bool → Bool × BitVec w :=
iunfoldr fun (i : Fin w) c => adcb (x.getLsbD i) (y.getLsbD i) c
theorem getLsbD_add_add_bool {i : Nat} (i_lt : i < w) (x y : BitVec w) (c : Bool) :
getLsbD (x + y + setWidth w (ofBool c)) i =
(getLsbD x i ^^ (getLsbD y i ^^ carry i x y c)) := by
let ⟨x, x_lt⟩ := x
let ⟨y, y_lt⟩ := y
simp only [getLsbD, toNat_add, toNat_setWidth, i_lt, toNat_ofFin, toNat_ofBool,
Nat.mod_add_mod, Nat.add_mod_mod]
apply Eq.trans
rw [← Nat.div_add_mod x (2^i), ← Nat.div_add_mod y (2^i)]
simp only
[ Nat.testBit_mod_two_pow,
Nat.testBit_mul_two_pow_add_eq,
i_lt,
decide_True,
Bool.true_and,
Nat.add_assoc,
Nat.add_left_comm (_%_) (_ * _) _,
testBit_limit (mod_two_pow_add_mod_two_pow_add_bool_lt_two_pow_succ x y i c)
]
simp [testBit_to_div_mod, carry, Nat.add_assoc]
theorem getLsbD_add {i : Nat} (i_lt : i < w) (x y : BitVec w) :
getLsbD (x + y) i =
(getLsbD x i ^^ (getLsbD y i ^^ carry i x y false)) := by
simpa using getLsbD_add_add_bool i_lt x y false
theorem getElem_add_add_bool {i : Nat} (i_lt : i < w) (x y : BitVec w) (c : Bool) :
(x + y + setWidth w (ofBool c))[i] =
(x[i] ^^ (y[i] ^^ carry i x y c)) := by
simp only [← getLsbD_eq_getElem]
rw [getLsbD_add_add_bool]
omega
theorem getElem_add {i : Nat} (i_lt : i < w) (x y : BitVec w) :
(x + y)[i] = (x[i] ^^ (y[i] ^^ carry i x y false)) := by
simpa using getElem_add_add_bool i_lt x y false
theorem adc_spec (x y : BitVec w) (c : Bool) :
adc x y c = (carry w x y c, x + y + setWidth w (ofBool c)) := by
simp only [adc]
apply iunfoldr_replace
(fun i => carry i x y c)
(x + y + setWidth w (ofBool c))
c
case init =>
simp [carry, Nat.mod_one]
cases c <;> rfl
case step =>
simp [adcb, Prod.mk.injEq, carry_succ, getLsbD_add_add_bool]
theorem add_eq_adc (w : Nat) (x y : BitVec w) : x + y = (adc x y false).snd := by
simp [adc_spec]
/-! ### add -/
/-- Adding a bitvector to its own complement yields the all ones bitpattern -/
@[simp] theorem add_not_self (x : BitVec w) : x + ~~~x = allOnes w := by
rw [add_eq_adc, adc, iunfoldr_replace (fun _ => false) (allOnes w)]
· rfl
· simp [adcb, atLeastTwo]
/-- Subtracting `x` from the all ones bitvector is equivalent to taking its complement -/
theorem allOnes_sub_eq_not (x : BitVec w) : allOnes w - x = ~~~x := by
rw [← add_not_self x, BitVec.add_comm, add_sub_cancel]
/-- Addition of bitvectors is the same as bitwise or, if bitwise and is zero. -/
theorem add_eq_or_of_and_eq_zero {w : Nat} (x y : BitVec w)
(h : x &&& y = 0#w) : x + y = x ||| y := by
rw [add_eq_adc, adc, iunfoldr_replace (fun _ => false) (x ||| y)]
· rfl
· simp only [adcb, atLeastTwo, Bool.and_false, Bool.or_false, bne_false, getLsbD_or,
Prod.mk.injEq, and_eq_false_imp]
intros i
replace h : (x &&& y).getLsbD i = (0#w).getLsbD i := by rw [h]
simp only [getLsbD_and, getLsbD_zero, and_eq_false_imp] at h
constructor
· intros hx
simp_all [hx]
· by_cases hx : x.getLsbD i <;> simp_all [hx]
/-! ### Negation -/
theorem bit_not_testBit (x : BitVec w) (i : Fin w) :
getLsbD (((iunfoldr (fun (i : Fin w) c => (c, !(x.getLsbD i)))) ()).snd) i.val = !(getLsbD x i.val) := by
apply iunfoldr_getLsbD (fun _ => ()) i (by simp)
theorem bit_not_add_self (x : BitVec w) :
((iunfoldr (fun (i : Fin w) c => (c, !(x.getLsbD i)))) ()).snd + x = -1 := by
simp only [add_eq_adc]
apply iunfoldr_replace_snd (fun _ => false) (-1) false rfl
intro i; simp only [ BitVec.not, adcb, testBit_toNat]
rw [iunfoldr_replace_snd (fun _ => ()) (((iunfoldr (fun i c => (c, !(x.getLsbD i)))) ()).snd)]
