This PR fixes `ST.Ref.ptrEq` to act as described in the docs. This fixes
two bugs:
1. The recent `IO.RealWorld` elimination PR overlooked this function
(afaik this is the only one),
causing its return value to be generally wrong.
2. The implementation of `ptrEq` would previously always consider two
different cells with pointer
equivalent value to be pointer equal. However, the function is supposed
to check whether two
`Ref` are the same cell, not whether the contained elements are.
This PR implements equality propagation for `Nat` in `grind order`.
`grind order` supports offset equalities for rings, but it has an
adapter for `Nat`. Example:
```lean
example (a b : Nat) (f : Nat → Int) : a ≤ b + 1 → b + 1 ≤ a → f (1 + a) = f (1 + b + 1) := by
grind -offset -mbtc -lia -linarith (splits := 0)
```
This PR implements (nested term) equality propagation in `grind order`.
That is, it propagates implied equalities from `grind order` back to the
`grind` core. Examples:
```lean
open Lean Grind Std
example [LE α] [IsPartialOrder α] (a b : α) (f : α → Nat) : a ≤ b → b ≤ c → c ≤ a → f a = f b := by
grind (splits := 0)
example [CommRing α] [LE α] [LT α] [LawfulOrderLT α] [IsPartialOrder α] [OrderedRing α]
(a b : α) (f : α → Int) : a ≤ b + 1 → b ≤ a - 1 → f a = f (2 + b - 1) := by
grind -mbtc -lia -linarith (splits := 0)
example (a b : Int) (f : Int → Int) : a ≤ b + 1 → b ≤ a - 1 → f a = f (2 + b - 1) := by
grind -mbtc -lia -linarith (splits := 0)
```
`prelude-injectivity.lean` was testing inj thm generation for all
inductives in core, including private ones, which could lead to name
clashes that should not be able to occur in actual files. Put it under
the module system to not load private decls in the first place.
This PR fixes a case of overeager constant folding on Nat where the
compiler would mistakenly assume `0 - x = x` (see also #11042 for the
same bug on UInts).
This PR adds a new suggestion to `finish?`. It now generates the `grind`
tactic script as before, and a `finish only` tactic. Example:
```lean
/--
info: Try these:
[apply] ⏎
instantiate only [findIdx, insert, = mem_indices_of_mem]
instantiate only [= getElem?_neg, = getElem?_pos]
cases #1bba
· instantiate only [findIdx]
· instantiate only
instantiate only [= HashMap.mem_insert, = HashMap.getElem_insert]
[apply] finish only [findIdx, insert, = mem_indices_of_mem, = getElem?_neg, = getElem?_pos, = HashMap.mem_insert,
= HashMap.getElem_insert, #1bba]
-/
example (m : IndexMap α β) (a : α) (b : β) :
(m.insert a b).findIdx a = if h : a ∈ m then m.findIdx a else m.size := by
grind => finish?
```
This PR adds a library suggestion engine for local theorems. To be
useful, I still need to write more combinators to re-rank and combine
suggestions from multiple engines.
This is stacked on top of #11029.
This PR changes the terminology used from "premise selection" to
"library suggestions". This will be more understandable to users (we
don't assume anyone is familiar with the premise selection literature),
and avoids a conflict with the existing use of "premise" in Lean
terminology (e.g. "major premise" in induction, as well as generally the
synonym for "hypothesis"/"argument").
This PR improves match compilation: Branch on variables in the order
suggested by the first remaining alternative, and do not branch when the
first remaining alternative does not require it. This fixes
https://github.com/leanprover/lean4/issues/10749. With `set_option
backwards.match.rowMajor false` the old behavior can be turned on.
(For now this is an experiment to get familiar with the code and the
whole
problem domain. It is likely overly naive.)
This PR improves the detection of situations where we branch multiple
times on the same value in the
code generator. Previously this would only consider repeated branching
on function arguments, now on
arbitrary values.
Closes: #11018
This PR improves join point finding in the compiler through two means:
1. We now handle situations where a function `f` can only become a join
point when a function `g`
becomes a join point as well correctly.
2. We introduce a second join point finding pass after specialisation
and before the following
simplification pass, as the specialiser might have introduced new join
point opportunities for
the simplifier to exploit.
