max/cubical-transport-hott-lean4

Fork 0

Lean 4 cubical-transport HoTT engine + Rust naga-IR FFI. The backend that powers topolei. Cells-spec implementation; the proof IS the implementation.

Find a file

Maximus Gorog 567d8722d5 Some checks are pending Lean Action CI / build (push) Waiting to run Details Refactor Phase 4: Rust ABI v6 → v7 (modal tag unification) Mirrors the Lean-side modal unification (commits `6e4936d` / `cfabca3` / 2b7564e) at the native level. ABI version constant CUBICAL_TRANSPORT_ABI_VERSION = 7. Tag unification (15 v6 modal tags → 5 v7 unified tags): · TY_MODAL = 9 (was: TY_FLAT/SHARP/SHAPE = 9-11) · TERM_MODAL_INTRO = 17, TERM_MODAL_ELIM = 18 (was: TERM__INTRO/ELIM = 17-22) · VAL_VMODAL_INTRO = 12 (was: VAL_V_INTRO = 12-14) · NEU_NMODAL_ELIM = 12 (was: NEU_N*_ELIM = 12-14) · MODKIND_FLAT/SHARP/SHAPE = 0/1/2 (u32 — matches existing tag- namespace convention) Tag-id design choice (TERM_MODAL_ELIM = 18, not 20): Lean compiler assigns constructor indices in declaration order; Syntax.lean declares modalIntro then modalElim consecutively, so the runtime tag is 18, not 20. Using 20 would break FFI marshalling. Same precedent as v4's sigma/path tag ordering. Reserved freed slots (no v7 reassignment per directive — gaps documented in tags.rs for future v8+ extensions): · CType: 10, 11 · TERM: 19, 20, 21, 22 · VAL: 13, 14 · NEU: 13, 14 Layout tables in cubical_transport.h v7 entry: · TY_MODAL [ℓ, k, A] · TERM_MODAL_INTRO [k, a] · TERM_MODAL_ELIM [k, f, m] · VAL_VMODAL_INTRO [k, inner_value] · NEU_NMODAL_ELIM [k, eliminator_value, stuck_scrutinee] Cascade across: · tags.rs — 15 deletions + 5 unified consts + 3 ModalityKind + reservation block · value.rs — CVal::VModalIntro(ModalityKind, _) and CNeu::NModalElim(ModalityKind, _, _) variants · eval.rs — 9-arm modal dispatch → 3-arm; β-rule fires only on matching kinds (k == k') with marker neutral on mismatch · subst.rs — 9 modal substDim arms → 3 unified · readback.rs — 6 modal readback arms → 2 · dim_absent.rs — 9 dim-absent modal arms → 3 Files NOT touched (verified): transport.rs, composition.rs, glue.rs, beta.rs, ffi.rs. Wildcard catchalls handle modal types as stuck neutrals (same as v6). Refcount discipline for the new k field documented at every mk_vmodal_intro / mk_nmodal_elim / mk_ty_modal call site. Verification: · cargo build --release: clean · cargo test --release: 0 passed, 0 failed (no in-crate tests) · lake build: 48 jobs PASS · lake build CubicalTransport: 43 jobs PASS · lake exe cubical-test: 49/49 + 46/46 = 95/95 PASS · 0 new unsafe / unimplemented! / todo! / panic! catchalls Net: -58 lines across 7 files (-397 deletions, +339 insertions). Source-code reduction concentrated in cascade modules; tags.rs and header grew due to reservation documentation and v7 layout tables. Honest gap: existing test suite doesn't exercise modal terms via FFI (modal Lean-side use sits behind sorry'd Phase 3 axioms). The 95/95 confirms no regression of existing paths but doesn't directly validate v7 modal marshalling. An FFI smoke test exercising modal terms is queued for when modal cohesion sorries discharge. Layer 0 + elegance refactor pass COMPLETE. All four phases landed: Phase 1 (derive_reflect_reify macro), Phase 2 (Lean modal unification), Phase 3 (Modal.lean rewrite + topolei consumer cascade), Phase 4 (Rust ABI v7). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>		2026-05-06 02:25:10 -06:00
.forgejo/workflows	Add proprietary LICENSE; move CI to .forgejo/workflows/, run cubical-test	2026-04-27 23:04:16 -06:00
CubicalTransport	Refactor Phase 3a: Modal.lean rewrite — forModalityKind unification	2026-05-06 02:13:16 -06:00
docs	REL2 universe stratification + topolei naming cleanup + Rust ABI v4	2026-05-04 00:21:14 -06:00
native/cubical	Refactor Phase 4: Rust ABI v6 → v7 (modal tag unification)	2026-05-06 02:25:10 -06:00
.gitignore	Initial commit: topolei — cubical-transport HoTT in Lean 4 + Rust FFI	2026-04-27 20:40:45 -06:00
build.sh	Split: engine = cubical-transport HoTT only	2026-04-27 21:35:01 -06:00
CubicalBench.lean	REL2 universe stratification + topolei naming cleanup + Rust ABI v4	2026-05-04 00:21:14 -06:00
CubicalTest.lean	Split: engine = cubical-transport HoTT only	2026-04-27 21:35:01 -06:00
CubicalTransport.lean	Modal cascade Phase 3: Modal.lean module proper — Layer 0 complete	2026-05-05 23:28:01 -06:00
lake-manifest.json	Drop Infoductor dependency; cubical-transport is now pure cubical engine	2026-05-01 07:52:59 -06:00
lakefile.toml	REL2 universe stratification + topolei naming cleanup + Rust ABI v4	2026-05-04 00:21:14 -06:00
lean-toolchain	Initial commit: topolei — cubical-transport HoTT in Lean 4 + Rust FFI	2026-04-27 20:40:45 -06:00
LICENSE	Add proprietary LICENSE; move CI to .forgejo/workflows/, run cubical-test	2026-04-27 23:04:16 -06:00
NOTICE	Add NOTICE: AI tooling acknowledgement; proprietary rights stand on human-directed whole	2026-04-27 23:10:32 -06:00
README.md	REL2 universe stratification + topolei naming cleanup + Rust ABI v4	2026-05-04 00:21:14 -06:00

