-
Notifications
You must be signed in to change notification settings - Fork 0
Home
Source of truth: docs/echo-types/MAP.adoc
in the repo. This wiki mirrors it for browsing; it never forks content —
always trust the in-repo file.
flowchart TD
F["<b>Foundation</b><br/>Echo f y := Σ (x : A) , (f x ≡ y)<br/>= homotopy fibre"]
AD["<b>Pillars A–D</b> (establishment plan, LANDED 2026-05-17)<br/>A: echo↔fib · B: pullback UP · C: separating model · D: carrier-parametric"]
PF["<b>Pillar F gates</b> (earn-back, ALL PASSED)<br/>F1 graded-comonad witness · F2 2nd Echo-functor model<br/>F3 2nd non-iso grade-monoid · F4 funext-qualified pullback UP<br/>F5 funext-qualified full OFS"]
T1["<b>Tier 1 · Canonical identity layer</b> (2026-05-27, extended 2026-05-28)<br/>EchoTotalCompletion (A ≃ Σ B Echo f) · OFS-witness · Image · no-section-of-collapsing-map · ImageFactorizationProp (epi,mono earn-back)"]
T2["<b>Tier 2 · Classification grid</b><br/>LossTaxonomy (function-side) · ResidueTaxonomy (residue-side, 8 instances incl. Indexed, Cost, Search, Epistemic) · DecorationStructure · _≡m_"]
T3["<b>Tier 3 · Qualified universal property</b> (Gate F5)<br/>echo-factorisation-strict · diagonal lifting · factorisation uniqueness up to iso"]
AUD["<b>Audience surfaces</b><br/>EchoProvenance · EchoSecurity · EchoProbabilisticSupport · EchoDifferential"]
SUITE["<b>EchoCanonicalIdentitySuite</b><br/>(curated single-file entry point)"]
F --> AD
AD --> PF
F --> T1
T1 --> T2
T2 --> T3
PF -.->|F5 underpins| T3
T1 --> AUD
T2 --> AUD
T3 --> AUD
T1 --> SUITE
T2 --> SUITE
T3 --> SUITE
AUD --> SUITE
- MAP.adoc — master content map (status-tagged).
-
EchoCanonicalIdentitySuite.agda— curated single-file Agda entry point. - universal-property.adoc — categorical-UP arc end-to-end (pullback + F4 + F5/OFS), with ASCII diagrams.
-
fibration-package.adoc — fibration-side arc (
map-over+ composition + cancellation + pentagon). - paper.adoc — Pillar E paper (§"Post-establishment structural extensions" weaves Tier-1+2+3).
- retractions.adoc + earn-back-plan.adoc — honesty + Pillar F gate ledger.
Foundation: Echo functor / loss-with-residue [REAL]
Canonical identity layer [REAL] (2026-05-27): the named-theorem
load-bearing artefacts that promote Echo from a useful construction
to a named structural object — the canonical fibre-completion
functor + one leg of the (equivalence, projection) orthogonal
factorisation system on Type. K-free skeletons under
--safe --without-K; full HoTT-strength upgrades documented as
the next earn-back gates (Pillar F4 template).
