This PR changes `Lean.Grind.NoNatZeroDivisors` so that it is
parametrised by a `NatModule` instance rather than just a `HMul`
instance. This is sufficiently general for our purposes, and is a
band-aid (~40% improvement) for the performance problems we've been
seeing coming from inference here. The problems observed in Mathlib may
not see much improvement, however.
This PR lets the equation compiler unfold abstracted proofs again if
they would otherwise hide recursive calls.
This fixes#8939.
---------
Co-authored-by: Sebastian Ullrich <sebasti@nullri.ch>
This PR uses `withAbstractAtoms` to prevent the kernel from accidentally
reducing the atoms in the arith normlizer while typechecking. This PR
also sets `implicitDefEqProofs := false` in the `grind` normalizer
(Almost) only typos in constant names and doc-strings were considered;
grammar was not considered. Also, along others,
`mkDefinitionValInferrringUnsafe` has been fixed :-)
This PR surfaces kernel diagnostics even in `example`.
The problem was that the kernel checking happens asynchronously. We
cannot use `reportDiag` in `addDecl`, which spawns that task, due to the
module hierarchy. For non `example`-declaration, `reportDiag` is called
somewhere else later, but for `example`, the `withoutModifyingEnv` in
`elabMutualDef` hid the kernel diagnostics. (But only the kernel
diagnostics; they are in the `Environment`, while the others are in the
`State`).
I also observed that the `reportDiag` in `elabAsync` (but not in
`elabSync`) duplicated the reporting, so without `elab.Async true` you
get the message twice. To fix this, `reportDiag` now resets the
diagnostics. This should avoid reporting counts twice in general (at
least within a linear use of the state).
---------
Co-authored-by: Sebastian Ullrich <sebasti@nullri.ch>
This PR adds a `HPow \a Int \a` field to `Lean.Grind.Field`, and
sufficient axioms to connect it to the operations, so that in future we
can reason about exponents in `grind`. To avoid collisions, we also move
the `HPow \a Nat \a` field in `Semiring` from the extends clause to a
field. Finally, we add some failing tests about normalizing exponents.
This PR resolves an issue where the `Meta.Context.configKey` field is
private but we still want to use the constructor of the structure for
setting other fields, which would be prevented by the module system
checks:
```lean
structure Context where
private config : Config := {}
private configKey : UInt64 := config.toKey
...
def ContextInfo.runMetaM (info : ContextInfo) (lctx : LocalContext) (x : MetaM α) : IO α := do
-- cannot call private constructor of `Meta.Context`!
(·.1) <$> info.runCoreM (x.run { lctx := lctx } { mctx := info.mctx })
```
Instead, the private field is extracted into an (existing) structure
that applies its default value:
```lean
/-- Configuration with key produced by `Config.toKey`. -/
structure ConfigWithKey where
private mk ::
config : Config := {}
key : UInt64 := config.toKey
structure Context where
keyedConfig : ConfigWithKey := default
```
Thus `Context`'s constructor remains public without exposing a way to
set `key` directly.
This PR fixes a kernel type mismatch that occurs when using `grind` on
goals containing non-standard `OfNat.ofNat` terms. For example, in issue
#9477, the `0` in the theorem `range_lower` has the form:
```lean
(@OfNat.ofNat
(Std.PRange.Bound (Std.PRange.RangeShape.lower (Std.PRange.RangeShape.mk Std.PRange.BoundShape.closed Std.PRange.BoundShape.open)) Nat)
(nat_lit 0)
(instOfNatNat (nat_lit 0)))
```
instead of the more standard form:
```lean
(@OfNat.ofNat
Nat
(nat_lit 0)
(instOfNatNat (nat_lit 0)))
```
Closes#9477
This PR improves the `evalInt?` function, which is used to evaluate
configuration parameters from the `ToInt` type class. This PR also adds
a new `evalNat?` function for handling the `IsCharP` type class, and
introduces a configuration option:
```
grind (exp := <num>)
```
This option controls the maximum exponent size considered during
expression evaluation. Previously, `evalInt?` used `whnf`, which could
run out of stack space when reducing terms such as `2^1024`.
closes#9427
This PR makes `PProdN.reduceProjs` also look for projection functions.
Previously, all redexes were created by the functions in `PProdN`, which
used primitive projections. But with `mkAdmProj` the projection
functions creep in via the types of the `admissible_pprod_fst` theorem.
So let's just reduce both of them.
Fixes#9462.
This PR removes uses of `Lean.RBMap` in Lean itself.
Furthermore some massaging of the import graph is done in order to avoid
having `Std.Data.TreeMap.AdditionalOperations` (which is quite
expensive) be the critical path for a large chunk of Lean. In particular
we can build `Lean.Meta.Simp` and `Lean.Meta.Grind` without it thanks to
these changes.
We did previously not conduct this change as `Std.TreeMap` was not
outperforming `Lean.RBMap` yet, however this has changed with the new
code generator.
This PR addresses the lean crash (stack overflow) with nested induction
and the generation of the `SizeOf` spec lemmas, reported at #9018.
It does not address the underlying issue that in these cases, the
generated SizeOf code does not match the the SizeOf function found by
instance search, and thus the generation fails.
This seem hard to fix: `mkSizeOfMinors` would have to recognize that
given, say, `List (Id Tree)`, the derived instance (assuming there was
one for `Tree`) is not the same as for `List Tree`.
The problem seems just about as hard as getting derived SizeOf right in
the presence of nested induction and non-canonical SizeOf instances.
