Skip to content

PeterHiggins19/higgins-decomposition

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

60 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Hˢ — Higgins Decomposition on the Simplex

A deterministic compositional inference instrument operating within Aitchison geometry on the simplex. Welcome — this repository is fully public and open for use, citation, and collaboration.

🎤 CoDaWork 2026 — follow along with the talk

In the room (or watching from anywhere) for Compositional monitoring of energy-mix drift on the simplex (Coimbra, Portugal · 1–5 June 2026)?

Open CODA-Association/CONFERENCE_ATTENDEES.md. It's a slide-by-slide follow-along, ordered to the talk, with every link you need open on your phone or laptop. Manuscript, deck, raw-output scroll, and an interactive 3-D projector you can run live in your browser — no install, no network call. If the screen is hard to see, or you're remote, the whole talk runs from that page.

🌐 Available in UN-6 locales — now a complete 2-side ambassador (v11). The community handout Higgins_Decomposition_Handout_CoDaCommunity.pdf is a single A4 sheet, both sides printed. Side 1 = the operationalization pitch (what / why / who / how to engage). Side 2 = the full operations reference (CoDa core operations, Hˢ supplementary operations, CNQ quaternion operations, closure constraints across domains, apparatus map of who reads what, symbols legend). Ships with Markdown twins in EN · FR · ES · RU · ZH · AR under CODA-Association/. Non-English locales are drafts pending native expert review per HCI-CNQ/wrappers/WRAPPER_SCHEMA.md §11.1.

🎯 Conference Status. Talk material complete and validated across five reviews; repository in formal pre-conference lockdown through 2026-06-06. See PRE_CONFERENCE_LOCKDOWN.md for what's locked and the S0-defect protocol. Conference authority folder: CODA-Association/CODAwork2026/. Recent change history: CHANGELOG.md. Live state: HS_FAST_REFRESH.json. Conference standards: HUF-STD-001 (Publication) · HUF-STD-002 (Tensor Train I/O) · HUF-STD-003 (Linear Algebra Foundations).

🎉 Public and publication-grade. Both engines (CNT v3.1.0 and CNQ v2.0.0) ship in four forms: Python reference, R reference, language-agnostic pseudocode (CNT_PSEUDOCODE.md + CNQ_PSEUDOCODE.md), and formal specification (HUF-STD-002). A 43-test suite plus three IEEE-floor confirmation datasets establish the determinism contract. The user entry point is PUBLICATION_READY.md; plain-English licensing in LICENSING.md. Free to use, free to cite, help is available — open an issue or follow the contact in PUBLICATION_READY.md.

🛡️ For skeptical users — TRUST_AND_VERIFICATION.md. Trust in open-source code is earned, not expected. If you do not yet trust the published Python or R, you do not have to run it. The engine is published in language-agnostic pseudocode you can re-implement in any language; three canonical reference inputs (Backblaze, Planck CMB, SM neutrino) have pinned content_sha256 values; your re-implementation should produce the same hash byte-identically. That is the verification protocol. The framework's central claim — that closure on the simplex is a real invariant — is testable at the implementation layer by the hash discipline, without you having to trust this repository at all.

Hˢ = R ∘ M ∘ E ∘ C ∘ T ∘ V ∘ S

Seven operators — Simplex closure, Variance trajectory, Transcendental squeeze, Classification, Entropy test, Mode synthesis, and Report — compose into a single decomposition function derived from a single axiom: same input, same output, always. Validated across 11 domains and 101 reference datasets (push #34 full-corpus suite), spanning 44 orders of magnitude. The instrument reads structure without creating or destroying it.

Validate Repository Code: Apache-2.0 Docs: CC BY 4.0 Licensing HCI-CNT engine 3.0.0 HCI-CNT schema 3.0.0 HCI-CNQ engine 2.0.0 HCI-AUDIO doctrine HCI-ULTRASOUND doctrine Reference experiments CodaWork 2026


At a Glance

Measure Value
Physical domains 18
Experiments 25
Distinct systems 36
Devices under test (DUTs) 53
Pipeline files 13
Interactive tools 9
Diagnostic codes 78
Structural modes 10
Transcendental constants 35
Conjugate pairs validated 13
Reference standards 15
Diagnostic-report languages 5 (en, zh, hi, pt, it) — tools/pipeline/locales/
UN-6 wrapper locales 6 (en, fr, es, ru, zh, ar) — HCI-CNQ/wrappers/ (push #32)
Languages supported (union) 9 (en, fr, es, ru, zh, ar, hi, pt, it)
Scale range 10⁻¹⁸ m to 10²⁶ m (44 orders of magnitude)
Framework version 3.0
Deterministic Yes (Gauge R&R bit-identical, SHA-256 verified)
Instrument metrology QUALIFIED (6/6 metrics pass)
License Code: Apache-2.0 (LICENSE) · Docs: CC BY 4.0 (LICENSE-DOCS)

What's New — May 2026

🌟 Flagship paper: The Isotropic Radiation Ground State and the Traction Engine (v2.2, 2026-05-22)