-
Total completion —
A ≃ Σ B (Echo f)(EchoTotalCompletion.agda). Every irreversible map's domain is canonically equivalent to the total space of its echoes. -
Orthogonal factorisation system
(
EchoOrthogonalFactorizationSystem.agda). Everyf : A → Bfactors asA ─encode→ Σ B (Echo f) ─proj₁→ B; left leg an equivalence, right leg a projection whose fibre atyisEcho f y. -
Image factorisation
(
EchoImageFactorization.agda).Image f := Σ B (Echo f); three classifications (Surjective, Injective, projection-uniqueness). The (epi, mono) truncation form lands asEchoImageFactorizationProp.agda(2026-05-28):TruncInterfacerecord packaging the four∥_∥obligations +module ImagePropwith mono right leg (is-propon truncated second component) + mere-surjective left leg (recinto truncated preimage Σ). Module-parameterised inTruncInterface— consumer plugs in Cubical∥_∥₋₁/ HIT / postulated interface; this module stays--safe --without-K, zero postulates. -
Generic no-section
(
EchoNoSectionGeneric.agda). For any collapsing map with two distinct elements landing on the same image, no section exists — structural fact, not an example. -
Loss taxonomy (function-side)
(
EchoLossTaxonomy.agda). Four-axis classification: EQUIV ⇒ centre + projection-unique; INJ ⇒ projection-unique; SURJ ⇒ inhabited; CONST ⇒ section. -
Residue taxonomy (residue-side)
(
EchoResidueTaxonomy.agda).record ResidueForm f Rpackaging the per-output residue carrier + lowering shared by the eight decoration modules. All eight decoration modules now packaged (2026-05-28 completion): trivial (⊤) / identity (Echo f) / generic Σ-cert (echoR-residue) / linear-affine / Indexed / Cost (parametric over CostAlgebra) / Search (parametric over search-completeness witness, mirrors Cost) / Epistemic (identity-residue (obs r)— indistinguishability class IS the fibre). -
Decoration structure (observation-side)
(
EchoDecorationStructure.agda).record DecorationStructure G— seven-field recipe (order + refl + trans + prop + join + UMP) shared across the eight decoration modules. Four instances: Graded (3-grade), Linear (2-grade), Access (5-grade), Choreo (2-role). Plusmodule DegradeAbstractproving abstractdegrade-compose+degrade-via-joinonce over the record. -
Observational equivalence
(
EchoObservationalEquivalence.agda). Mode-indexed equality_≡m_onLEcho. Linear witness-aware; affine witness-blind. Headline pins a linear-distinct / affine-equal pair as a checked theorem.
Qualified universal property (Pillar F, Gate F5 FULL PASS,
2026-05-27): the (equivalence, projection) orthogonal factorisation
system on Type, earned back at the qualified level via three slices
in the F4 template (funext as explicit module parameter, never a
postulate; unconditional pointwise corollaries kept funext-free):
-
F5-1 strict factorisation triangle
(
EchoOFSUnivF5.agda) -
F5-2 diagonal lifting property
(
EchoOFSUnivF5Diag.agda) -
F5-3 factorisation uniqueness up to iso, via composition
φ.to = encode f ∘ g⁻¹(EchoOFSUnivF5Iso.agda).
Retraction follow-up F-2026-05-27a logged in
docs/retractions.adoc.
Full Pillar F gate ledger:
docs/echo-types/earn-back-plan.adoc.
Audience-facing modules (Tier 3, all landed 2026-05-27 per GPT-recommended order; each ships the abstract record + parametric theorems + worked instance + honest-bound matched-negatives):
-
Provenance
(
EchoProvenance.agda).Provenancerecord + four parametric theorems. Workedbool-over-nat-provenanceinstance. -
Security
(
EchoSecurity.agda).Securityrecord + per-region collapse + audit-no-recovery (factored throughno-section-of-collapsing-map). Workedregion-exit-audit-instance. Matched-negatives: NOT bytes-zeroed / side-channel / tamper-evident / oracle. -
ProbabilisticSupport
(
EchoProbabilisticSupport.agda).Samplingrecord + marginal non-injectivity + Echo-carries-index + residue-loses-index. Matched-negatives: NOT measure-preserving / probability monad / Kantorovich / coupling / randomness extraction. -
Differential
(
EchoDifferential.agda).Sensitivityrecord + blur non-injectivity + Echo-carries-perturbation- residue-loses-perturbation. Matched-negatives: NOT ε-DP / Lipschitz / noise calibration / adversary model.
Curated suite — single-file entry point:
EchoCanonicalIdentitySuite.agda.
Re-exports the load-bearing headlines from every Tier-1 / Tier-2 /
Tier-3 module in the "why Echo deserves a name" reading order.
Consolidation docs (2026-05-27): two doc-only narratives
threading source modules into end-to-end stories with ASCII
diagrams:
universal-property.adoc
(pullback + F4 + F5 / OFS arc) and
fibration-package.adoc
(map-over + composition + cancellation + pentagon).
