This PR improves the E-matching pattern inference procedure in `grind`.
Consider the following theorem:
```lean
@[grind →]
theorem eq_empty_of_append_eq_empty {xs ys : Array α} (h : xs ++ ys = #[]) : xs = #[] ∧ ys = #[] :=
append_eq_empty_iff.mp h
```
Before this PR, `grind` inferred the following pattern:
```lean
@HAppend.hAppend _ _ _ _ #2#1
```
Note that this pattern would match any `++` application, even if it had
nothing to do with arrays. With this PR, the inferred pattern becomes:
```lean
@HAppend.hAppend (Array #3) (Array _) (Array _) _ #2#1
```
With the new pattern, the theorem will not be considered by `grind` for
goals that do not involve `Array`s.
This PR implements **stepwise proof terms** in the commutative ring
procedure used by `grind`. These terms serve as an alternative
representation to the traditional Nullstellensatz certificates, aiming
to address the **exponential worst-case complexity** often associated
with certificate construction.
While various compression techniques for Nullstellensatz certificates
exist, they are not implemented in our procedure. Moreover, many of
these techniques rely on additional properties not available in
arbitrary commutative rings. In contrast, the stepwise proof terms
encode the **actual derivation** used during simplification, offering
significantly better scalability in practice.
Here is a motivating example:
```lean
example {α} [CommRing α] [IsCharP α 0] (d t c : α) (d_inv PSO3_inv : α)
(Δ40 : d^2 * (d + t - d * t - 2) * (d + t + d * t) = 0)
(Δ41 : -d^4 * (d + t - d * t - 2) *
(2 * d + 2 * d * t - 4 * d * t^2 + 2 * d * t^4 + 2 * d^2 * t^4 - c * (d + t + d * t)) = 0)
(_ : d * d_inv = 1)
(_ : (d + t - d * t - 2) * PSO3_inv = 1) :
t^2 = t + 1 := by grind +ring
```
In this case, the Nullstellensatz certificate generated by our procedure
contains **over 20,000 terms**, which overwhelms the Lean kernel during
verification. @kim-em also computed certificates using Mathematica with
various variable orderings, producing results between **500 and 2,000
terms**: still quite large.
By switching to stepwise derivations:
- `grind` completes the goal in **under 10 ms**
- The Lean kernel checks the resulting proof term in **under 1 second**
This change dramatically improves both the performance and robustness of
`grind` for nontrivial algebraic goals.
This PR adds support for inductive and coinductive predicates defined
using lattice theoretic structures on `Prop`. These are syntactically
defined using `greatest_fixpoint` or `least_fixpoint` termination
clauses for recursive `Prop`-valued functions. The functionality relies
on `partial_fixpoint` machinery and requires function definitions to be
monotone. For non-mutually recursive predicates, an appropriate
(co)induction proof principle (given by Park induction) is generated.
Summary of changes:
- `Interal.Order.Basic` now contains `CompleteLattice` class, as well as
version of Knaster-Tarski fixpoint theorem (with an associated Park
induction principle) for the internal use for defining (co)inductive
predicates. `Prop` is shown to have two complete lattice structures (one
given by implication order for defining inductive predicates, and one
given by reverse implication for defining coinductive predicates).
Additionally, proofs that lattices are closed under products and
function spaces are included.
- Partial fixpoint's `EqnInfo` now additionally carries an information
whether something is defined as a lattice-theoretic fixpoint or via
CCPOs.
- When constructing a (co)inductive predicate,`PartialFixpoint/Main`
builds an appropriate lattice structure on the type of the predicate
using product lattice, function space lattice and an appropriate lattice
instance on `Prop`.
- `PartialFixpoint/Eqns` is modified to be able to perform rewrite under
lattice-theoretic fixpoint construction
- `PartialFixpoint/Induction`contains a case split for handling of the
(co)inductive predicates. In the case of lattice-theoretic fixpoints, it
appropriately desugars the Park induction principle.
This PR lets `induction` accept eliminator where the motive application
in the conclusion has complex arguments; these are abstracted over using
`kabstract` if possible. This feature will go well with unfolding
induction principles (#8088).
This PR fixes a bug when constructing the proof term for a
Nullstellensatz certificate produced by the new commutative ring
procedure in `grind`. The kernel was rejecting the proof term.
This PR changes `Lean.Grind.CommRing` to inline the `NatCast` instance
(i.e. to be provided by the user) rather than constructing one from the
existing data. Without this change we can't construct instances in
Mathlib that `grind` can use.
This PR adds the “unfolding” variant of the functional induction and
functional cases principles, under the name `foo.induct_unfolding` resp.
`foo.fun_cases_unfolding`. These theorems combine induction over the
structure of a recursive function with the unfolding of that function,
and should be more reliable, easier to use and more efficient than just
case-splitting and then rewriting with equational theorems.