The unified-formula statement of the framework's foundation, consolidated against the Rogue-Wave-Audio archive cross-check. 31-page master standard linking BTL acoustic measurement experience to the present-day Hˢ simplex framework, with the full lemma chain in support (Banach fixed-point convergence, Helmholtz reciprocity, Rayleigh-Sommerfeld diffraction, Gershgorin invertibility, group-delay-as-rotation on S³, closure invariance under the log-ratio transform). Every component of equation (13) — the unified isotropic-radiation ground-state formula — has been measured at BTL continuously and never failed a closure check.

v2.2 consolidation folds in eight architectural details from the RWA archive cross-check: the HUF-GOV/HUF-CLS fork at ADAC, the Paired Measurement Doctrine, DADI as failure-direction diagnostic, date precision (DADC formal paper 2024-12-05; the November 2025 Grok-collaboration generalization moment where MC-4 was born), the non-monotonic H₁ abstraction path (concrete → abstract → back to concrete), the RWA concepts/ folder anticipations of HUF concepts, expanded acknowledgements, and §18 — "The recursion test" — documenting that v2.1 was reconstructed bottom-up by AI synthesis from public artefacts and v2.2 is the version where the recomposition agrees with the canonical record. The system sums to one.

The flagship answers a question reviewers have repeatedly asked: "why are you so confident?" The answer is now provable, not faith-based. The acoustic ground state (6.02 dB, 4π → 2π baffle-step transition) is the physical instance; the unified formula generalizes it via Theorem 2 to any compositional time-series with a conserved budget and a log-carrier. The CoDaWork 2026 energy-mix manuscript is the first non-acoustic application of an apparatus with continuous BTL empirical validation behind it. The framework is an extension of a human-machine partnership built up continuously across seven domains — acoustics, governance, electronics, robotics, X-ray procedural, mass production automation, and man-machine interface engineering. See AI_AGENTS.md §1.5 for the partnership context.

Recent pushes

🗂️ Push #58 (2026-05-20) — Refinement-trail archive: 10-slide is the only talk (ec9a3c6, CI #55 "Validate Repository" green 50s). With the 10-slide compressed final talk adopted as the conference deck, the 22-slide narrative and 12-slide intermediate compression (and their builders, the ChatGPT 22→12 compression-plan JSON, and the 22-slide speaking script) were moved into CODA-Association/CODAwork2026/archive/talk_decks_pre_10slide_2026-05-20/ with a folder-level README. README chain refreshed across CODA-Association/, CODAwork2026/, and data_outputs/. Stale slide-number references in How to run the presentation and Standards conformance closed. Single deck at the active surface.

🎤 Push #57 (2026-05-20) — Talk deck compression: 22 → 10 slides (09696d5, CI #54 "10-slide deck" green 56s). The 10-slide compressed final-talk deck (CodaWork2026_FinalTalk_10Slide_2026-05-20.pptx) becomes the conference talk. ~8 min spoken; slides 6 / 7 / 8 (Germany / Japan / UK case studies) deliberately weighted at 75 sec each. Beat-by-beat script: CODA-Association/CODAwork2026/SPEAKING_SCRIPT_10slide.md.

🌐 Push #56 (2026-05-20) — UN-6 PDF ambassador bundle (4e0e1a9, CI #53 "UN-6 Ambassador" green 52s). Five new print-ready PDFs at CODA-Association/Higgins_Decomposition_Handout_CoDaCommunity.{fr,es,ru,zh,ar}.pdf — same v10 layout, locale-specific line-height tuning, Arabic RTL with code-span LTR overrides, Chinese with CJK font embed. EN PDF unchanged.

Earlier in the conference-prep arc

🧭 Push #51 — Routing + Terms + Activation Coefficient (6d2e492, CI #48 "Routing + Terms" green 52s, 2026-05-16). Six-category bundle: (a) AI-refresh routing surfaces (README banner, llms.txt, HS_FAST_REFRESH.json) now point at CODA-Association/CODAwork2026/ as the conference-authority folder; (b) HUF-STD-001 v1.0 → v1.1 adds the person-noun convention (human → researcher / user / reader / participant) with an exception list for authorship rules, AI-safety vocabulary, anthropology, and regulatory disclosure; (c) HUF-STD-002 post-conference target reorder — Power Share / Activation Coefficient promoted to Order 1 (was the CNQ vector PDF exporter); (d) NOTATION_AND_TERMINOLOGY.md v2.0 + GLOSSARY.md v2.0 full refresh — 8/9 new sections each, Helmsman family promoted PROPOSED → CANONICAL per schema 3.1.0; (e) INV-060 title sharpened, Activation Coefficient formal name recorded; (f) CodaWork 2026 talk deck five-slide polish per commitment audit (8-simplex notation + EMBER CC BY 4.0 + four-category monitoring frame + "Mathematics is not new; the monitoring application may be" working-posture line). Lockdown-compliant; engine code, schemas, and INV catalog dispositions all untouched.