Plus two cementing matched-negatives against shadow encodings:
Shannon-entropy non-distinguishing
(EchoEntropy.agda)
and the linear-logic !A := 1 shallow-encoding gap
(EchoLLEncoding.agda).
Directions: Thermodynamic (Landauer/Bennett) [REAL*] ·
Buchholz/Veblen ordinals [REAL] (head-Ω domination LANDED
2026-05-27 PRs #130/#131/#133/#134; Slice 3 closes at ψ-rank-level
2026-05-28 via PRs #137 prereqs + #141 headline + #142 _<ᵇ¹_
umbrella + #143 RankLexSlice3 ψ-rank discharge — bplus-chain-level
extension to rank-pow structurally blocked under current Brouwer
arithmetic, documented in RankLexSlice3.agda; Slice 4 narrowed
umbrella _<ᵇ⁻ⁿ_ LANDED 2026-05-28 via PR #149
(RankMonoUmbrellaSlice4) covering 10 inherited _<ᵇ⁰_ cases plus
the strict-head joint-bplus; rank-lex pivot for the
bpsi-source-at-equality sub-case scaffolded via PR #147
(RankLexJointBplus); two shortfalls pinned as ⊤-aliases
(<ᵇ-ψΩ≤ boundary at lex-rank only; <ᵇ-+1 at equal-head gated on
pivot completion); unbudgeted wf-<ᵇʳᶠ + full-order
internalisation remain) ·
Characteristic/EI — epistemic, choreographic, roles [REAL] ·
CNO/absolute-zero [REAL] · JanusKey [REAL] ·
Tropical/Dyadic [REAL*] · MAA Framework [ROADMAP] ·
Shannon/information theory [REAL*] (discrete non-distinguishing
landed; real-valued / mutual-information open) ·
Decoration Bridge (R5 exploratory, PR #129).
Tutorial / pedagogy [REAL]: three worked walkthroughs under
tutorial/
— certified region-exit audit (Walkthrough 1, killer-app candidate),
epistemic erasure (Walkthrough 2, KDF / no-section-via-residue),
provenance / debugging (Walkthrough 3, section/no-section contrast
across three residue layers). Each carries an honest-bound
disclosure at the top and a matched-negative block at the bottom.
Tooling: ArghDA proof-ladder [REAL*]
Establishment track (Pillar E): internal programme (Pillars A–D) complete since 2026-05-17 within R-2026-05-18 narrowings. Paper
- TYPES extended abstract under
docs/echo-types/. Offline plan (venue + Zenodo + outreach) atpillar-e-offline.adoc. Paper §"Post-establishment structural extensions" weaves the 2026-05-27 Tier-1+2+3 spine into the central argument.
Governance: retraction R-2026-05-18 + earn-back ledger (F1–F5 ALL PASSED) — read before citing any claim as final.
Executable finite-domain companion — runnable shadow of the
Agda library, NOT a proof. Each Julia testset = finite shadow of
one named Agda lemma. Latest pin: echo-types eed4250 (2026-05-28).
- Repo: hyperpolymath/EchoTypes.jl
- Current release:
v0.3.0(Tier-3 audience-facing spine + cementing LL-encoding gap) - Hands-on exercises bridging both repos: Julia-Companion-Exercises — walk through each tier with a Julia REPL session, an Agda-side link, and an "extend this" prompt.
The companion explicitly does NOT mirror:
- Funext-qualified surfaces (F5 strict / pullback strict, F4 template) — Julia has no funext.
- Retracted (R-2026-05-18) framings (graded-comonad, conservativity, two-models, universal-property).
- HoTT-strength upgrades (UIP, propositional truncation, HITs).
- Ordinal-lane work (Slice 3/4,
RankPow*,HeadOmegaInversion,RankLexJointBplus) — separate pillar; would need a newBordcarrier design.
The mirror covers exactly the K-free Tier 1+2+3 spine + the cementing matched-negatives (Entropy, LL-encoding gap). 253 passing test assertions across 18 testsets.
Edit the in-repo MAP.adoc (via a PR), not this page.