<;> simp [bit_not_testBit, negOne_eq_allOnes, getLsbD_allOnes]
theorem bit_not_eq_not (x : BitVec w) :
((iunfoldr (fun i c => (c, !(x.getLsbD i)))) ()).snd = ~~~ x := by
simp [←allOnes_sub_eq_not, BitVec.eq_sub_iff_add_eq.mpr (bit_not_add_self x), ←negOne_eq_allOnes]
theorem bit_neg_eq_neg (x : BitVec w) : -x = (adc (((iunfoldr (fun (i : Fin w) c => (c, !(x.getLsbD i)))) ()).snd) (BitVec.ofNat w 1) false).snd:= by
simp only [← add_eq_adc]
rw [iunfoldr_replace_snd ((fun _ => ())) (((iunfoldr (fun (i : Fin w) c => (c, !(x.getLsbD i)))) ()).snd) _ rfl]
· rw [BitVec.eq_sub_iff_add_eq.mpr (bit_not_add_self x), sub_toAdd, BitVec.add_comm _ (-x)]
simp [← sub_toAdd, BitVec.sub_add_cancel]
· simp [bit_not_testBit x _]
/-! ### Inequalities (le / lt) -/
theorem ult_eq_not_carry (x y : BitVec w) : x.ult y = !carry w x (~~~y) true := by
simp only [BitVec.ult, carry, toNat_mod_cancel, toNat_not, toNat_true, ge_iff_le, ← decide_not,
Nat.not_le, decide_eq_decide]
rw [Nat.mod_eq_of_lt (by omega)]
omega
theorem ule_eq_not_ult (x y : BitVec w) : x.ule y = !y.ult x := by
simp [BitVec.ule, BitVec.ult, ← decide_not]
theorem ule_eq_carry (x y : BitVec w) : x.ule y = carry w y (~~~x) true := by
simp [ule_eq_not_ult, ult_eq_not_carry]
/-- If two bitvectors have the same `msb`, then signed and unsigned comparisons coincide -/
theorem slt_eq_ult_of_msb_eq {x y : BitVec w} (h : x.msb = y.msb) :
x.slt y = x.ult y := by
simp only [BitVec.slt, toInt_eq_msb_cond, BitVec.ult, decide_eq_decide, h]
cases y.msb <;> simp
/-- If two bitvectors have different `msb`s, then unsigned comparison is determined by this bit -/
theorem ult_eq_msb_of_msb_neq {x y : BitVec w} (h : x.msb ≠ y.msb) :
x.ult y = y.msb := by
simp only [BitVec.ult, msb_eq_decide, ne_eq, decide_eq_decide] at *
omega
/-- If two bitvectors have different `msb`s, then signed and unsigned comparisons are opposites -/
theorem slt_eq_not_ult_of_msb_neq {x y : BitVec w} (h : x.msb ≠ y.msb) :
x.slt y = !x.ult y := by
simp only [BitVec.slt, toInt_eq_msb_cond, Bool.eq_not_of_ne h, ult_eq_msb_of_msb_neq h]
cases y.msb <;> (simp; omega)
theorem slt_eq_ult (x y : BitVec w) :
x.slt y = (x.msb != y.msb).xor (x.ult y) := by
by_cases h : x.msb = y.msb
· simp [h, slt_eq_ult_of_msb_eq]
· have h' : x.msb != y.msb := by simp_all
simp [slt_eq_not_ult_of_msb_neq h, h']
theorem slt_eq_not_carry (x y : BitVec w) :
x.slt y = (x.msb == y.msb).xor (carry w x (~~~y) true) := by
simp only [slt_eq_ult, bne, ult_eq_not_carry]
cases x.msb == y.msb <;> simp
theorem sle_eq_not_slt (x y : BitVec w) : x.sle y = !y.slt x := by
simp only [BitVec.sle, BitVec.slt, ← decide_not, decide_eq_decide]; omega
theorem sle_eq_carry (x y : BitVec w) :
x.sle y = !((x.msb == y.msb).xor (carry w y (~~~x) true)) := by
rw [sle_eq_not_slt, slt_eq_not_carry, beq_comm]
/-! ### mul recurrence for bitblasting -/
/--
A recurrence that describes multiplication as repeated addition.
Is useful for bitblasting multiplication.
-/
def mulRec (x y : BitVec w) (s : Nat) : BitVec w :=
let cur := if y.getLsbD s then (x <<< s) else 0
match s with
| 0 => cur
| s + 1 => mulRec x y s + cur
theorem mulRec_zero_eq (x y : BitVec w) :
mulRec x y 0 = if y.getLsbD 0 then x else 0 := by
simp [mulRec]
theorem mulRec_succ_eq (x y : BitVec w) (s : Nat) :
mulRec x y (s + 1) = mulRec x y s + if y.getLsbD (s + 1) then (x <<< (s + 1)) else 0 := rfl
/--
Recurrence lemma: truncating to `i+1` bits and then zero extending to `w`
equals truncating upto `i` bits `[0..i-1]`, and then adding the `i`th bit of `x`.
