From 153246f78a8afc0352a4046b1af0842a4a7917bd Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Mon, 8 Jun 2026 21:42:48 +0000 Subject: [PATCH 01/11] =?UTF-8?q?docs:=20add=20DISCOVERY-MAP=20=E2=80=94?= =?UTF-8?q?=20the=20shape-preserving=20index=20of=20every=20discovery?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Per operator request: "map all discoveries and ideas out so that they don't dilute shape once materialized in architecture." The anti-dilution thesis: shape dilutes from FRAGMENTATION, not from too few documents — each idea pinned in isolation loses its links, and a future reader sees N disconnected ADRs instead of one doctrine. This map is the one place the topology lives. Terse by discipline (pointers + structure + status, never re-explanation); append-only + status-tracked, mirroring the substrate's own versioned-audit shape (ADR-008/013). Structure: §1 doctrine spine — ADR-022/023/024/025 (+ 026 pending) + the one- sentence gestalt the individual ADRs lose §2 the discovery ledger — ~40 D-* entries across 9 areas, each: [ID] shape — grade (G/H/S) — status (IDEA/EPIPHANY/ADR/CODED) — home — depends. Covers addressing/cascade, selection/bounds, codec/codebooks, no-collapse precondition, amortization gate, storage, synergy catalog, the coded IR+adapters, domain instances. §3 materialization pipeline — what's READY to materialize now (G + EPIPHANY, no runtime gate -> draft ADR-026), what's BLOCKED on runtime confirmation, what's CODED, what's queued, what's do-not-build (OLED [S]). §4 the OGAR/runtime boundary — owned-vs-[per rt]; the five open ADR-026 confirmations; the two operator-closed items. §5 the shape graph — the ASCII topology the individual ADRs lose. §6 maintenance discipline — append-only, terse, grade-honest, index- not-content; the map mirrors the substrate (recursive = the tell the shape is intact). §7 cross-references. Status column makes the materialization state of every idea visible at a glance, so when ADR-026 (or future ADRs/crates) pin a subset, the rest stay tracked and linked instead of forgotten. The map is the durable shape; the ADRs/crates/synergy-doc are the content it indexes. Docs-only. PII abort-guard (word-boundary): CLEAN. cargo check: clean. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- docs/DISCOVERY-MAP.md | 268 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 268 insertions(+) create mode 100644 docs/DISCOVERY-MAP.md diff --git a/docs/DISCOVERY-MAP.md b/docs/DISCOVERY-MAP.md new file mode 100644 index 0000000..48c2545 --- /dev/null +++ b/docs/DISCOVERY-MAP.md @@ -0,0 +1,268 @@ +# Discovery Map — the shape‑preserving index + +> **Purpose.** A single append‑only index of every discovery, idea, and +> decision in the substrate arc — each with its **grade**, **materialization +> status**, **home**, and **dependencies** — so the *shape* (the connections +> between ideas) survives when ideas materialize into individual ADRs and +> crates. +> +> **The anti‑dilution thesis.** Shape dilutes not from too few documents but +> from **fragmentation**: each idea pinned in isolation loses its links, and +> a future reader sees N disconnected ADRs instead of one doctrine. This map +> is the one place the **topology** lives. It is **terse by discipline** — +> pointers + structure + status, never re‑explanation. The ADRs / crates / +> docs are the *content*; this is the *index*. It is **append‑only + +> status‑tracked**, mirroring the substrate's own versioned‑audit shape +> (ADR‑008/013) — fitting, since a map of an append‑only