This PR improves the error messages produced by the `split` tactic,
including suggesting syntax fixes and related tactics with which it
might be confused.
Note that, to avoid clashing with the new error message styling
conventions used in these messages, this PR also updates the formatting
of the message produced by `throwTacticEx`.
Closes#6224
This PR updates the formatting of, and adds explanations for, "unknown
identifier" errors as well as "failed to infer type" errors for binders
and definitions.
It attempts to ameliorate some of the confusion encountered in #1592 by
modifying the wording of the "header is elaborated before body is
processed" note and adding further discussion and examples of this
behavior in the corresponding error explanation.
This PR increases the number of cases where `isArrowProposition` returns
a result other than `.undef`. This function is used to implement the
`isProof` predicate, which is invoked on every subterm visited by
`simp`.
This PR adds improves the "invalid named argument" error message in
function applications and match patterns by providing clickable hints
with valid argument names. In so doing, it also fixes an issue where
this error message would erroneously flag valid match-pattern argument
names.
This PR implements a simple optimization: dependent implications are no
longer treated as E-matching theorems in `grind`. In
`grind_bitvec2.lean`, this change saves around 3 seconds, as many
dependent implications are generated. Example:
```lean
∀ (h : i + 1 ≤ w), x.abs.getLsbD i = x.abs[i]
```
This PR fixes a bug at `mkCongrSimpCore?`. It fixes the issue reported
by @joehendrix at #9388.
The fix is just commit: afc4ba617fe2ca5828e0e252558d893d7791d56b. The
rest of the PR is just cleaning up the file.
closes#9388
This PR fixes a performance issue that occurs when generating equation
lemmas for functions that use match-expressions containing several
literals. This issue was exposed by #9322 and arises from a combination
of factors:
1. Literal values are compiled into a chain of dependent if-then-else
expressions.
2. Dependent if-then-else expressions are significantly more expensive
to simplify than regular ones.
3. The `split` tactic selects a target, splits it, and then invokes
`simp` on the resulting subgoals. Moreover, `simp` traverses the entire
goal bottom-up and does not stop after reaching the target.
This PR addresses the issue by introducing a custom simproc that avoids
recursively simplifying nested if-then-else expressions. It does **not**
alter the user-facing behavior of the `split` tactic because such a
change would be highly disruptive. Instead, the PR adds a new flag,
`backward.split` to control the behavior of the user-facing `split`
tactic. It is currently set to `true`, i.e., the old behavior is still
the default one. In a future PR, we should set this flag to `false` by
default and begin repairing all affected proofs.
closes#9322
This PR modifies the encoding from `Nat` to `Int` used in `grind
cutsat`. It is simpler, more extensible, and similar to the generic
`ToInt`. After update stage0, we will be able to delete the leftovers.
This PR uses the `mkCongrSimpForConst?` API in `simp` to reduce the
number of times the same congruence lemma is generated. Before this PR,
`grind` would spend `1.5`s creating congruence theorems during
normalization in the `grind_bitvec2.lean` benchmark. It now spends
`0.6`s. This PR should make an even bigger difference after we merge
#9300.
This PR replaces the `reduceCtorEq` simproc used in `grind` by a much
more efficient one. The default one use in `simp` is just overhead
because the `grind` normalizer is already normalizing arithmetic.
In a separate PR, we will push performance improvements to the default
`reduceCtorEq`.
This PR optimizes support for `Decidable` instances in `grind`. Because
`Decidable` is a subsingleton, the canonicalizer no longer wastes time
normalizing such instances, a significant performance bottleneck in
benchmarks like `grind_bitvec2.lean`. In addition, the
congruence-closure module now handles `Decidable` instances, and can
solve examples such as:
```lean
example (p q : Prop) (h₁ : Decidable p) (h₂ : Decidable (p ∧ q)) : (p ↔ q) → h₁ ≍ h₂ := by
grind
```
This PR makes `isDefEq` detect more stuck definitional equalities
involving smart unfoldings. Specifically, if `t =?= defn ?m` and `defn`
matches on its argument, then this equality is stuck on `?m`. Prior to
this change, we would not see this dependency and simply return `false`.
Fixes#8766.
Co-authored-by: Kyle Miller <kmill31415@gmail.com>
This PR improves the `congr` tactic so that it can handle function
applications with fewer arguments than the arity of the head function.
This also fixes a bug where `congr` could not make progress with
`Set`-valued functions in Mathlib, since `Set` was being unfolded and
making such functions have an apparently higher arity.
This addresses issue #2128 for the `congr` tactic, but not `simp` and
others.
This PR migrates usages of `Std.Range` to the new polymorphic ranges.
This PR unfortunately increases the transitive imports for
frequently-used parts of `Init` because the ranges now rely on iterators
in order to provide their functionality for types other than `Nat`.
However, iteration over ranges in compiled code is as efficient as
before in the examples I checked. This is because of a special
`IteratorLoop` implementation provided in the PR for this purpose.
There were two issues that were uncovered during migration:
* In `IndPredBelow.lean`, migrating the last remaining range causes
`compilerTest1.lean` to break. I have minimized the issue and came to
the conclusion it's a compiler bug. Therefore, I have not replaced said
old range usage yet (see #9186).
* In `BRecOn.lean`, we are publicly importing the ranges. Making this
import private should theoretically work, but there seems to be a
problem with the module system, causing the build to panic later in
`Init.Data.Grind.Poly` (see #9185).
* In `FuzzyMatching.lean`, inlining fails with the new ranges, which
would have led to significant slowdown. Therefore, I have not migrated
this file either.
