summaryrefslogtreecommitdiff
path: root/backends/lean (follow)
Commit message (Collapse)AuthorAgeFilesLines
* Merge branch 'main' into prop-has-imp-sortSon HO2024-06-125-45/+118
|\
| * Revert "Update the scalar notation for the Lean backend"Son Ho2024-06-121-25/+25
| | | | | | | | This reverts commit c8272aeea205ca9cb36e22757473ca2a931a4933.
| * Revert "Update CoreConvertNum.lean"Son Ho2024-06-121-22/+22
| | | | | | | | This reverts commit fe59d3afe131e9f83cbdf73b1c8089ad090d92fb.
| * Update CoreConvertNum.leanSon Ho2024-06-121-22/+22
| |
| * Update the scalar notation for the Lean backendSon Ho2024-06-121-25/+25
| |
| * Deactivate the coercion from Nat to ScalarSon Ho2024-06-122-0/+17
| |
| * Slightly simplify the progress tacticSon Ho2024-06-121-8/+4
| |
| * Add wrappers around the rewriterSon Ho2024-06-121-0/+61
| |
| * Add the Simp.Config to the simp wrappersSon Ho2024-06-124-37/+36
| |
* | feat: `PropHasImp` can have `Sort u` as premissesRyan Lahfa2024-06-111-1/+1
|/ | | | | | | This makes it possible to have `InBounds ... : Type 1` for example as `x`. Signed-off-by: Ryan Lahfa <ryan.lahfa@inria.fr>
* feat: add small pieces of max theoryRyan Lahfa2024-05-241-0/+39
| | | | | | | | | `0#ty` is neutral for `max` for unsigned integers. Without the `Fact` instances, those theorems are not as automatic as they could be. Signed-off-by: Ryan Lahfa <ryan.lahfa@inria.fr>
* Update scalar_tac to use omega instead of linarithSon Ho2024-05-244-104/+82
|
* Fix an issue in the progress tacticSon Ho2024-05-211-0/+13
|
* feat(backends/lean): make `max`-related coercions nicerRyan Lahfa2024-05-211-0/+12
| | | | | | | | | | Situations where you have `coe (max a b) = max (coe a) (coe b)` are often stuck during verification because of the lack of this theorem. With this theorem, `push_cast` works as intended and normalizes even further. Signed-off-by: Ryan Lahfa <ryan.lahfa@inria.fr>
* Merge pull request #164 from RaitoBezarius/ofnat_scalarSon HO2024-05-141-0/+5
|\ | | | | | | feat(backend/lean): Raw Lean literals can be parsed into scalars automatically
| * feat(backend/lean): Raw Lean literals can be parsed into scalars automaticallyRyan Lahfa2024-04-301-0/+5
| | | | | | | | | | | | | | We can use coercion to go to integers, build a scalar out of an int via unification and let the user clear up bounds or let Lean decide them. Signed-off-by: Ryan Lahfa <ryan.lahfa@inria.fr>
* | backends/lean: repair definition of `core.mem.replace`Ryan Lahfa2024-05-131-1/+4
|/ | | | | | Oops, it is supposed to do something with the second argument! Signed-off-by: Ryan Lahfa <ryan.lahfa@inria.fr>
* Update the backend and ExtractBuiltin.mlSon Ho2024-04-251-1/+1
|
* Merge branch 'main' into core-option-unwrapSon Ho2024-04-251-0/+4
|\
| * Merge branch 'main' into option-takeSon Ho2024-04-251-0/+15
| |\
| * | compiler: map `core::mem::swap` to the pure swapRyan Lahfa2024-04-241-0/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | In the pure functional model, `swap` is mostly about borrow checking and should simplify to the pure swap in our backends. Other backends than Lean are not done in this commit. Signed-off-by: Ryan Lahfa <ryan.lahfa@inria.fr>
| * | compiler: map `core::option::Option::take` to identity functionRyan Lahfa2024-04-171-0/+2
| | | | | | | | | | | | | | | | | | | | | | | | `take` in a pure functional model is the identity function and everything related to borrow checking is handled by the forward/backward mechanism. Signed-off-by: Ryan Lahfa <ryan.lahfa@inria.fr>
* | | Add core::option::unwrap builtinZyad Hassan2024-04-243-5/+13
| |/ |/|
* | feat(backends/lean): scalars form a linear orderRyan Lahfa2024-04-231-0/+15
|/ | | | | | | More than c1c33de8, actually, scalars form a linear order with a decidable ≤ operation which is induced by the integer (Z) model. Signed-off-by: Ryan Lahfa <ryan.lahfa@inria.fr>
* lean: scalars form a preorderRyan Lahfa2024-04-121-0/+10
| | | | | | | | | | Via the canonical injection, we can easily define an induced preorder on scalars and inherit all nice properties. It's useful to reason on specific scalar preorders w.r.t. Ordering, see the binary search tree verification example. Signed-off-by: Ryan Lahfa <ryan.lahfa@inria.fr>
* Add more definitions to the Lean librarySon Ho2024-04-124-0/+127
|
* Reorganize the files in the Lean backend a bitSon Ho2024-04-125-241/+266
|
* Start adding integer functions to the Lean librarySon Ho2024-04-123-3/+266
|
* Fix a proofSon Ho2024-04-121-1/+0
|
* Update the core.num.checked_* definitionsSon Ho2024-04-121-31/+233
|
* Merge branch 'son/clean' into checked-opsSon Ho2024-04-1113-240/+241
|\
| * Merge remote-tracking branch 'origin/main' into son/cleanSon Ho2024-04-1110-23/+26
| |\
| | * Update the lean toolchain and fix the proofsSon Ho2024-04-059-20/+12
| | |
| | * Fix the coerce notation for scalars and update some lemmasSon Ho2024-04-041-4/+15
| | |
| * | Rename Result.ret as Result.ok in the backendsSon Ho2024-04-048-217/+215
| |/
* / Add builtins for some checked ops such as checked_addZyad Hassan2024-04-031-0/+29
|/
* backend(/tests)/lean: 4.6.0-rc1 → 4.6.1Ryan Lahfa2024-03-251-1/+1
| | | | | | 4.6.0 has been released in https://github.com/leanprover/lean4/releases/tag/v4.6.0 Signed-off-by: Ryan Lahfa <ryan.lahfa@inria.fr>
* Fix a minor issueSon Ho2024-03-181-1/+3
|
* Regenerate the constants tests and update Primitives/Base.leanSon Ho2024-03-181-1/+1
|
* Fix an issue with the divergent encodingSon Ho2024-03-091-3/+27
|
* Fix tuple indexing for Lean backendZyad Hassan2024-03-081-1/+1
|
* Update the generation of constant bodies for LeanSon Ho2024-03-081-2/+2
|
* Update the tuples syntaxSon Ho2024-03-081-1/+3
|
* Update the tuples notationsSon Ho2024-03-083-51/+81
|
* Remove some commentsSon Ho2024-03-081-31/+0
|
* Update the handling of notations like #u32 or #isizeSon Ho2024-03-084-70/+104
|
* Update the notation for heterogeneous negationSon Ho2024-03-081-3/+12
|
* Introduce a notation for constant scalars in match patternsSon Ho2024-03-071-0/+29
|
* Add a notation for tuple field accesses in LeanSon Ho2024-03-071-0/+51
|
* Cleanup a bit Scalar.leanSon Ho2024-02-091-198/+198
|