Previous: Push #50 (2026-05-14) — Conference-prep monster push

47cecc9, CI #47 "Foundations" green 48s. Twelve work products consolidated into a single coordinated commit under PRE_CONFERENCE_LOCKDOWN discipline:

  1. Hs/huf-gov/ — circuit-breaker structural addition with BREAKER_INVENTORY.md, 2 candidate DCPs (DCP-002 CHK-CNQ regex upgrade, DCP-003 CHK-DISPOSITION-001), and a breaker-test runner.
  2. Hs/CODA-Association/CODAwork2026/ — conference-authority folder with 8 versioned speaker docs, 13-slide talk deck, and a complete data_outputs/ Premier Data Output package.
  3. HUF-STD-001 Publication Standards — ICMJE/COPE/Nature/Science/WAME/EU-AI-Act/arXiv/ACM/IEEE-compliant AI Use Declaration template; human-only authorship.
  4. HUF-STD-002 Tensor Train I/O Standard — names the data → CNT → CNQ → vector output chain. PDF/PNG/SVG are standard; PPTX is conference-only.
  5. HUF-STD-003 Hs Linear Algebra Foundations — the seven components (Symmetric Matrix · Property of Transpose · Matrix Decomposition · Eigenvectors/Eigenvalues · Spectral Theorem · Spectral Decomposition · Visualization) named, with Stage-0 (Foundations Plate) as the visualization tier. Companion: FOUNDATIONS.md + FOUNDATIONS_TRACEABILITY.md.
  6. ILR-Helmert Triplet Plate generator (HCI/codawork2026/stage1_plates/ilr_triplet_plate.py) — orthonormal companion to the Section Plate.
  7. Stage-0 Foundations Plate generator (HCI/codawork2026/stage0_foundations/foundations_plate.py) — visualizes the seven foundations directly with machine-precision verification of the Spectral Theorem on actual data (Germany rank-k = 60.5% / 90.4% / 99.9%).
  8. Premier Data Output v2.0CodaWork2026_PremierDataOutput_2026-05-13.pdf (325 pages) + CodaWork2026_PremierDataOutput_2026-05-13.pptx (66 slides with corrected CNQ + new Triplet slide per country).
  9. Dual-View Stage 1 OutputCodaWork2026_DualViewStage1Output_2026-05-13.pdf (503 pages, Section + Triplet paired per country).
  10. Foundations Plates master PDFCodaWork2026_FoundationsPlates_2026-05-14.pdf (19 pages, cover + 9 countries × 2-page Stage-0 plate).
  11. CNQ dashboard fix — JSON-key-path corrections; all 9 countries now show real Hs(t), ω(t), K_eff+TV, helmsman σ(t), spike-detector, and diagnostics box.
  12. papers/ additionsEITT_CANONICAL_EXPLANATION_2026-05-12.md, BREAD_THE_HS_WAY_2026-05-12.md, HUF_GOV_BREAKER_TEST_2026-05-12.md, POST_CODA_PARTNERSHIP_TARGETS.md.

All under lockdown discipline: engine code, schemas, INV catalog dispositions (still 63 entries / 33 CANONICAL / 8 STAGED), six NO-CREATE files, and papers/codawork2026/talk/ untouched.

→ Push #50 summary: ai-refresh/PUSH50_READY_FOR_COMMIT.md


What's New — May 2026 (CoDaWork 2026 Conference-Prep Arc)

🎯 CoDaWork 2026 talk material complete (pushes #38–#43, 2026-05-10 / 2026-05-11). Six pushes in 48 hours delivered the full conference-prep bundle: published abstract honored, MC-4 claim sharpened to three-conjunct form, five named investigations (INV-049 through INV-055) catalogued CANONICAL, the talk delivery infrastructure shipped phone-readable at papers/codawork2026/talk/, and six post-conference research entries (INV-056 through INV-061) filed STAGED for promotion after 2026-06-06.

The talk material is structured in five layers — strategic compass (SPEAKER_BRIEF.md), spoken oratory (README.md), study guide (STUDY_PAGE.md), backstage scanner (CHEAT_SHEET.md), and AV-failure backup (BACKUP_PRESENTATION.md). Plus 10 slide files and 5 Q&A bench cards.

Catalog state at push #43: 61 total / 33 CANONICAL / 6 STAGED (new disposition for "canonical-content, deferred-ripple") / 12 DEFERRED / 8 OPEN / 1 FALSIFIED / 1 CLOSED. Sources: USER 25, GROK 18, CHATGPT 10, CLAUDE 8.

External-review validation: the talk's humble-invitation methods-challenge framing has been independently validated by two external models (ChatGPT session 2, Grok round 5) reading the MC-4 packet cold via the narrowed re-prompt template in ai-refresh/cross_check_archive/chatgpt_deep_research_2026-05-10_INDEX.md. Cross-model convergence on the same posture across three internal Claude reviews + two independent external models. See INV-059 (CANONICAL).