-/
theorem setWidth_setWidth_succ_eq_setWidth_setWidth_add_twoPow (x : BitVec w) (i : Nat) :
setWidth w (x.setWidth (i + 1)) =
setWidth w (x.setWidth i) + (x &&& twoPow w i) := by
rw [add_eq_or_of_and_eq_zero]
· ext k
simp only [getLsbD_setWidth, Fin.is_lt, decide_True, Bool.true_and, getLsbD_or, getLsbD_and]
by_cases hik : i = k
· subst hik
simp
· simp only [getLsbD_twoPow, hik, decide_False, Bool.and_false, Bool.or_false]
by_cases hik' : k < (i + 1)
· have hik'' : k < i := by omega
simp [hik', hik'']
· have hik'' : ¬ (k < i) := by omega
simp [hik', hik'']
· ext k
simp only [and_twoPow, getLsbD_and, getLsbD_setWidth, Fin.is_lt, decide_True, Bool.true_and,
getLsbD_zero, and_eq_false_imp, and_eq_true, decide_eq_true_eq, and_imp]
by_cases hi : x.getLsbD i <;> simp [hi] <;> omega
@[deprecated setWidth_setWidth_succ_eq_setWidth_setWidth_add_twoPow (since := "2024-09-18"),
inherit_doc setWidth_setWidth_succ_eq_setWidth_setWidth_add_twoPow]
abbrev zeroExtend_truncate_succ_eq_zeroExtend_truncate_add_twoPow :=
@setWidth_setWidth_succ_eq_setWidth_setWidth_add_twoPow
/--
Recurrence lemma: multiplying `x` with the first `s` bits of `y` is the
same as truncating `y` to `s` bits, then zero extending to the original length,
and performing the multplication. -/
theorem mulRec_eq_mul_signExtend_setWidth (x y : BitVec w) (s : Nat) :
mulRec x y s = x * ((y.setWidth (s + 1)).setWidth w) := by
induction s
case zero =>
simp only [mulRec_zero_eq, ofNat_eq_ofNat, Nat.reduceAdd]
by_cases y.getLsbD 0
case pos hy =>
simp only [hy, ↓reduceIte, setWidth_one_eq_ofBool_getLsb_zero,
ofBool_true, ofNat_eq_ofNat]
rw [setWidth_ofNat_one_eq_ofNat_one_of_lt (by omega)]
simp
case neg hy =>
simp [hy, setWidth_one_eq_ofBool_getLsb_zero]
case succ s' hs =>
rw [mulRec_succ_eq, hs]
have heq :
(if y.getLsbD (s' + 1) = true then x <<< (s' + 1) else 0) =
(x * (y &&& (BitVec.twoPow w (s' + 1)))) := by
simp only [ofNat_eq_ofNat, and_twoPow]
by_cases hy : y.getLsbD (s' + 1) <;> simp [hy]
rw [heq, ← BitVec.mul_add, ← setWidth_setWidth_succ_eq_setWidth_setWidth_add_twoPow]
@[deprecated mulRec_eq_mul_signExtend_setWidth (since := "2024-09-18"),
inherit_doc mulRec_eq_mul_signExtend_setWidth]
abbrev mulRec_eq_mul_signExtend_truncate := @mulRec_eq_mul_signExtend_setWidth
theorem getLsbD_mul (x y : BitVec w) (i : Nat) :
(x * y).getLsbD i = (mulRec x y w).getLsbD i := by
simp only [mulRec_eq_mul_signExtend_setWidth]
rw [setWidth_setWidth_of_le]
· simp
· omega
theorem getElem_mul {x y : BitVec w} {i : Nat} (h : i < w) :
(x * y)[i] = (mulRec x y w)[i] := by
simp [mulRec_eq_mul_signExtend_setWidth]
/-! ## shiftLeft recurrence for bitblasting -/
/--
`shiftLeftRec x y n` shifts `x` to the left by the first `n` bits of `y`.
The theorem `shiftLeft_eq_shiftLeftRec` proves the equivalence of `(x <<< y)` and `shiftLeftRec`.
Together with equations `shiftLeftRec_zero`, `shiftLeftRec_succ`,
this allows us to unfold `shiftLeft` into a circuit for bitblasting.
-/
def shiftLeftRec (x : BitVec w₁) (y : BitVec w₂) (n : Nat) : BitVec w₁ :=
let shiftAmt := (y &&& (twoPow w₂ n))
match n with
| 0 => x <<< shiftAmt
| n + 1 => (shiftLeftRec x y n) <<< shiftAmt
@[simp]
theorem shiftLeftRec_zero {x : BitVec w₁} {y : BitVec w₂} :
shiftLeftRec x y 0 = x <<< (y &&& twoPow w₂ 0) := by
simp [shiftLeftRec]
@[simp]
theorem shiftLeftRec_succ {x : BitVec w₁} {y : BitVec w₂} :
shiftLeftRec x y (n + 1) = (shiftLeftRec x y n) <<< (y &&& twoPow w₂ (n + 1)) := by
simp [shiftLeftRec]
/--
If `y &&& z = 0`, `x <<< (y ||| z) = x <<< y <<< z`.
This follows as `y &&& z = 0` implies `y ||| z = y + z`,
and thus `x <<< (y ||| z) = x <<< (y + z) = x <<< y <<< z`.
-/
theorem shiftLeft_or_of_and_eq_zero {x : BitVec w₁} {y z : BitVec w₂}
(h : y &&& z = 0#w₂) :
x <<< (y ||| z) = x <<< y <<< z := by
rw [← add_eq_or_of_and_eq_zero _ _ h,
shiftLeft_eq', toNat_add_of_and_eq_zero h]
simp [shiftLeft_add]
/--
`shiftLeftRec x y n` shifts `x` to the left by the first `n` bits of `y`.