For example instead of
```
ackermann.induct
(motive : Nat → Nat → Prop)
(case1 : ∀ (m : Nat), motive 0 m)
(case2 : ∀ (n : Nat), motive n 1 → motive (Nat.succ n) 0)
(case3 : ∀ (n m : Nat), motive (n + 1) m → motive n (ackermann (n + 1) m) → motive (Nat.succ n) (Nat.succ m))
(x x : Nat) : motive x x
```
one gets
```
ackermann.fun_cases_unfolding
(motive : Nat → Nat → Nat → Prop)
(case1 : ∀ (m : Nat), motive 0 m (m + 1))
(case2 : ∀ (n : Nat), motive n.succ 0 (ackermann n 1))
(case3 : ∀ (n m : Nat), motive n.succ m.succ (ackermann n (ackermann (n + 1) m)))
(x✝ x✝¹ : Nat) : motive x✝ x✝¹ (ackermann x✝ x✝¹)
```
This PR fixes the `grind +splitImp` and the arrow propagator. Given `p :
Prop`, the propagator was incorrectly assuming `A` was always a
proposition in an arrow `A -> p`. This PR also adds a missing
normalization rule to `grind`.
This PR improves equality propagation (also known as theory combination)
and polynomial simplification for rings that do not implement the
`NoZeroNatDivisors` class. With these fixes, `grind` can now solve:
```lean
example [CommRing α] (a b c : α) (f : α → Nat)
: a + b + c = 3 →
a^2 + b^2 + c^2 = 5 →
a^3 + b^3 + c^3 = 7 →
f (a^4 + b^4) + f (9 - c^4) ≠ 1 := by
grind +ring
```
This example uses the commutative ring procedure, the linear integer
arithmetic solver, and congruence closure.
For rings that implement `NoZeroNatDivisors`, a polynomial is now also
divided by the greatest common divisor (gcd) of its coefficients when it
is inserted into the basis.
This PR ensures that `set_option grind.debug true` works properly when
using `grind +ring`. It also adds the helper functions `mkPropEq` and
`mkExpectedPropHint`.
This PR implements equality propagation in the new commutative ring
procedure in `grind`. The idea is to propagate implied equalities back
to the `grind` core module that does congruence closure. In the
following example, the equalities: `x^2*y = 1` and `x*y^2 - y = 0` imply
that `y*x` is equal to `y*x*y`, which implies by congruence that `f
(y*x) = f (y*x*y)`.
```lean
example [CommRing α] (x y : α) (f : α → Nat) : x^2*y = 1 → x*y^2 - y = 0 → f (y*x) = f (y*x*y) := by
grind +ring
```
This PR implements the main loop of the new commutative ring procedure
in `grind`. In the main loop, for each polynomial `p` in the todo queue,
the procedure:
- Simplifies it using the current basis.
- Computes critical pairs with polynomials already in the basis and adds
them to the queue.
After the queue is empty, the disequalities are re-simplified using the
new basis. `grind` can now solve examples such as:
```lean
example [CommRing α] (x y : α) : x*y*x = 1 → x*y*y = y → y = 1 := by
grind +ring
example [CommRing α] (x y : α) : x^2*y = 1 → x*y^2 = y → y*x = 1 := by
grind +ring
example (x y : BitVec 16) : x^2*y = 1 → x*y^2 = y → y*x = 1 := by
grind +ring
```
This PR implements the generation of compact proof terms for
Nullstellensatz certificates in the new commutative ring procedure in
`grind`. Some examples:
```lean
example [CommRing α] (x y : α) : x = 1 → y = 2 → 2*x + y = 4 := by
grind +ring
example [CommRing α] [IsCharP α 7] (x y : α) : 3*x = 1 → 3*y = 2 → x + y = 1 := by
grind +ring
example [CommRing α] [NoZeroNatDivisors α] (x y : α) : 3*x = 1 → 3*y = 2 → x + y = 1 := by
grind +ring
example (x y z : BitVec 8) : z = y → (x + 1)*(x - 1)*y + y = z*x^2 + 1 → False := by
grind +ring
```
This PR adds the helper type class `NoZeroNatDivisors` for the
commutative ring procedure in `grind`. Core only implements it for
`Int`. It can be instantiated in Mathlib for any type `A` that
implements `NoZeroSMulDivisors Nat A`.
See `findSimp?` and `PolyDerivation` for details on how this instance
impacts the commutative ring procedure.
This PR fixes the generation of functional induction principles for
functions with nested nested well-founded recursion and late fixed
parameters. This is a follow-up for #7166. Fixes#8093.
This PR makes sure that the functional induction priciples for mutually
recursive structural functions with extra parameters are split deeply,
as expected.