→ Conference-prep arc summary: ai-refresh/REPO_STATE_2026-05-11_post-push43.md → Push-by-push traceability: ai-refresh/PUSHES_INDEX.md


What's New — May 2026 (earlier arc)

Two protocols shipped that make this repo dramatically more usable for both researchers and AI assistants:

🆕 CCTT v1.0 — CNT Compositional Tensor Train. A 7-phase protocol that takes any compositional dataset (CSV/XLSX) and produces a CNT-grade analysis with full hash-chained provenance — even if you have never heard of Aitchison geometry. Works in two interchangeable modes:

  • User-mode — researcher walks the runbook by hand
  • User + AI-mode — AI assistant (Claude, ChatGPT, Gemini, in-house) executes the same 7 phases; user confirms at every gate

The protocol is identical in both modes. Pilot acceptance test: an AI given only the spec and a raw CSV reproduced the canonical content_sha256 byte-for-byte. → ai-refresh/CCTT_QUICKSTART.md

🆕 Volume IV — The Quaternion View (May 7, 2026, push #22). Names the algebra CNT has been computing in. Three IEEE-floor confirmations on drive failures, Planck CMB photons, and Standard Model neutrino oscillation establish that compositional dynamics on the simplex carries three structural invariances simultaneously — simplex rotation, mass-flow handedness, time-reversal symmetry — which is exactly the definition of a quaternion. Central claim: CNT measures invariance. CNQ names the algebra it lives in. Engine unchanged; 25-experiment determinism gate unchanged; what changes is what we can say about what the engine is doing. → HCI-CNT/handbook/VOLUME_4_QUATERNION_VIEW.md

🆕 HCI-CNQ — Compositional Navigation Quaternion (May 7-8, 2026, pushes #23–#27). The CNQ tier is live, canonical, and the engine is shipped. Promoted from experimental status after the third IEEE-floor confirmation (push #23); the production cnq.py engine landed in push #26; the R port cnq.R, language-agnostic pseudocode, and 43-test suite landed in push #27. The Hs system now ships a three-tier compositional analytics stack (CoDa → CNT → CNQ) plus the HCI instrument family. Both engines (cnt.py + cnq.py + cnt.R + cnq.R) are deterministic and hash-chained, producing identical content_sha256 across consecutive runs. Three reproducible IEEE-floor demonstrations on Backblaze, Planck CMB, and SM neutrino. Cross-platform reproduction challenge open. Open code, hash-chained outputs, doctrine published, build-to-spec help available, free.HCI-CNQ/README.md

🆕 HCI-AUDIO + HCI-ULTRASOUND — applied sibling tiers (May 8, 2026, push #24). Two new canonical tiers landed alongside the third AI cross-check pass (Grok). HCI-AUDIO/ is the canonical home for psychoacoustic 4-way active loudspeaker alignment with ERB-band carriers, quaternion phase mapping, and listening-position diffraction — the modern descendant of the original DADC compositional work. HCI-ULTRASOUND/ is the canonical home for non-contact medical and industrial ultrasound, with a geometry lock probe as the headline use case. Both are doctrine-only scaffolds; first pilots are the next milestones. → HCI-AUDIO/README.md, HCI-ULTRASOUND/README.md

🆕 DADC origin lineage documented (May 8, 2026, push #24). The Grok cross-check pass surfaced and verified the historical origin of the entire framework: DADC (Dimension-Apportioned Diffraction Correction) at the Binaural Test Lab in Markham, with a fixed 6.02 dB diffraction budget apportioned across cabinet dimensions — the first natural simplex constraint in the Higgins lineage. The lineage runs DADC → H₁ → HUF → Hˢ → CNT → CNQ. Original work: Rogue-Wave-Audio repository. Canonical lineage doc: HCI-CNT/handbook/ORIGIN_DADC_LINEAGE.md.

🆕 OPERATIONS_PROTOCOL v1.0 — Gawande meta-checklist for the whole repo. A single map of 12 transition points (starting an analysis, pushing, cowork session start/end, push failure, corpus drift, …) each with a binary pass/fail local checklist pointing at the canonical document holding the deeper detail. → OPERATIONS_PROTOCOL.md

🆕 EXPERIMENTS_JOURNAL.md — full sequential lineage of every experiment (push #34). Single citation-grade markdown that documents every experiment ever run under HUF / CNT v1 / CNT v2.0.4 / CNT v3.0.0 + CNQ v2.0.0, with the engine-version transitions, what each version added, what each version revealed that the predecessors did not, and direct links to every artefact. → EXPERIMENTS_JOURNAL.md

🆕 Full-corpus validation reference suite (push #34). 101 datasets across 11 domains; 100 ran end-to-end through CNT v3 + CNQ v2 with citation-grade Stage 1 (pure CoDa) + Advanced (full Hˢ + CNQ v2) reports per dataset. The definitive worked-examples set as of push #34. → experiments/2026-05-10_full-corpus-validation/README.md

🆕 Project doctrines — policy index (push #32–#33). Three model-agnostic doctrines bind every engine, wrapper, and published artifact:

  • docs/SUSPICION_OF_EVERY_ASSUMPTION.mdSEA-1.0: every public function and claim enumerates its failure modes with mitigation evidence; the engine is guilty until proven innocent (push #32).
  • docs/SELF_TEST_PROTOCOL.mdSTP-1.0: every engine carries a frozen reference corpus and a runner that produces dated, hash-chained receipts of pass/fail status (push #32).
  • docs/COHERENT_RANGE_DOCTRINE.mdCRD-1.0: every multi-carrier comparison is computed on the intersection of all members' time ranges; the shortest-coverage member sets the binding window; every output declares its coherent-range manifest in its header (push #33, INV-047).