-/
theorem shiftLeftRec_eq {x : BitVec w₁} {y : BitVec w₂} {n : Nat} :
shiftLeftRec x y n = x <<< (y.setWidth (n + 1)).setWidth w₂ := by
induction n generalizing x y
case zero =>
ext i
simp only [shiftLeftRec_zero, twoPow_zero, Nat.reduceAdd, setWidth_one,
and_one_eq_setWidth_ofBool_getLsbD]
case succ n ih =>
simp only [shiftLeftRec_succ, and_twoPow]
rw [ih]
by_cases h : y.getLsbD (n + 1)
· simp only [h, ↓reduceIte]
rw [setWidth_setWidth_succ_eq_setWidth_setWidth_or_twoPow_of_getLsbD_true h,
shiftLeft_or_of_and_eq_zero]
simp [and_twoPow]
· simp only [h, false_eq_true, ↓reduceIte, shiftLeft_zero']
rw [setWidth_setWidth_succ_eq_setWidth_setWidth_of_getLsbD_false (i := n + 1)]
simp [h]
/--
Show that `x <<< y` can be written in terms of `shiftLeftRec`.
This can be unfolded in terms of `shiftLeftRec_zero`, `shiftLeftRec_succ` for bitblasting.
-/
theorem shiftLeft_eq_shiftLeftRec (x : BitVec w₁) (y : BitVec w₂) :
x <<< y = shiftLeftRec x y (w₂ - 1) := by
rcases w₂ with rfl | w₂
· simp [of_length_zero]
· simp [shiftLeftRec_eq]
/-! # udiv/urem recurrence for bitblasting
In order to prove the correctness of the division algorithm on the integers,
one shows that `n.div d = q` and `n.mod d = r` iff `n = d * q + r` and `0 ≤ r < d`.
Mnemonic: `n` is the numerator, `d` is the denominator, `q` is the quotient, and `r` the remainder.
This *uniqueness of decomposition* is not true for bitvectors.
For `n = 0, d = 3, w = 3`, we can write:
- `0 = 0 * 3 + 0` (`q = 0`, `r = 0 < 3`.)
- `0 = 2 * 3 + 2 = 6 + 2 ≃ 0 (mod 8)` (`q = 2`, `r = 2 < 3`).
Such examples can be created by choosing different `(q, r)` for a fixed `(d, n)`
such that `(d * q + r)` overflows and wraps around to equal `n`.
This tells us that the division algorithm must have more restrictions than just the ones
we have for integers. These restrictions are captured in `DivModState.Lawful`.
The key idea is to state the relationship in terms of the toNat values of {n, d, q, r}.
If the division equation `d.toNat * q.toNat + r.toNat = n.toNat` holds,
then `n.udiv d = q` and `n.umod d = r`.
Following this, we implement the division algorithm by repeated shift-subtract.
References:
- Fast 32-bit Division on the DSP56800E: Minimized nonrestoring division algorithm by David Baca
- Bitwuzla sources for bitblasting.h
-/
private theorem Nat.div_add_eq_left_of_lt {x y z : Nat} (hx : z x) (hy : y < z) (hz : 0 < z) :
(x + y) / z = x / z := by
refine Nat.div_eq_of_lt_le ?lo ?hi
· apply Nat.le_trans
· exact div_mul_le_self x z
· omega
· simp only [succ_eq_add_one, Nat.add_mul, Nat.one_mul]
apply Nat.add_lt_add_of_le_of_lt
· apply Nat.le_of_eq
exact (Nat.div_eq_iff_eq_mul_left hz hx).mp rfl
· exact hy
/-- If the division equation `d.toNat * q.toNat + r.toNat = n.toNat` holds,
then `n.udiv d = q`. -/
theorem udiv_eq_of_mul_add_toNat {d n q r : BitVec w} (hd : 0 < d)
(hrd : r < d)
(hdqnr : d.toNat * q.toNat + r.toNat = n.toNat) :
n.udiv d = q := by
apply BitVec.eq_of_toNat_eq
rw [toNat_udiv]
replace hdqnr : (d.toNat * q.toNat + r.toNat) / d.toNat = n.toNat / d.toNat := by
simp [hdqnr]
rw [Nat.div_add_eq_left_of_lt] at hdqnr
· rw [← hdqnr]
exact mul_div_right q.toNat hd
· exact Nat.dvd_mul_right d.toNat q.toNat
· exact hrd
· exact hd
/-- If the division equation `d.toNat * q.toNat + r.toNat = n.toNat` holds,
then `n.umod d = r`. -/
theorem umod_eq_of_mul_add_toNat {d n q r : BitVec w} (hrd : r < d)
(hdqnr : d.toNat * q.toNat + r.toNat = n.toNat) :
n.umod d = r := by
apply BitVec.eq_of_toNat_eq
rw [toNat_umod]
replace hdqnr : (d.toNat * q.toNat + r.toNat) % d.toNat = n.toNat % d.toNat := by
simp [hdqnr]
rw [Nat.add_mod, Nat.mul_mod_right] at hdqnr
simp only [Nat.zero_add, mod_mod] at hdqnr
replace hrd : r.toNat < d.toNat := by
simpa [BitVec.lt_def] using hrd
rw [Nat.mod_eq_of_lt hrd] at hdqnr
simp [hdqnr]
/-! ### DivModState -/
/-- `DivModState` is a structure that maintains the state of recursive `divrem` calls. -/
structure DivModState (w : Nat) : Type where
/-- The number of bits in the numerator that are not yet processed -/
wn : Nat
/-- The number of bits in the remainder (and quotient) -/
wr : Nat
/-- The current quotient. -/
q : BitVec w
/-- The current remainder. -/
r : BitVec w
/-- `DivModArgs` contains the arguments to a `divrem` call which remain constant throughout
execution. -/
structure DivModArgs (w : Nat) where
/-- the numerator (aka, dividend) -/
n : BitVec w
/-- the denumerator (aka, divisor)-/
d : BitVec w
/-- A `DivModState` is lawful if the remainder width `wr` plus the numerator width `wn` equals `w`,
and the bitvectors `r` and `n` have values in the bounds given by bitwidths `wr`, resp. `wn`.