README.md

cubical-transport-hott-lean4

A Lean 4 implementation of cubical-transport homotopy type theory (CCHM-flavor), with a fast Rust kernel exposed through C ABI.

The Lean side defines the syntax, semantics, and soundness theorems. The Rust side discharges the per-step β-rules of the evaluator. Lean axioms are routed through @[implemented_by] to Rust functions that return Lean objects in the same shape Lean would have produced; the soundness layer (CubicalTransport/Soundness.lean) certifies the backend at the boundary, so the kernel speed of the Rust code preserves the Lean-level proofs.

What's here

CubicalTransport/ — 22 Lean modules for syntax, substitution, dimensional structure, faces, typing, evaluation (eval / value / readback), transport, Glue, composition, and the soundness theorems.
native/cubical/ — Rust kernel (#![no_std], dual-target native staticlib + cdylib, wasm32 cdylib).
CubicalTest.lean, CubicalBench.lean — engine smoke + property tests (62/62 passing) and microbenchmarks.

Consuming the engine (with permission)

This Software is proprietary. See LICENSE — no use is granted by virtue of the repository being public. The instructions below are for the copyright holder and any party with prior written consent from mgorog@gmail.com.

Add as a Lake dependency from another Lean 4 project:

[[require]]
name = "cubicalTransport"
path = "../cubical-transport-hott-lean4"   # or git = "..."

Then import CubicalTransport.Syntax, import CubicalTransport.Eval, etc. Link against native/cubical/target/release/libcubical_transport.a in your own moreLinkArgs so the FFI symbols resolve.

Build

(cd native/cubical && cargo build --release)
lake build
./.lake/build/bin/cubical-test    # 62/62 tests pass

Reference

docs/FFI_DESIGN.md — C ABI contract between Lean and the Rust kernel.
docs/FFI_COMPLETENESS.md — per-function axiom audit.
docs/KERNEL_BOUNDARY.md — what this delivers in unmodified Lean 4 vs. what would need upstream Lean kernel work.
docs/NUMERICAL.md — numerical implementation principles.
docs/PHASE1_HISTORY.md — original transport/composition formalisation plan (archived; Phase 1 closed; preserved as methodology template for future phases).
docs/ZIGZAG_PORT.md — step-by-step Lean port plan for the n-category combinatorial engine (Phase 2+ higher-cell backend). Lands here in the engine. Read the cascade caveat at the top before editing: changes in the ported zigzag layer cascade to the sibling topolei repo.
native/cubical/README.md — Rust crate build + dev guide.
native/cubical/WASM.md — wasm32 ABI integration contract.

Used by

max/topolei — the cells-spec workspace interface, built on this engine. See its docs/cells-spec.md and docs/STATUS.md for the application-side picture.