Notably in the code from #10995 we now correctly detect the missing join
point which required both
of these changes to be made.
Closes: #10995
This PR extracts some refactorings from #10763, including dropping dead
code and not failing in `inaccessibleAsCtor`, which leadas to (slightly)
better error messages, and also on the grounds that the failing
alternative may actually be unreachable.
This PR inlines several Decidable instances for performance reasons.
Unlike the previous #10934 it does not attempt to also simplify the
Decidable instance system as
that has proven to have non trivial performance impact.
Co-authored-by: Rob23oba <152706811+Rob23oba@users.noreply.github.com>
This PR defines `String.Slice.replace` and redefines `String.replace` to
use the `Slice` version.
The new implementation is generic in the pattern, so it supports things
like `"education".replace isVowel "☃!" = "☃!d☃!c☃!t☃!☃!n"`. Since it
uses the `ForwardSearcher` infrastructure, `String` patterns are
searched using KMP, unlike the previous implementation which had
quadratic runtime. As a side effect, the behavior when replacing an
empty string now matches that of most other programming languages,
namely `"abc".replace "" "k" = "kakbkck"`.
This PR fixes the KMP implementation, which did incorrect bookkeeping of
the backtracking process, leading to incorrect starting ranges of
matches.
The new implementation does not require `partial` anywhere.
This PR adds support for specifying anchors to restrict the search space
in `grind` when using `grind only`. Anchors can limit which case splits
are performed and which local lemmas are instantiated.
This PR adds the `set_config` tactic for setting `grind` configuration
options. It uses the same syntax used for setting configuration options
in the `grind` main tactic.
This PR tries to preserve names of pattern variables in match
alternatives in `decreasing_by`, by telescoping into the concrete
alternative rather than the type of the matcher's alt. Fixes#10976.
This PR ensures that searching for an empty string returns the expected
pattern of alternating size-zero matches and size-one rejects.
In particular, splitting by an empty string returns an array formed of
the empty string, all of the string's characters as singleton strings,
followed by another empty string. This matches the [Rust
behavior](https://doc.rust-lang.org/std/primitive.str.html#method.split),
for example.
This PR adds inline annotations to several `Decidable` instances.
Additionally, it removes the `Decidable` instance for `p → q` which is
made redundant by `forall_prop_decidable`.
This PR changes the closure allocator to use the general allocator
instead of the small object one.
This is because users may create closures with a gigantic amount of
closed variables which in turn
boost the size of the closure beyond the small object threshold.
This issue was uncovered by #10979. Detecting that the small object
threshold is at fault requires
building mimalloc in debug mode at which point it yields:
```
mimalloc: assertion failed: at "/home/henrik/lean4/build/debug/mimalloc/src/mimalloc/src/alloc.c":132, mi_heap_malloc_small_zero
assertion: "size <= MI_SMALL_SIZE_MAX"
```
The generated code at fault here looks as follows:
```c
LEAN_EXPORT lean_object* l_initExec___at___00res_spec__0(lean_object* x_1) {
_start:
{
lean_object* x_2; lean_object* x_3; lean_object* x_4; lean_object* x_5; lean_object* x_6; lean_object* x_7; lean_object* x_8; lean_object* x_9; lean_object* x_10; lean_object* x_11; lean_object* x_12; lean_object* x_13; lean_object* x_14;
x_2 = lean_alloc_closure((void*)(l_initializer_ext___at___00initExec___at___00res_spec__0_spec__0___lam__0___boxed), 3, 0);
x_3 = l_initExec___redArg___closed__0;
x_4 = l_initExec___redArg___closed__1;
x_5 = l_instMonadLiftNonDetT___closed__0;
x_6 = l_initExec___redArg___closed__2;
x_7 = l_initExec___at___00res_spec__0___closed__0;
lean_inc_ref(x_2);
x_8 = lean_alloc_closure((void*)(l_initExec___at___00res_spec__0___lam__29___boxed), 213, 212);
lean_closure_set(x_8, 0, x_3);
lean_closure_set(x_8, 1, x_2);
lean_closure_set(x_8, 2, x_4);
lean_closure_set(x_8, 3, x_3);
lean_closure_set(x_8, 4, x_4);
lean_closure_set(x_8, 5, x_3);
lean_closure_set(x_8, 6, x_4);
lean_closure_set(x_8, 7, x_3);
lean_closure_set(x_8, 8, x_4);
lean_closure_set(x_8, 9, x_3);
lean_closure_set(x_8, 10, x_4);
lean_closure_set(x_8, 11, x_3);
lean_closure_set(x_8, 12, x_4);
lean_closure_set(x_8, 13, x_3);
lean_closure_set(x_8, 14, x_4);
lean_closure_set(x_8, 15, x_5);
lean_closure_set(x_8, 16, x_6);
lean_closure_set(x_8, 17, x_5);
lean_closure_set(x_8, 18, x_5);
lean_closure_set(x_8, 19, x_5);
lean_closure_set(x_8, 20, x_5);
lean_closure_set(x_8, 21, x_5);
lean_closure_set(x_8, 22, x_5);
...