This is a proof engineering choice: an alternative world could have been
`r : BitVec wr` and `n : BitVec wn`, but this required much more dependent typing coercions.
Instead, we choose to declare all involved bitvectors as length `w`, and then prove that
the values are within their respective bounds.
We start with `wn = w` and `wr = 0`, and then in each step, we decrement `wn` and increment `wr`.
In this way, we grow a legal remainder in each loop iteration.
-/
structure DivModState.Lawful {w : Nat} (args : DivModArgs w) (qr : DivModState w) : Prop where
/-- The sum of widths of the dividend and remainder is `w`. -/
hwrn : qr.wr + qr.wn = w
/-- The denominator is positive. -/
hdPos : 0 < args.d
/-- The remainder is strictly less than the denominator. -/
hrLtDivisor : qr.r.toNat < args.d.toNat
/-- The remainder is morally a `Bitvec wr`, and so has value less than `2^wr`. -/
hrWidth : qr.r.toNat < 2^qr.wr
/-- The quotient is morally a `Bitvec wr`, and so has value less than `2^wr`. -/
hqWidth : qr.q.toNat < 2^qr.wr
/-- The low `(w - wn)` bits of `n` obey the invariant for division. -/
hdiv : args.n.toNat >>> qr.wn = args.d.toNat * qr.q.toNat + qr.r.toNat
/-- A lawful DivModState implies `w > 0`. -/
def DivModState.Lawful.hw {args : DivModArgs w} {qr : DivModState w}
{h : DivModState.Lawful args qr} : 0 < w := by
have hd := h.hdPos
rcases w with rfl | w
· have hcontra : args.d = 0#0 := by apply Subsingleton.elim
rw [hcontra] at hd
simp at hd
· omega
/-- An initial value with both `q, r = 0`. -/
def DivModState.init (w : Nat) : DivModState w := {
wn := w
wr := 0
q := 0#w
r := 0#w
}
/-- The initial state is lawful. -/
def DivModState.lawful_init {w : Nat} (args : DivModArgs w) (hd : 0#w < args.d) :
DivModState.Lawful args (DivModState.init w) := by
simp only [BitVec.DivModState.init]
exact {
hwrn := by simp only; omega,
hdPos := by assumption
hrLtDivisor := by simp [BitVec.lt_def] at hd ⊢; assumption
hrWidth := by simp [DivModState.init],
hqWidth := by simp [DivModState.init],
hdiv := by
simp only [DivModState.init, toNat_ofNat, zero_mod, Nat.mul_zero, Nat.add_zero];
rw [Nat.shiftRight_eq_div_pow]
apply Nat.div_eq_of_lt args.n.isLt
}
/--
A lawful DivModState with a fully consumed dividend (`wn = 0`) witnesses that the
quotient has been correctly computed.
-/
theorem DivModState.udiv_eq_of_lawful {n d : BitVec w} {qr : DivModState w}
(h_lawful : DivModState.Lawful {n, d} qr)
(h_final : qr.wn = 0) :
n.udiv d = qr.q := by
apply udiv_eq_of_mul_add_toNat h_lawful.hdPos h_lawful.hrLtDivisor
have hdiv := h_lawful.hdiv
simp only [h_final] at *
omega
/--
A lawful DivModState with a fully consumed dividend (`wn = 0`) witnesses that the
remainder has been correctly computed.
-/
theorem DivModState.umod_eq_of_lawful {qr : DivModState w}
(h : DivModState.Lawful {n, d} qr)
(h_final : qr.wn = 0) :
n.umod d = qr.r := by
apply umod_eq_of_mul_add_toNat h.hrLtDivisor
have hdiv := h.hdiv
simp only [shiftRight_zero] at hdiv
simp only [h_final] at *
exact hdiv.symm
/-! ### DivModState.Poised -/
/--
A `Poised` DivModState is a state which is `Lawful` and furthermore, has at least
one numerator bit left to process `(0 < wn)`
The input to the shift subtractor is a legal input to `divrem`, and we also need to have an
input bit to perform shift subtraction on, and thus we need `0 < wn`.