Together with the engine-independence policy (push #32), these are the four binding doctrines of the framework.

Both protocols are model-agnostic, registered in ai-refresh/HS_ADMIN.json for future cold-start discovery, and proven end-to-end on the live repo (see OPERATIONS_PROTOCOL_PILOT_REPORT.md and ai-refresh/CCTT_PILOT_REPORT.md).


Start Here

If you want to see what the framework has actually been run on (NEW): EXPERIMENTS_JOURNAL.md — the full sequential lineage from HUF 12-step → CNT v3 + CNQ v2, every experiment dated and linked, every engine-version transition explained, every cross-version diff named. The single document that answers "what have we actually run, on what version, and what did each version change?"

If you have a compositional dataset and want a CNT-grade analysis right now (NEW): CCTT_QUICKSTART.md → walk the CCTT_RUNBOOK.md yourself, or paste the AI-mode prompt into Claude/ChatGPT/Gemini.

If you are working on or with this repo (NEW): OPERATIONS_PROTOCOL.md — the front-door map of every operational transition (analysis, push, cowork start/end, AI cold-start, recovery paths). One row per transition, each pointing at its canonical checklist.

If you are a person: Learning PathArchitecture OverviewApplications GuideHigh Index Platform

If you are a machine: ai-refresh/HS_MACHINE_MANIFEST.json — identity, navigation, protocol, governance, and authority resolution in a single file. Follow the onboarding sequence defined there. Then read OPERATIONS_PROTOCOL.md and ai-refresh/CCTT_RUNBOOK.md.

If you want to run Hs right now: Standards Edition Notebook — self-contained Jupyter notebook, auto-installs dependencies, auto-fetches pipeline from GitHub, includes 3 built-in reference standards, runs all advanced analyses. The conference handout tool.

If you are reviewing for CoDaWork 2026: Abstract (PDF)Executive SummaryCCTT_QUICKSTART.mdStandards Edition NotebookCollaboration Path


HCI-CNT — Compositional Navigation Tensor (active development line)

The HCI-CNT/ subsystem extends Hˢ with the Compositional Navigation Tensor (CNT) — a deterministic, hash-traceable instrument for compositional time series and cross-sections. Engine 2.0.4 / Schema 2.1.0 / 25 reference experiments, all passing the determinism gate.

CNT shares Hˢ's foundations (Aitchison geometry, simplex closure, same input → same output → always) and adds trajectory-native operators (bearings, angular velocity, helmsman, period-2 attractor, IR class), a four-stage paged report family, end-to-end hash provenance, and cross-dataset inference reports.

Three handbook volumes in HCI-CNT/handbook/ cover the system in full:

Volume Audience
VOLUME_1_THEORY_AND_MATHEMATICS.md math, schema, doctrine, balance vs classical CoDa
VOLUME_2_PRACTITIONER_AND_OPERATIONS.md engine, atlas, mission command, demo, ROI, integrations
VOLUME_3_VERIFICATION_REFERENCE_AND_RELEASE.md determinism, hash chain, talk plan, public-trial readiness

Quickstart: see HCI-CNT/README.md.

Three CoDa-community preprint papers live at HCI-CNT/coda_community/, and the CodaWork 2026 demo package at HCI-CNT/conference_demo/ is self-contained.

The previous standalone HUF-CNT-System package outside the Hˢ repo is preserved as archived history. Active CNT development from this point forward happens inside HCI-CNT/.


HCI-CNQ — Compositional Navigation Quaternion (live tier, push #23)

The HCI-CNQ/ subsystem is the quaternion-native sibling tier above CNT in the three-tier Hs stack (CoDa → CNT → CNQ). Promoted to canonical on 2026-05-07 after three independent IEEE-floor confirmations of the quaternion identification on real datasets. Doctrine, demonstrations, comparisons with CoDa and CNT, and the engineering proposal for a compiled cnq.py engine all live in this folder, in public.

The CNQ tier is what comes above CNT for problems CNT was not designed for: D ≥ 8, large T, multi-trajectory bundles, cross-dataset structure as the primary observable. Climate modeling, multi-decade economic flows, large industrial composition, microbiome cohorts.

Folder Contents
HCI-CNQ/doctrine/ Central claim, deeper connections, concepts-for-test, corpus comparison plan, post-CoDa benefits
HCI-CNQ/tier_system/ CoDa → CNT → CNQ tier explanation, ROI/use cases, engine proposal, three-way comparison
HCI-CNQ/experiments/ Three IEEE-floor demonstrations: backblaze drive failures, Planck CMB photons, SM neutrino oscillation

Three reproducible demonstrations are in the experiments folder. Each is self-contained — script, input data, CNT JSON output, results, report. Anyone can re-run.