```
With the crash happening in `lean_alloc_closure` where we
unconditionally invoke the small allocator
which cannot cope with closures this large. Hopefully changing this to
the general purpose allocator
doesn't have too much of an impact on performance.
Closes: #10979
This PR adds the basic infrastructure to perform termination proofs
about `String.ValidPos` and `String.Slice.Pos`.
We choose approach where the intended way to do termination arguments is
to argue about the position itself rather than some projection of it
like `remainingBytes`.
The types `String.ValidPos` and `String.Slice.Pos` are equipped with a
`WellFoundedRelation` instance given by the greater-than relation. This
means that if a function takes a position `p` and performs a recursive
call on `q`, then the decreasing obligation will be `p < q`. This works
well in the common case where `q` is `p.next h`, in which case the goal
`p < p.next h` is solved by the simplifier.
For stepping through a string backwards, we introduce a type synonym
with a `WellFoundedRelation` instance given by the less-than relation.
This means that if a function takes a position `p` and performs a
recursive call on `q` and specifies `termination_by p.down`, then the
decreasing obligation will be `q < p`. This works well in the case where
`q` is `p.prev h`, in which case the goal `p.prev h < p` is solved by
the simplifier.
For termination arguments invoving multiple strings, the lower-level
primitive `p.remainingBytes` (landing in `Nat`) is also available.
In a future PR, we will additionally provide the necessary typeclasses
instances to register `String.ValidPos` and `String.Slice.Pos` with
`grind` to make complex termination arguments more convenient in user
code.
This PR implements the following `grind` improvements:
1. `set_option` can now be used to set `grind` configuration options in
the interactive mode.
2. Fixes a bug in the repeated theorem instantiation detection.
3. Adds the macro `use [...]` as a shorthand for `instantiate only
[...]`.
This PR adds the combinator ` · t_1 ... t_n` to the `grind` interactive
mode. The `finish?` tactic now generates scripts using this combinator
to conform to Mathlib coding standards. The new format is also more
compact. Example:
```lean
/--
info: Try this:
[apply] ⏎
instantiate only [= mem_indices_of_mem, insert, = getElem_def]
instantiate only [= getElem?_neg, = getElem?_pos]
cases #f590
· cases #ffdf
· instantiate only
instantiate only [= Array.getElem_set]
· instantiate only
instantiate only [size, = HashMap.mem_insert, = HashMap.getElem_insert, = Array.getElem_push]
· instantiate only [= mem_indices_of_mem, = getElem_def]
instantiate only [usr getElem_indices_lt]
instantiate only [size]
cases #ffdf
· instantiate only [=_ WF]
instantiate only [= getElem?_neg, = getElem?_pos, = Array.getElem_set]
instantiate only [WF']
· instantiate only
instantiate only [= HashMap.mem_insert, = HashMap.getElem_insert, = Array.getElem_push]
-/
#guard_msgs in
example (m : IndexMap α β) (a a' : α) (b : β) (h : a' ∈ m.insert a b) :
(m.insert a b)[a'] = if h' : a' == a then b else m[a'] := by
grind => finish?
```
This PR ensures that model-based theory combination in `grind cutsat`
considers nonlinear terms. Nonlinear multiplications such as `x * y` are
treated as uninterpreted symbols in `cutsat`.
Closes#10885
This PR adds support for scientific literals for `Rat` in `grind`.
`grind` does not yet add support for this kind of literal in arbitrary
fields.
closes#10489