-/
structure DivModState.Poised {w : Nat} (args : DivModArgs w) (qr : DivModState w)
extends DivModState.Lawful args qr : Type where
/-- Only perform a round of shift-subtract if we have dividend bits. -/
hwn_lt : 0 < qr.wn
/--
In the shift subtract input, the dividend is at least one bit long (`wn > 0`), so
the remainder has bits to be computed (`wr < w`).
-/
def DivModState.wr_lt_w {qr : DivModState w} (h : qr.Poised args) : qr.wr < w := by
have hwrn := h.hwrn
have hwn_lt := h.hwn_lt
omega
/-! ### Division shift subtractor -/
/--
One round of the division algorithm, that tries to perform a subtract shift.
Note that this should only be called when `r.msb = false`, so we will not overflow.
-/
def divSubtractShift (args : DivModArgs w) (qr : DivModState w) : DivModState w :=
let {n, d} := args
let wn := qr.wn - 1
let wr := qr.wr + 1
let r' := shiftConcat qr.r (n.getLsbD wn)
if r' < d then {
q := qr.q.shiftConcat false, -- If `r' < d`, then we do not have a quotient bit.
r := r'
wn, wr
} else {
q := qr.q.shiftConcat true, -- Otherwise, `r' ≥ d`, and we have a quotient bit.
r := r' - d -- we subtract to maintain the invariant that `r < d`.
wn, wr
}
/-- The value of shifting right by `wn - 1` equals shifting by `wn` and grabbing the lsb at `(wn - 1)`. -/
theorem DivModState.toNat_shiftRight_sub_one_eq
{args : DivModArgs w} {qr : DivModState w} (h : qr.Poised args) :
args.n.toNat >>> (qr.wn - 1)
= (args.n.toNat >>> qr.wn) * 2 + (args.n.getLsbD (qr.wn - 1)).toNat := by
show BitVec.toNat (args.n >>> (qr.wn - 1)) = _
have {..} := h -- break the structure down for `omega`
rw [shiftRight_sub_one_eq_shiftConcat args.n h.hwn_lt]
rw [toNat_shiftConcat_eq_of_lt (k := w - qr.wn)]
· simp
· omega
· apply BitVec.toNat_ushiftRight_lt
omega
/--
This is used when proving the correctness of the division algorithm,
where we know that `r < d`.
We then want to show that `((r.shiftConcat b) - d) < d` as the loop invariant.
In arithmetic, this is the same as showing that
`r * 2 + 1 - d < d`, which this theorem establishes.
-/
private theorem two_mul_add_sub_lt_of_lt_of_lt_two (h : a < x) (hy : y < 2) :
2 * a + y - x < x := by omega
/-- We show that the output of `divSubtractShift` is lawful, which tells us that it
obeys the division equation. -/
theorem lawful_divSubtractShift (qr : DivModState w) (h : qr.Poised args) :
DivModState.Lawful args (divSubtractShift args qr) := by
rcases args with ⟨n, d⟩
simp only [divSubtractShift, decide_eq_true_eq]
-- We add these hypotheses for `omega` to find them later.
have ⟨⟨hrwn, hd, hrd, hr, hn, hrnd⟩, hwn_lt⟩ := h
have : d.toNat * (qr.q.toNat * 2) = d.toNat * qr.q.toNat * 2 := by rw [Nat.mul_assoc]
by_cases rltd : shiftConcat qr.r (n.getLsbD (qr.wn - 1)) < d
· simp only [rltd, ↓reduceIte]
constructor <;> try bv_omega
case pos.hrWidth => apply toNat_shiftConcat_lt_of_lt <;> omega
case pos.hqWidth => apply toNat_shiftConcat_lt_of_lt <;> omega
case pos.hdiv =>
simp [qr.toNat_shiftRight_sub_one_eq h, h.hdiv, this,
toNat_shiftConcat_eq_of_lt (qr.wr_lt_w h) h.hrWidth,
toNat_shiftConcat_eq_of_lt (qr.wr_lt_w h) h.hqWidth]
omega
· simp only [rltd, ↓reduceIte]
constructor <;> try bv_omega
case neg.hrLtDivisor =>
simp only [lt_def, Nat.not_lt] at rltd
rw [BitVec.toNat_sub_of_le rltd,
toNat_shiftConcat_eq_of_lt (hk := qr.wr_lt_w h) (hx := h.hrWidth),
Nat.mul_comm]
apply two_mul_add_sub_lt_of_lt_of_lt_two <;> bv_omega
case neg.hrWidth =>
simp only
have hdr' : d ≤ (qr.r.shiftConcat (n.getLsbD (qr.wn - 1))) :=
BitVec.not_lt_iff_le.mp rltd
have hr' : ((qr.r.shiftConcat (n.getLsbD (qr.wn - 1)))).toNat < 2 ^ (qr.wr + 1) := by
apply toNat_shiftConcat_lt_of_lt <;> bv_omega
rw [BitVec.toNat_sub_of_le hdr']
omega
case neg.hqWidth =>
apply toNat_shiftConcat_lt_of_lt <;> omega
case neg.hdiv =>
have rltd' := (BitVec.not_lt_iff_le.mp rltd)
simp only [qr.toNat_shiftRight_sub_one_eq h,