The compiled cnq.py engine shipped in push #26 (2026-05-08) and now runs at CNQ v2.0.0 / schema cnq/2.0.0 alongside CNT v3.1.0 / schema 3.1.0. The experiments remain the working proofs at the IEEE-floor residual (4.441 × 10⁻¹⁶ on Backblaze and Planck CMB; 7.4 × 10⁻¹⁷ on SM neutrino oscillation). Quickstart: see HCI-CNQ/README.md. Current engine state is authoritative in HS_FAST_REFRESH.json.

How we work — demonstration first. Every tool in the Hs family — CoDa methods (community-standard), CNT (HCI-CNT/), CNQ (HCI-CNQ/), HCI-AUDIO (HCI-AUDIO/), HCI-ULTRASOUND (HCI-ULTRASOUND/), and the HCI instrument family (HCI/) — is built and tested in public on the same terms: open code, hash-chained outputs, doctrine published. We show what each tool is, what it does (by demonstration on real datasets), when to use it, how to use it, and we offer to help you build it to specification on your own data, free. Open an issue on the repository, or follow the contact in PUBLICATION_READY.md.


HCI-AUDIO — applied sibling, doctrine-only (push #24)

The HCI-AUDIO/ subsystem is the canonical home for applied audio work: 4-way active loudspeaker alignment with ERB psychoacoustic band carriers, quaternion phase mapping, and listening-position diffraction.

This is the direct modern descendant of the original DADC (Dimension-Apportioned Diffraction Correction) work in Rogue-Wave-Audio — the BTL loudspeaker-laboratory work that created the first natural simplex constraint in the Higgins lineage. Where DADC apportioned a fixed 6.02 dB diffraction budget across three cabinet dimensions, HCI-AUDIO apportions perceptual energy across 40 ERB bands × 4 drivers at the listening position. Same closure principle, applied at the right scale.

Folder Contents
HCI-AUDIO/doctrine/ ERB band mapping, quaternion phase mapping, helmsman at listening position, alignment targets
HCI-AUDIO/spec/ Psychoacoustic 4-way adapter spec, pipeline spec

Status: doctrine-only. First pilot (real measurement against the project's reference 4-way system) is the next milestone. Quickstart: HCI-AUDIO/README.md.


HCI-ULTRASOUND — applied sibling, doctrine-only (push #24)

The HCI-ULTRASOUND/ subsystem is the canonical home for non-contact medical and industrial ultrasound, with a geometry lock probe as the headline use case. The lock probe uses CNT/CNQ-driven feedback (Joint Helmsman + Helmsman Stability + M² = I) to actively maintain measurement on a specific geometric feature of the target — an edge, a tissue interface, a defect, a specular reflector — under relative motion or noise.

This is the active-sensing descendant of DADC: the same closure principle (apportioning a fixed return-signal total across carriers), plus a control loop that steers the probe to keep the helmsman locked on the desired feature.

Folder Contents
HCI-ULTRASOUND/doctrine/ Geometry lock probe, object detection, autofocus and stabilisation, medical vs industrial
HCI-ULTRASOUND/spec/ Ultrasound adapter spec

Status: doctrine-only. Recommended first pilot is industrial composite inspection on a public dataset (lower regulatory overhead). Quickstart: HCI-ULTRASOUND/README.md.


Lineage

The simplex / compositional thinking that underpins Hˢ → CNT → CNQ originated in earlier loudspeaker work at the Binaural Test Lab (BTL) in Markham, Ontario — a single-identity lab with canonical machine-readable identity card RWA-001.json. The BTL work is documented in the Rogue-Wave-Audio repository (live site) and mirrored locally at ../RWA/ for reference. Specifically, DADC (Dimension-Apportioned Diffraction Correction) discovered that the cabinet-edge diffraction gain was a fixed 6.02 dB budget that had to be apportioned across the three cabinet dimensions — the first natural simplex constraint in the Higgins lineage. The lineage runs DADC → H₁ (Higgins Operator, a nonlinear unity-normalization map on Hilbert space) → HUF (Higgins Unity Framework, MC-4 + EITT) → Hˢ (Higgins Decomposition, this repo) → CNT (engine v3.0.0) → CNQ (engine v2.0.0) → HCI-AUDIO + HCI-ULTRASOUND (applied tiers). Full canonical narrative: HCI-CNT/handbook/ORIGIN_DADC_LINEAGE.md. RWA-side reciprocal: ../RWA/HUF_RELATIONSHIP.json.

This tool also emerged from the Higgins Unity Framework, which remains the governance, application, and historical development sibling. The mathematical foundations build on Aitchison (1982/1986) simplex geometry, Shannon (1948) entropy, and Varley (2025) information theory for complex systems.


Prime Documents

These are the governing documents of the Hˢ system — the ones that define what it is, what it does, and what it claims.

