Test end.offset <= 1 before revTruncLeOneLow in the reversed crossed
dispatch, and specialize revTruncLeOneLow under that path fact. The
helper drops the now-tautological end.offset conjuncts and the key
comparison implied by a failed optLe.
The crossed truncated reverse benchmark drops from about 67 ns to
about 61 ns. Sweep, corpus, SMT cases, and 100% line and branch
coverage still pass.
Drop the redundant begin.offset split in crossedCompleteOvershoot and
the rE > 0 guard in closedFormForwardSaturated; both are implied by
the call-site invariants.
Add an SMT-derived empty-interval case as the smt_cases CTest and as a
committed fuzz_readable seed. run-fuzz.sh now loads the read-only seed
corpus alongside the writable one.
Rename exhaustive to sweep and document that it is exhaustive only
over its bounded family of instances.
- Correct stale references and headline-path wording in READMEs and Lean docstrings
- Document build steps, comparator requirements, sweep scope, and coverage
- Unify Apache headers across Lean and C++ sources
- Add a bwrap-free axiom audit for the headline theorems
- Add a Gitea workflow running lake build, the axiom audit, C++ tests, and coverage
- Match LLVM coverage tool versions to the compiler and recreate stale build dirs
Drop predicate operands and unreachable arms that dispatch and readable invariants fix, and extend the exhaustive sweep to cover every branch side. Gate coverage on branch coverage as well.
Add an exhaustive readable-instance sweep and a coverage gate requiring
100% line coverage of the summarised implementation. Delete branches that
are unreachable under `Readable`, centralise and assert the 32-bit offset
contract at the public entry points, and stop materialising `n` zero bytes
when deciding `keyAfterN n k < upper`.
Also add the extreme-offset regression test, harden the corpus parsers,
remove unused model helpers, document the literal-total / fast-readable
split, and refresh the validation and benchmark numbers.
- Literal transliteration of `Statement.closedForm` plus an equivalent
summarised implementation
- Borrowed input keys (`std::string_view` / `std::span`) and owned result
keys (`std::string`)
- Golden-vector test, differential/safety fuzzers, and a nanobench
comparison
The literal and summarised implementations agree on 40,000 Lean golden
vectors, 18,971 sorted corpus cases, and the readable fuzzer; the raw fuzzer
runs both implementations under ASan/UBSan. `build/` and the generated
`oracle/golden.txt` are ignored.
Collapse duplicated cells and share their conditions:
- Merge the crossed complete clamp cells. They all fall through to
`end_.key`, and only the bottom / infinite / head-bottom / last-ge cells
conflict with the long nonpositive cell, so test `negCrossClampNonposLastGe`
explicitly and let the rest share the fall-through.
- Allow `r.end_.offset ≤ 1` in the crossed truncated nonpositive cell and
merge it with the old `...Small` cell into `negCrossTruncRevCrossedLow`.
- Use `overshoot r res` in `negCrossOvershoot` and
`negCrossTruncRevBelowEnd` instead of spelling out the same key bound.
- Drop the `r.begin.key ≤ r.end_.key` shortcut in `emptyValue`; the
cardinality branch below already returns `bottom..end_.key` in that case.
- Inline the single-use helpers: `...TruncRevBelowEnd`,
`...TruncRevCrossedNonposLarge`, `...ResidualFiniteUnder`,
`...ResidualTopLargeSentinel`, `...TruncRevUnder` and
`...TruncTopTopEq`.
`Bridge.lean` follows the removed cells, keeping the deleted reference
predicate as a private legacy abbreviation for the bridge proof.
`lake build` is clean.
On a clamped nonpositive begin the reference clamped rule `compactClamped` and
the positive rule `compactPositive` coincide: the begin offset is nonpositive,
so `residualB = 0`, and the reversed-truncated rule collapses to its no-skip
case.
So `closedForm` now dispatches on `1 ≤ r.begin.offset ∨ clampedNonpos r res`
and drops `closedFormClamped`, `closedFormNonposRevTrunc` and
`nonposRevReduced`, together with their bridges. `closedForm_eq` splits
explicitly on the positive and clamped regions, using the new reference lemmas
`ref_closedFormNonposRevTrunc_eq` and `ref_compactClamped_eq_compactPositive`.
The headlines, their statements and the comparator inputs are unchanged.
- Define `ConflictRange.toSet`, the set of keys lying in a range
- Add the headline `MinimalUnique`: any two minimal sound ranges for the
same read and result denote the same set of keys
- Transport `FormalConflictRange.Minimal.toSet_eq` across the bridge; the
statement is unconditional, since `Minimal` already carries soundness
- State `Statement.minimalUnique` in the challenge and add it to the
comparator's `theorem_names`
`lake build` is warning-free and `lake comparator` accepts both headlines.
- Move to lean-toolchain v4.35.0 and mathlib v4.35.0
- Opt out of the module-system warning with `allowNonModules`
- Adapt to Lean/Mathlib 4.35 changes: drop now-redundant
`List.filter_congr` blocks and no-op `ring` calls
- Add `Challenge.lean`, `Solution.lean`, and `comparator.json`
`lake comparator` now accepts the solution against the trusted challenge.
- Update deprecated lemmas (ite_eq_left/right, ENat casts, List.Sublist, ...)
- Replace push_neg with push Not and resolve unused-variable warnings
- Convert flexible simp/simp_all calls to explicit simp only lists
- Remove unscoped maxHeartbeats options, open scoped Classical, and unused tactics
- Add missing module docstrings and clean up style linter hits
- Wrap over-long lines and flatten the deeply nested
minimal_conflictRangeClosedForm_total proof
`lake build` now completes with zero warnings.
- Vendor the FormalConflictRange reference library needed by the proof
- Add Bridge.lean with simp lemmas connecting the Statement definitions
to the reference library
- Prove Statement.closedFormMinimal in Proof.lean
- Register the new libraries in the Lake config