BitVec.toNat_sub_of_le rltd',
toNat_shiftConcat_eq_of_lt (qr.wr_lt_w h) h.hrWidth]
simp only [BitVec.le_def,
toNat_shiftConcat_eq_of_lt (qr.wr_lt_w h) h.hrWidth] at rltd'
simp only [toNat_shiftConcat_eq_of_lt (qr.wr_lt_w h) h.hqWidth, h.hdiv, Nat.mul_add]
bv_omega
/-! ### Core division algorithm circuit -/
/-- A recursive definition of division for bitblasting, in terms of a shift-subtraction circuit. -/
def divRec {w : Nat} (m : Nat) (args : DivModArgs w) (qr : DivModState w) :
DivModState w :=
match m with
| 0 => qr
| m + 1 => divRec m args <| divSubtractShift args qr
@[simp]
theorem divRec_zero (qr : DivModState w) :
divRec 0 args qr = qr := rfl
@[simp]
theorem divRec_succ (m : Nat) (args : DivModArgs w) (qr : DivModState w) :
divRec (m + 1) args qr =
divRec m args (divSubtractShift args qr) := rfl
/-- The output of `divRec` is a lawful state -/
theorem lawful_divRec {args : DivModArgs w} {qr : DivModState w}
(h : DivModState.Lawful args qr) :
DivModState.Lawful args (divRec qr.wn args qr) := by
generalize hm : qr.wn = m
induction m generalizing qr
case zero =>
exact h
case succ wn' ih =>
simp only [divRec_succ]
apply ih
· apply lawful_divSubtractShift
constructor
· assumption
· omega
· simp only [divSubtractShift, hm]
split <;> rfl
/-- The output of `divRec` has no more bits left to process (i.e., `wn = 0`) -/
@[simp]
theorem wn_divRec (args : DivModArgs w) (qr : DivModState w) :
(divRec qr.wn args qr).wn = 0 := by
generalize hm : qr.wn = m
induction m generalizing qr
case zero =>
assumption
case succ wn' ih =>
apply ih
simp only [divSubtractShift, hm]
split <;> rfl
/-- The result of `udiv` agrees with the result of the division recurrence. -/
theorem udiv_eq_divRec (hd : 0#w < d) :
let out := divRec w {n, d} (DivModState.init w)
n.udiv d = out.q := by
have := DivModState.lawful_init {n, d} hd
have := lawful_divRec this
apply DivModState.udiv_eq_of_lawful this (wn_divRec ..)
/-- The result of `umod` agrees with the result of the division recurrence. -/
theorem umod_eq_divRec (hd : 0#w < d) :
let out := divRec w {n, d} (DivModState.init w)
n.umod d = out.r := by
have := DivModState.lawful_init {n, d} hd
have := lawful_divRec this
apply DivModState.umod_eq_of_lawful this (wn_divRec ..)
theorem divRec_succ' (m : Nat) (args : DivModArgs w) (qr : DivModState w) :
divRec (m+1) args qr =
let wn := qr.wn - 1
let wr := qr.wr + 1
let r' := shiftConcat qr.r (args.n.getLsbD wn)
let input : DivModState _ :=
if r' < args.d then {
q := qr.q.shiftConcat false,
r := r'
wn, wr
} else {
q := qr.q.shiftConcat true,
r := r' - args.d
wn, wr
}
divRec m args input := by
simp [divRec_succ, divSubtractShift]
/- ### Arithmetic shift right (sshiftRight) recurrence -/
/--
`sshiftRightRec x y n` shifts `x` arithmetically/signed to the right by the first `n` bits of `y`.
The theorem `sshiftRight_eq_sshiftRightRec` proves the equivalence of `(x.sshiftRight y)` and `sshiftRightRec`.
Together with equations `sshiftRightRec_zero`, `sshiftRightRec_succ`,
this allows us to unfold `sshiftRight` into a circuit for bitblasting.
-/
def sshiftRightRec (x : BitVec w₁) (y : BitVec w₂) (n : Nat) : BitVec w₁ :=
let shiftAmt := (y &&& (twoPow w₂ n))
match n with
| 0 => x.sshiftRight' shiftAmt
| n + 1 => (sshiftRightRec x y n).sshiftRight' shiftAmt
@[simp]
theorem sshiftRightRec_zero_eq (x : BitVec w₁) (y : BitVec w₂) :
sshiftRightRec x y 0 = x.sshiftRight' (y &&& 1#w₂) := by
simp only [sshiftRightRec, twoPow_zero]
@[simp]
theorem sshiftRightRec_succ_eq (x : BitVec w₁) (y : BitVec w₂) (n : Nat) :
sshiftRightRec x y (n + 1) = (sshiftRightRec x y n).sshiftRight' (y &&& twoPow w₂ (n + 1)) := by
simp [sshiftRightRec]
/--
If `y &&& z = 0`, `x.sshiftRight (y ||| z) = (x.sshiftRight y).sshiftRight z`.
This follows as `y &&& z = 0` implies `y ||| z = y + z`,
and thus `x.sshiftRight (y ||| z) = x.sshiftRight (y + z) = (x.sshiftRight y).sshiftRight z`.