Document Purpose
EXECUTIVE_SUMMARY.md Running log of all development, decisions, results, and principles
Decomposition Function (v3.0) Formal derivation: axiom → decimation → seven operators → Hˢ
Logic Map and State Machine Complete symbolic logic of the pipeline
Symbolic Logic Definition Pure mathematical definition — no prose
Reference v3.0 (docx) Formal reference document with full operator specifications
Character Analysis (docx) Atomic-level disassembly — the pipeline as DUT
Instrument Metrology Quantified instrument qualification (6 metrics)
Naming Convention File naming rules, branding, and terminology migration
CITATION.cff How to cite this work

The Pipeline (13 Files)

All code lives in tools/pipeline/. No external dependencies beyond numpy.

Core Engine

File Role
higgins_decomposition_12step.py The 12-step pipeline — simplex closure through helix projection
higgins_transcendental_pretest.py Transcendental constant proximity against 35-constant library
hs_amalgamation.py Subcompositional recursion engine — amalgamation stability testing

Diagnostics

File Role
hs_codes.py 78 diagnostic codes + 10 structural modes
hs_fingerprint.py Seven-dimensional compositional fingerprint generator + matcher
hs_sensitivity.py Component Power Mapper — leverage, phase, power scores per carrier
hs_metrology.py Instrument meta-evaluation — Gauge R&R, self-consistency

Ingestion

File Role
hs_ingest.py Universal CSV/JSON loader — any composition, automatic closure
hs_hepdata.py HEPData fetch — 8 curated HEP datasets with validated pipeline runs

Infrastructure

File Role
hs_reporter.py Multilingual diagnostic reporter (5 reporter languages — see also UN-6 wrapper system in HCI-CNQ/wrappers/)
hs_testgen.py Secondary test tools — adversarial, boundary, and regression tests
hs_audit.py Audit trail + 16 configurable breakpoints
hs_controller.py Industrial state machine controller with Hˢ-GOV supervisor

Interactive Tools (9 HTML Demos)

Download any HTML file and open in a browser. No installation, no server, no dependencies.

Tool What It Does
CoDaWork Demo Dual-dataset live demo — SEMF + Radionuclides, full pipeline strip, structural modes
Cosmic Composition Slider Planck 2018 cosmic energy budget — slide from z=0 to z=3400, watch dark energy vanish
Cosmic Cone Loop 5-minute inflation cone animation — cosmic composition evolution from Big Bang
Cosmic Duality Dance Black hole / white hole compositional duality across amalgamation levels
Cosmic Future Projection ΛCDM Friedmann model — dark energy dominance trajectory from 1 Myr to heat death
Simplex Scope Real-time Fourier conjugate pair decomposition — all 12 pipeline steps visualised
Spring-Mass Simulator Damped oscillator decomposed into KE/PE/Damping with chaos detection
Conjugate Preservation Theorem Mathematical proof — 3 theorems + 1 corollary, interactive walkthrough
Hˢ Spectrum Analyzer Universal JSON reader — 5 readings from any pipeline output file

Quick Start

Have a CSV? One command:

python tools/pipeline/hs_ingest.py mydata.csv --all-languages

Have HEPData? Published high-energy physics measurements:

python tools/pipeline/hs_hepdata.py --list                    # see 8 curated HEP datasets
python tools/pipeline/hs_hepdata.py --fetch higgs_br --run    # Higgs branching ratios → pipeline
python tools/pipeline/hs_hepdata.py --fetch-all --run         # all 8 → full pipeline runs

Python API:

from tools.pipeline.higgins_decomposition_12step import HigginsDecomposition

hd = HigginsDecomposition("MY-01", "My System", "MY_DOMAIN",
    carriers=["A", "B", "C"])
hd.load_data(my_matrix)  # numpy array, shape (N, D)
result = hd.run_full_extended()

from tools.pipeline.hs_codes import generate_codes
from tools.pipeline.hs_reporter import report
codes = generate_codes(result)
print(report(codes, lang="pt"))  # en, zh, hi, pt, it

Amalgamation stability test:

from tools.pipeline.hs_amalgamation import AmalgamationEngine
engine = AmalgamationEngine(hd)
results = engine.run_all_schemes()  # tests all valid carrier merges