This PR implements `EqCnstr.mkNullCertExt`. Given an implied polynomial
equation `p = 0`, it generates the certificate:
```
q₁ * h₁ + … + qₙ * hₙ
```
for `d * p = 0`, where each `qᵢ`s are polynomials and each `hᵢ` is an
equational hypothesis of the form `lhsᵢ = rhsᵢ`. `d` is a numeral.
This PR makes `IntCast` a field of `Lean.Grind.CommRing`, along with
additional axioms relating it to negation of `OfNat`. This allows use to
use existing instances which are not definitionally equal to the
previously given construction.
---------
Co-authored-by: Leonardo de Moura <leomoura@amazon.com>
This PR implements tactics called `extract_lets` and `lift_lets` that
manipulate `let`/`let_fun` expressions. The `extract_lets` tactic
creates new local declarations extracted from any `let` and `let_fun`
expressions in the main goal. For top-level lets in the target, it is
like the `intros` tactic, but in general it can extract lets from deeper
subexpressions as well. The `lift_lets` tactic moves `let` and `let_fun`
expressions as far out of an expression as possible, but it does not
extract any new local declarations. The option `extract_lets +lift`
combines these behaviors.
This is a re-implementation of `extract_lets` and `lift_lets` from
mathlib. The new `extract_lets` is like doing `lift_lets; extract_lets`,
but it does not lift unextractable lets like `lift_lets`. The
`lift_lets; extract_lets` behavior is now handled by `extract_lets
+lift`. The new `lift_lets` tactic is a frontend to `extract_lets +lift`
machinery, which rather than creating new local definitions instead
represents the accumulated local declarations as top-level lets.
There are also conv tactics for both of these. The `extract_lets` has a
limitation due to the conv architecture; it can extract lets for a given
conv goal, but the local declarations don't survive outside conv. They
get zeta reduced immediately upon leaving conv.
This PR makes the following modifications to the new comm ring procedure
in `grind`
1. Adds data-structures for representing equations (and their
justifications), basis, and queue of equations to be processed.
2. Adds `RingM` helper monad.
3. Adds equation simplification main loop
This PR adds support to `grind` for detecting unsatisfiable commutative
ring equations when the ring characteristic is known. Examples:
```lean
example (x : Int) : (x + 1)*(x - 1) = x^2 → False := by
grind +ring
example (x y : Int) : (x + 1)*(x - 1)*y + y = y*x^2 + 1 → False := by
grind +ring
example (x : UInt8) : (x + 1)*(x - 1) = x^2 → False := by
grind +ring
example (x y : BitVec 8) : (x + 1)*(x - 1)*y + y = y*x^2 + 1 → False := by
grind +ring
```
This PR implements basic support for `CommRing` in `grind`. Terms are
already being reified and normalized. We still need to process the
equations, but `grind` can already prove simple examples such as:
```lean
open Lean.Grind in
example [CommRing α] (x : α) : (x + 1)*(x - 1) = x^2 - 1 := by
grind +ring
open Lean.Grind in
example [CommRing α] [IsCharP α 256] (x : α) : (x + 16)*(x - 16) = x^2 := by
grind +ring
example (x : Int) : (x + 1)*(x - 1) = x^2 - 1 := by
grind +ring
example (x : UInt8) : (x + 16)*(x - 16) = x^2 := by
grind +ring
example (x : Int) : (x + 1)^2 - 1 = x^2 + 2*x := by
grind +ring
example (x : BitVec 8) : (x + 16)*(x - 16) = x^2 := by
grind +ring
example (x : BitVec 8) : (x + 1)^2 - 1 = x^2 + 2*x := by
grind +ring
```
This PR ensures that `mkAppM` can be used to construct terms that are
only type-correct at at default transparency, even if we are in
`withReducible` (e.g. in simp), so that simp does not stumble over
simplifying `let` expression with simplifiable type.reliable.
Here is a reproducer of the issue this solves:
```
example (a b : Nat) (h : a = b):
(let _ : id Bool := true; a) = (let _ : Bool := true; b) := by
simp -zeta -zetaDelta [h]
```
This fixes#7826.
This PR adds the option `debug.terminalTacticsAsSorry`. When enabled,
terminal tactics such as `grind` and `omega` are replaced with `sorry`.
Useful for debugging and fixing bootstrapping issues.
This PR removes all type annotations (optional paramters, auto
parameters, out params, semi-out params, not just optional parameters as
before) from the type of functional induction principles.
This PR adds some docstrings to clarify the functions of
`Lean.mkFreshId`, `Lean.Core.mkFreshUserName`,
`Lean.Elab.Term.mkFreshBinderName`, and
`Lean.Meta.mkFreshBinderNameForTactic`.
This PR adds the attribute `[grind ext]`. It is used to select which
`[ext]` theorems should be used by `grind`. The option `grind +extAll`
instructs `grind` to use all `[ext]` theorems available in the
environment.
After update stage0, we need to add the builtin `[grind ext]`
annotations to key theorems such as `funext`.