-/
theorem sshiftRight'_or_of_and_eq_zero {x : BitVec w₁} {y z : BitVec w₂}
(h : y &&& z = 0#w₂) :
x.sshiftRight' (y ||| z) = (x.sshiftRight' y).sshiftRight' z := by
simp [sshiftRight', ← add_eq_or_of_and_eq_zero _ _ h,
toNat_add_of_and_eq_zero h, sshiftRight_add]
theorem sshiftRightRec_eq (x : BitVec w₁) (y : BitVec w₂) (n : Nat) :
sshiftRightRec x y n = x.sshiftRight' ((y.setWidth (n + 1)).setWidth w₂) := by
induction n generalizing x y
case zero =>
ext i
simp [twoPow_zero, Nat.reduceAdd, and_one_eq_setWidth_ofBool_getLsbD, setWidth_one]
case succ n ih =>
simp only [sshiftRightRec_succ_eq, and_twoPow, ih]
by_cases h : y.getLsbD (n + 1)
· rw [setWidth_setWidth_succ_eq_setWidth_setWidth_or_twoPow_of_getLsbD_true h,
sshiftRight'_or_of_and_eq_zero (by simp [and_twoPow]), h]
simp
· rw [setWidth_setWidth_succ_eq_setWidth_setWidth_of_getLsbD_false (i := n + 1)
(by simp [h])]
simp [h]
/--
Show that `x.sshiftRight y` can be written in terms of `sshiftRightRec`.
This can be unfolded in terms of `sshiftRightRec_zero_eq`, `sshiftRightRec_succ_eq` for bitblasting.
-/
theorem sshiftRight_eq_sshiftRightRec (x : BitVec w₁) (y : BitVec w₂) :
(x.sshiftRight' y).getLsbD i = (sshiftRightRec x y (w₂ - 1)).getLsbD i := by
rcases w₂ with rfl | w₂
· simp [of_length_zero]
· simp [sshiftRightRec_eq]
/- ### Logical shift right (ushiftRight) recurrence for bitblasting -/
/--
`ushiftRightRec x y n` shifts `x` logically to the right by the first `n` bits of `y`.
The theorem `shiftRight_eq_ushiftRightRec` proves the equivalence
of `(x >>> y)` and `ushiftRightRec`.
Together with equations `ushiftRightRec_zero`, `ushiftRightRec_succ`,
this allows us to unfold `ushiftRight` into a circuit for bitblasting.
-/
def ushiftRightRec (x : BitVec w₁) (y : BitVec w₂) (n : Nat) : BitVec w₁ :=
let shiftAmt := (y &&& (twoPow w₂ n))
match n with
| 0 => x >>> shiftAmt
| n + 1 => (ushiftRightRec x y n) >>> shiftAmt
@[simp]
theorem ushiftRightRec_zero (x : BitVec w₁) (y : BitVec w₂) :
ushiftRightRec x y 0 = x >>> (y &&& twoPow w₂ 0) := by
simp [ushiftRightRec]
@[simp]
theorem ushiftRightRec_succ (x : BitVec w₁) (y : BitVec w₂) :
ushiftRightRec x y (n + 1) = (ushiftRightRec x y n) >>> (y &&& twoPow w₂ (n + 1)) := by
simp [ushiftRightRec]
/--
If `y &&& z = 0`, `x >>> (y ||| z) = x >>> y >>> z`.
This follows as `y &&& z = 0` implies `y ||| z = y + z`,
and thus `x >>> (y ||| z) = x >>> (y + z) = x >>> y >>> z`.
-/
theorem ushiftRight'_or_of_and_eq_zero {x : BitVec w₁} {y z : BitVec w₂}
(h : y &&& z = 0#w₂) :
x >>> (y ||| z) = x >>> y >>> z := by
simp [← add_eq_or_of_and_eq_zero _ _ h, toNat_add_of_and_eq_zero h, shiftRight_add]
theorem ushiftRightRec_eq (x : BitVec w₁) (y : BitVec w₂) (n : Nat) :
ushiftRightRec x y n = x >>> (y.setWidth (n + 1)).setWidth w₂ := by
induction n generalizing x y
case zero =>
ext i
simp only [ushiftRightRec_zero, twoPow_zero, Nat.reduceAdd,
and_one_eq_setWidth_ofBool_getLsbD, setWidth_one]
case succ n ih =>
simp only [ushiftRightRec_succ, and_twoPow]
rw [ih]
by_cases h : y.getLsbD (n + 1) <;> simp only [h, ↓reduceIte]
· rw [setWidth_setWidth_succ_eq_setWidth_setWidth_or_twoPow_of_getLsbD_true h,
ushiftRight'_or_of_and_eq_zero]
simp [and_twoPow]
· simp [setWidth_setWidth_succ_eq_setWidth_setWidth_of_getLsbD_false, h]
/--
Show that `x >>> y` can be written in terms of `ushiftRightRec`.
This can be unfolded in terms of `ushiftRightRec_zero`, `ushiftRightRec_succ` for bitblasting.
-/
theorem shiftRight_eq_ushiftRightRec (x : BitVec w₁) (y : BitVec w₂) :
x >>> y = ushiftRightRec x y (w₂ - 1) := by
rcases w₂ with rfl | w₂
· simp [of_length_zero]
· simp [ushiftRightRec_eq]
end BitVec