The 25 Experiments

ID Domain System Highlight
Hs-01 Precious metals Gold/Silver ratio Transfer entropy: Au→Ag directed flow
Hs-02 Energy US primary energy mix Renewable carrier drift detection
Hs-03 Nuclear physics SEMF binding energy Flagship: δ = 5.87 × 10⁻⁶ at 1/(π^e), Z=38 strontium
Hs-04 Acoustics Bessel function decomposition Spectral mode analysis on simplex
Hs-05 Geochemistry Major oxide compositions CaO+MgO dominant (61%) — depletion carries variance
Hs-06 Nuclear fusion Plasma confinement Lawson criterion approached compositionally
Hs-07 QCD Quark/gluon decomposition Perturbative ↔ non-perturbative boundary
Hs-08 Particle physics CKM/PMNS mixing matrices Flavour mixing as composition
Hs-09 Stellar physics Main-sequence composition CNO cycle carrier detection
Hs-10 Gravitational waves GW150914 merger Chirp mass ratio decomposition
Hs-11 Nuclear mass AME2020 atomic masses Binding energy systematics across chart of nuclides
Hs-12 Classical mechanics Spring-mass oscillator KE/PE exchange — reversal under heavy damping
Hs-13 Metallurgy Steel alloy compositions Phase-boundary detection via variance trajectory
Hs-14 Mathematics Fourier conjugate pairs 12/12 preservation — 3 theorems + 1 corollary
Hs-15 Materials science hBN dielectric response Crystal field decomposition
Hs-16 Cosmology Planck 2018 cosmic budget Dark energy dominance, CDM/Baryon lock (CV=0)
Hs-17 Data engineering Backblaze HDD reliability Fleet composition drift, 4 sub-experiments
Hs-18 Urban planning Markham municipal budget Capital vs operating drift
Hs-19 Infrastructure Traffic signal timing Phase allocation as composition
Hs-20 AI/NLP Conversation drift Text-to-composition mapping (exploratory)
Hs-21 Calibration Reference standard library 15 standards: mathematical, diffraction, transcendental
Hs-22 Cross-domain Natural pairs baseline 12 systems, 7 domain pairs, cross-pair constant sharing
Hs-23 Nuclear decay Radionuclide chains (U-235, U-238, Th-232) Decay chain as compositional trajectory
Hs-24 Particle physics HEPData validation campaign 9 runs across 8 HEP systems, independent data source
Hs-25 Cosmology Planck 2018 cosmic energy budget CoDaWork centrepiece — amalgamation reveals conservation laws

Key Results

Finding Value Source
Tightest transcendental match δ = 5.87 × 10⁻⁶ (Nuclear SEMF → 1/(π^e) at Z=38) Hs-03
Classification rate 15/15 NATURAL across all physical systems All experiments
Fourier conjugate preservation 12/12 pairs bit-identical (3 theorems + 1 corollary) Hs-14
Amalgamation stability 58/58 schemes preserve classification (100%) Hs-25, cross-domain
EITT entropy invariance < 5% variation under geometric-mean decimation All natural systems
Adversarial robustness 21 attacks, 0 plausible-but-wrong outputs Character Analysis
Transfer entropy Detects directed causal flow between carriers All experiments
Ratio locks CDM/Baryon and Photon/Neutrino at CV=0 survive all amalgamation Hs-25

CoDaWork 2026 — Coimbra, Portugal (June 1–5)

The 11th International Workshop on Compositional Data Analysis. 11 days away as of this README revision. The conference-authority folder is CODA-Association/CODAwork2026/ — everything an attendee or reviewer needs is there, in one place.

Present at the talk

What File
Audience follow-along page CODA-Association/CONFERENCE_ATTENDEES.md — slide-by-slide, every link in talk order
Manuscript (PDF, 26 pp) CODA-Association/CODAwork2026/Compositional_Monitoring_2026.pdf
Manuscript (.docx) CODA-Association/CODAwork2026/Compositional_Monitoring_2026.docx
Talk deck — 13 slides (.pptx) CODA-Association/CODAwork2026/data_outputs/CodaWork2026_FinalTalk_13Slide_2026-05-24.pptx
Talk deck (.pdf) CODA-Association/CODAwork2026/data_outputs/CodaWork2026_FinalTalk_13Slide_2026-05-24.pdf
Speaking script (beat-by-beat) CODA-Association/CODAwork2026/SPEAKING_SCRIPT_13slide.md
Cinema scroll (66 slides, 325 pp) CodaWork2026_PremierDataOutput_2026-05-13.pptx · .pdf
Interactive HTML projector codawork2026_projector.html — three modes (RADAR / BARY / ALIGN) + SHOCK overlay; runs offline
Community handout (UN-6) EN · FR · ES · RU · ZH · AR — at CODA-Association/
Foundation paper (companion master standard) papers/flagship/GROUND_STATE_AND_TRACTION.md — the unified formula behind the framework

Conference timing

Piece Time
13-slide talk ~8 min 50 sec spoken (slides 6–11 are the three country pairs — share-and-work + navigation, 85 sec per country)
Cinema scroll ~1 min as a movie after slide 13
HTML projector ~1 min Q&A demo
Apparatus time ~11 min
Q&A ~4 min remaining in a 15-min slot

Source materials (working folder)

The working build folder is at papers/codawork2026/ — abstract, executive summary, manuscript build pipeline, talk-prep documents. The conference distribution lives in CODA-Association/CODAwork2026/; the working build remains in papers/codawork2026/. Both surfaces are documented and current.

Three open questions posed to the CoDa community: (1) Can the EITT entropy invariance be proved from Aitchison geometry? (2) Does classification survive ILR

About

Higgins Decomposition (Hs) — deterministic compositional inference instrument. 18 domains, 36 systems. CNT engine + CNQ quaternion verification (IEEE-floor on Backblaze, Planck CMB, SM Neutrino). Apache-2.0 (code) + CC BY 4.0 (docs).

Topics

Resources

License

Unknown, Unknown licenses found

Licenses found

Unknown
LICENSE
Unknown
LICENSE-DOCS

Code of conduct

Contributing

Security policy

Stars

Watchers

Forks

Packages

 
 
 

Contributors