e26ada2fbc
Some checks are pending
Lean Action CI / build (push) Waiting to run
The six generic methodology / repo-organization modules (Meta / Edit / Restructure / MacroAlias / MetaPath / Methodology) move out of CubicalTransport/Algebra/ into the new Infoductor repo at http://maxgit.wg:3000/max/infoductor. cubical-transport-hott-lean4 now `require`s `infoductor` from that forgejo URL. Imports updated: - import CubicalTransport.Algebra.X → import Infoductor.Foundation.X - open CubicalTransport.Algebra → open Infoductor Files that stay (cubical-domain-specific): - CubicalTransport/Algebra/EngineMethodologies.lean (cubical closing-form @[methodology] tags) - CubicalTransport/Algebra/Test.lean (integration tests) Files deleted (moved to Infoductor.Foundation): - CubicalTransport/Algebra/Meta.lean - CubicalTransport/Algebra/Edit.lean - CubicalTransport/Algebra/Restructure.lean - CubicalTransport/Algebra/MacroAlias.lean - CubicalTransport/Algebra/MetaPath.lean - CubicalTransport/Algebra/Methodology.lean Architecture rationale (per memory: "Infoductor — generic methodology / repo-organization project"): - Foundation primitives are domain-agnostic; anyone can register their own methodology atop them, regardless of cubical interest. - Cubical-transport keeps the question-form (CompQ etc.) and cubical-specific @[methodology] / @[metaPath] decls. - topolei (next, separate work) will consume both Infoductor.Foundation and cubical-transport, picking cubical as its methodology. - "Info-ductor" — conducts information through a codebase; pairs with Pantograph (the conductor sits atop the pantograph hardware on an electric train). 93/93 tests pass (47 smoke + 46 property). 53 build jobs total (43 cubical + 10 Infoductor.Foundation + linker stages). No new axioms, no behavioural change — pure code-organization refactor. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
85 lines
3.9 KiB
Text
85 lines
3.9 KiB
Text
/-
|
|
CubicalTransport.Algebra.EngineMethodologies
|
|
============================================
|
|
Engine-bound methodology registrations. Per ALGEBRA_PLAN.md §9:
|
|
|
|
> Every existing `eval_*` / `vTransp_*` / `vCompValue_*` / Glue /
|
|
> Soundness theorem has at least one corresponding `@[methodology]`
|
|
> registration that closes its representative question via
|
|
> `cubical_search`.
|
|
|
|
Each registration here is a *closing form* — a theorem that closes
|
|
a concrete-shape question goal directly, with its classifier
|
|
hypothesis discharged automatically (typically by `rfl` after
|
|
struct projection). This turns a hypothesis-bearing simp lemma
|
|
(`ask_of_full_face`) into a `cubical_search`-applicable closer
|
|
(`compq_top_concrete`).
|
|
|
|
Goals targeted:
|
|
- Full-face / empty-face / interval-line questions (all three
|
|
question shapes: CompQ, TranspQ, HCompQ).
|
|
- hcomp on `.pi` and on `.top` — closing forms with concrete body.
|
|
|
|
Out of scope here:
|
|
- Methodologies requiring non-trivial classifier discharge
|
|
(e.g., `ask_of_const_line` wants `¬IsFullFace ∧ ¬IsEmptyFace ∧
|
|
IsConstLine`). These would need a richer dispatch tactic; for
|
|
now `cubical_simp` covers them via the simp-routing path.
|
|
- Methodologies depending on Glue / Path body shape. The Glue
|
|
transport axioms have ~10 face-disjoint cases that each need
|
|
their own concrete closers; they land in a follow-up.
|
|
-/
|
|
|
|
import Infoductor.Foundation.Methodology
|
|
import CubicalTransport.Question
|
|
|
|
namespace CubicalTransport.Algebra.EngineMethodologies
|
|
|
|
open Infoductor
|
|
open Question
|
|
|
|
-- ── CompQ closers ───────────────────────────────────────────────────────────
|
|
|
|
/-- C1 closer: literal `.top` face on a CompQ → `eval env (u.substDim
|
|
binder .one)`. Classifier discharged by struct-projection rfl. -/
|
|
@[methodology IsCompQTopFace]
|
|
theorem compQ_top_face (env : CEnv) (i : DimVar) (A : CType) (u t : CTerm) :
|
|
(CompQ.mk env i A .top u t).ask = eval env (u.substDim i .one) :=
|
|
CompQ.ask_of_full_face _ rfl
|
|
|
|
/-- C2 closer: literal `.bot` face on a CompQ → `eval env (.transp
|
|
binder body .bot t)`. -/
|
|
@[methodology IsCompQBotFace]
|
|
theorem compQ_bot_face (env : CEnv) (i : DimVar) (A : CType) (u t : CTerm) :
|
|
(CompQ.mk env i A .bot u t).ask = eval env (.transp i A .bot t) :=
|
|
CompQ.ask_of_empty_face _ rfl
|
|
|
|
-- ── TranspQ closers ─────────────────────────────────────────────────────────
|
|
|
|
/-- T1 closer: literal `.top` face on a TranspQ → identity. -/
|
|
@[methodology IsTranspQTopFace]
|
|
theorem transpQ_top_face (env : CEnv) (i : DimVar) (A : CType) (t : CTerm) :
|
|
(TranspQ.mk env i A .top t).ask = eval env t :=
|
|
TranspQ.ask_of_full_face _ rfl
|
|
|
|
/-- T2 specialisation: any face on `.interval` → identity. -/
|
|
@[methodology IsTranspQIntervalLine]
|
|
theorem transpQ_interval_line (env : CEnv) (i : DimVar)
|
|
(φ : FaceFormula) (t : CTerm) :
|
|
(TranspQ.mk env i .interval φ t).ask = eval env t :=
|
|
TranspQ.ask_of_interval_line _ rfl
|
|
|
|
-- ── HCompQ closers ──────────────────────────────────────────────────────────
|
|
|
|
/-- HCompQ full-face closer: `vPApp tube .one`. -/
|
|
@[methodology IsHCompQTopFace]
|
|
theorem hcompQ_top_face (A : CType) (tube base : CVal) :
|
|
(HCompQ.mk A .top tube base).ask = vPApp tube .one :=
|
|
HCompQ.ask_of_full_face _ rfl
|
|
|
|
end CubicalTransport.Algebra.EngineMethodologies
|
|
|
|
-- (No `cubical_close` macro shipped here. A composite closer like
|
|
-- `simp; (rfl | cubical_search)` is a one-line user-side composition;
|
|
-- baking it in would impose an opinionated ordering and an
|
|
-- opinionated default tactic. Compose at the call site as needed.)
|