audited substrate +> should itself be append‑only audited. +> +> **How to read an entry:** `[ID] one‑line shape — grade — status — home — +> depends`. +> - **Grade:** `[G]` Grounded (real artifact/spec) · `[H]` Hypothesis +> (sound, one side unconfirmed) · `[S]` Speculative (catalog only). +> - **Status (materialization pipeline):** `IDEA` → `EPIPHANY` (graded, +> captured in a synergy/epiphany doc) → `ADR` (pinned contract) → `CODED` +> (crate + tests + CI). +> - **`[per runtime session]`** = runtime‑owned; OGAR cannot verify. +> +> **Status: LIVING INDEX** (2026‑06‑08). Update entry *status* on +> materialization; never delete (append‑only). One entry per discovery; +> cross‑link, do not duplicate. + +--- + +## 1. The doctrine spine — the load‑bearing floors + +The ADR canon is the spine; every discovery in §2 hangs off the floor it +specializes. + +| ADR | Floor | Status | Home | +|---|---|---|---| +| **022** | **boundary** — no serialization in the hot path | `ADR` | `THE-FIREWALL.md` | +| **023** | **IR** — `Class` is wire truth; dialects absorbed by adapters | `ADR` | `ARCH-DECISIONS` + `ogar-vocab/` | +| **024** | **codec** — palette256 + HHTL; 1‑byte index, sub‑µs decode, ρ ≥ 0.99 | `ADR` | `ARCH-DECISIONS` | +| **025** | **selection** — probe‑free closed‑form level (`r* = ⌈log₄(C/τ)⌉`) | `ADR` | `ARCH-DECISIONS` | +| **026** | **(pending)** cascade + amortization gate + meta→content gradient + no‑collapse precondition + storage | `EPIPHANY → ADR` | `CASCADE-SYNERGIES-EPIPHANY.md` | + +**The gestalt the spine forms (the one sentence the individual ADRs lose):** + +> *The substrate is an **immaterialized Morton cascade with templated +> payloads** — address is geometry (helix + HHTL), error is closed‑form +> (jc), the codec is exact in the θ‑window, materialization is lazy and +> columnar; the **only forbidden cost is non‑amortized per‑query.*** + +ADR‑022 (the boundary) + ADR‑025 (the probe) are both special cases of that +last clause. ADR‑023 (address vs payload) + ADR‑024 (the codec) describe the +two halves of a cell. ADR‑026 names the cascade that ties them. + +--- + +## 2. The discovery ledger + +### 2.1 Addressing & cascade + +| ID | Shape | Grade | Status | Home | Deps | +|---|---|:--:|:--:|---|---| +| D‑MORTON | NiblePath nibble = one 4×4 Morton tile (2 x‑bits + 2 y‑bits interleaved) | G | EPIPHANY | SYN §0,§7.5 | 023,024 | +| D‑XOR2 | one address‑XOR, two reductions: **popcount = perms** (024) / **CLZ = containment** (025) | G | EPIPHANY | SYN §0 | 024,025 | +| D‑CASCADE | 64→256→1024→4096→16k→64k→256k = immaterialized Morton enumeration; every level = +1 nibble | G | EPIPHANY | SYN §7.5 | D‑MORTON | +| D‑IMMAT | the cascade is a **coordinate transform, not a stored grid** (`(lat,lon)→quadkey` cheap) | G | EPIPHANY | SYN §7.5 | D‑CASCADE | +| D‑NEIGH | neighbor‑XOR walk + parent‑prefix = structured‑sparse stencil (block‑banded, not sparse GEMM) | H | EPIPHANY | SYN §6 | D‑MORTON, `[per rt]` blasgraph | + +### 2.2 Selection & bounds + +| ID | Shape | Grade | Status | Home | Deps | +|---|---|:--:|:--:|---|---| +| D‑PROBEFREE | hot path takes zero data‑dependent branches; level is closed‑form | G | ADR | ADR‑025 | 022 | +| D‑RSTAR | `r* = ⌈2·log_b(C/τ)⌉`; `b=16` ⟹ `⌈log₄(C/τ)⌉`; inclusive `≤ τ` | G | ADR | ADR‑025 (#45,#46) | D‑CASCADE | +| D‑BOUNDS | tile `bounding_volume` is **derived** (address arithmetic), not stored | H | ADR | ADR‑025 | D‑MORTON, `[per rt]` helix | +| D‑CESIUM‑PROBE | native Cesium SSE refine/collapse = the trial‑and‑error probe 025 removes | G | EPIPHANY | SYN §1 | D‑PROBEFREE | + +### 2.3 Codec & codebooks + +| ID | Shape | Grade | Status | Home | Deps | +|---|---|:--:|:--:|---|---| +| D‑PAL256 | palette256 = one PQ subspace's 256 centroids (CAM = 6×256, PR #477) | G | ADR | ADR‑024 | — | +| D‑CAM | CAM‑PQ = 6 roles × 256 = 48‑bit fingerprint per word (`cam_codes.bin`) | G | EPIPHANY | SYN §3 | D‑PAL256 | +| D‑CARD | cascade per‑axis cardinality ↔ codebook size: 256↔palette, 4096↔`MAX_VOCAB` | G | EPIPHANY | SYN §3,§7 | D‑CASCADE, D‑CAM | +| D‑META64 | `CausalEdge64` = meta: `2⁶` role‑mask **+** 48‑bit CAM + 16‑bit headroom, one 64‑bit word | H | EPIPHANY | SYN §9.6 | D‑CAM, `[per rt]` CausalEdge64 | +| D‑BITGATE | 48→64 headroom: store the seed **iff** irreducible‑beyond‑the‑address (else compute) | H | EPIPHANY | SYN §9.6 | D‑META64, D‑AMORT | +| D‑RHO | ρ = 0.9973 vs cosine (arm‑discovery aerial codebook) — the empirical anchor | G | ADR | ADR‑024 | D‑PAL256 | + +### 2.4 No‑collapse precondition (θ + golden) + +| ID | Shape | Grade | Status | Home | Deps | +|---|---|:--:|:--:|---|---| +| D‑THETA | θ ∈ [1.45,1.6] (near‑orthogonal, cosθ∈[−0.03,0.12]) = the conditioning window | H | EPIPHANY | SYN §2; ADR‑025 ref | D‑PAL256, `[per rt]` jc | +| D‑RHOENV | ρ ∈ [0.93 (band edge) .. 0.9973 (θ≈π/2)] = exactness envelope over θ | H | EPIPHANY | SYN §2 | D‑THETA, D‑RHO | +| D‑GOLDEN | golden‑stride helix placement controls θ into the window **by construction** | H | EPIPHANY | SYN §2 | `[per rt]` helix | +| D‑MOIRE | golden irrationality = X‑Trans‑grade anti‑moiré interlace ("x256 that can't collapse") | H | EPIPHANY | SYN §2 | D‑GOLDEN | +| D‑NOCOLLAPSE | unify: near‑orthogonal **codebook** + aperiodic **lattice** = no degenerate beat in value‑ or position‑space | H | EPIPHANY → ADR‑026 §2 | SYN §9.1 | D‑THETA, D‑MOIRE | + +### 2.5 Amortization gate + +| ID | Shape | Grade | Status | Home | Deps | +|---|---|:--:|:--:|---|---| +| D‑AMORT | admission test = **amortize‑or‑don't‑spend** (not "free"); unifies 022 + 025 | G | EPIPHANY → ADR‑026 §3 | SYN §7 | 022,025 | +| D‑AMORT‑AXES | two reuse axes: **mipmap cascade** (spatial) / **Semantik volumetric centroid** (semantic) | G | EPIPHANY | SYN §7 | D‑AMORT | +| D‑BOTHCASC | store **both** freq + semantic cascades (SoA headroom affords it); "pick one axis" fork dissolved | G | EPIPHANY | SYN §7 | D‑AMORT‑AXES | +| D‑FRACTAL | the gate is fractal: **storage** (§7.5) / **cascade** (§7) / **bit** (§9.6) — one rule, three scales | G | EPIPHANY | SYN §7 | D‑AMORT, D‑BITGATE | +| D‑MAXAMORT | enrich the **coarse/meta** levels lavishly (amortize over everything below); be frugal only at the per‑query leaf | H | EPIPHANY | SYN §9 | D‑AMORT, D‑META64 | + +### 2.6 Storage + +| ID | Shape | Grade | Status | Home | Deps | +|---|---|:--:|:--:|---|---| +| D‑COLUMNAR | Morton‑keyed columnar rows + prefix pushdown = tile fetch (Delta `ZORDER`/Iceberg/BigQuery — deployed) | G | EPIPHANY | SYN §7.5 | D‑IMMAT | +| D‑LANCE | "parquet‑shaped" = the columnar family; **Lance** is the instance (random tile access > Parquet row‑group scan) | G | EPIPHANY | SYN §7.5 | D‑COLUMNAR | +| D‑PYRAMID | the unifying shape: columnar (parquet) + Morton rows (grid) + level cascade (pyramid) + closed‑form cell (shader) | H | EPIPHANY | SYN §7.5 | D‑COLUMNAR, D‑CASCADE | +| D‑DELTA | delta frames = version‑diff = changed Morton cells = codec P‑frame (I/P map; B does not, append‑only log) | H | EPIPHANY | SYN §7.5,§1 | D‑LANCE | + +### 2.7 Cross‑domain synergy catalog + +| ID | Shape | Grade | Status | Home | Deps | +|---|---|:--:|:--:|---|---| +| D‑CTU | Morton cascade = x265/x266 CTU quadtree; codec RDO split = the probe | G | EPIPHANY | SYN §1 | D‑CASCADE, D‑CESIUM‑PROBE | +| D‑ATTN | attention (bgz‑tensor WeightPalette 256) ranks tiles → drives `r*`; τ = min(certificate, attention) | H | EPIPHANY | SYN §4 | D‑RSTAR, D‑PAL256 | +| D‑CONVERGE | 6 lineages (codecs/sensors/displays/attention/PQ/Cesium) → quadtree + 256‑palette + irrational | G | EPIPHANY | SYN §0 | — | +| D‑OLED | OLED exciton ↔ substrate = weakest leg; only candidate diffusion‑length ↔ neighborhood; **do not build on** | S | EPIPHANY | SYN §3 | — | + +### 2.8 IR & adapters — the coded layer + +| ID | Shape | Grade | Status | Home | Deps | +|---|---|:--:|:--:|---|---| +| D‑VOCAB | `Class`/`Attribute`/`Association`/`EnumDecl`/`ActionDef`/`KausalSpec`/`Identity` IR | G | CODED | `ogar-vocab/` | 023 | +| D‑IDENT | `class_identity(prefix,name)` = NiblePath; canonical‑form invariant | G | CODED | `ogar-ontology/` (#31) | D‑VOCAB | +| D‑EMIT | `TripleEmitter` — 129 RDF predicates; SPO + TeKaMoLo | G | CODED | `ogar-emitter/` | D‑VOCAB | +| D‑SURREALQL | SurrealQL DDL adapter, parse+emit, round‑trip, identifier‑quoted | G | CODED | `ogar-adapter-surrealql/` (#32,#36) | D‑VOCAB | +| D‑TTL | Turtle (RDF/OWL) adapter, parse+emit, round‑trip | G | CODED | `ogar-adapter-ttl/` (#37) | D‑EMIT | +| D‑CHDDL | ClickHouse DDL adapter, parse+emit, dotted‑name round‑trip | G | CODED | `ogar-adapter-clickhouse-ddl/` (#38,#40) | D‑VOCAB | +| D‑KNOWABLE | `KnowableFromStore` + `register_class_knowable_from`; `surrealql-hint`; **`vart-backend`** | G | CODED | `ogar-knowable-from/` (#25,#33,#43) | D‑IDENT | +| D‑HINT | `schema_ddl_hint` loop closed — self‑describing registry via emit | G | CODED | (#33) | D‑SURREALQL, D‑KNOWABLE | +| D‑ELIXIR | Elixir/HIRO SchemaSource scaffold (`gen_statem`→Rubicon) | G | CODED (scaffold) | `ogar-from-elixir/` | D‑VOCAB | +| D‑OSM | `ogar-from-osm-pbf` — Node/Way/Relation; quadkey NiblePath from resolved geometry | H | IDEA | (queued) | D‑VOCAB, `[per rt]` D‑OSM‑3 | +| D‑PATTERN | `ogar-pattern` — recognition library + confidence (FMA‑D/FIBO/SKR/PROV‑O) | H | IDEA | (queued) | D‑TTL | +| D‑ACTION | `ogar-actionable` — lifecycle → `ActionDef`/`KausalSpec` | H | IDEA | (queued) | D‑PATTERN | +| D‑NSM | 4096‑dim Deep‑NSM encoder (Wierzbicka primes, `NUM_PRIMES=63`) calibration | H | IDEA | RDF‑OWL §4.10 | D‑CAM, `[per rt]` | + +### 2.9 Domain instances (universality witnesses) + +| ID | Shape | Grade | Status | Home | +|---|---|:--:|:--:|---| +| D‑DOM | 6 instances: chess / OpenProject / Elixir‑HIRO / Odoo‑ERP / HIPAA / OSM | G | ADR/doc | `DOMAIN-INSTANCES.md` (#27,#41,#42) | +| D‑PII | label‑free contract IS the PII guarantee (HHTL leaf‑rename at Adapter) | G | doc | `HEALTHCARE-TRANSCODING.md §4` | +| D‑LITMUS‑FMA | FMA bones‑rendering = compile‑time HHTL litmus (~75K static classes) | H | doc | `RDF-OWL-ALIGNMENT.md §6` | +| D‑LITMUS‑GEO | OSM = geographic litmus; "Femur is_a LongBone AND Marienplatz is_in Munich" sub‑µs | H | doc | `DOMAIN-INSTANCES.md §2.6` | + +--- + +## 3. The materialization pipeline — what's ready, what's blocked + +``` +IDEA ──→ EPIPHANY ──→ ADR ──→ CODED +(floated) (graded) (pinned) (crate+CI) +``` + +**Ready to materialize now** (`G` + `EPIPHANY`, no runtime gate) → **draft +ADR‑026**: +- D‑AMORT, D‑AMORT‑AXES, D‑BOTHCASC, D‑FRACTAL (the amortization gate) +- D‑MORTON, D‑XOR2, D‑CASCADE, D‑IMMAT (the addressing) +- D‑COLUMNAR, D‑LANCE (the storage layer — almost entirely `[G]`) +- D‑CTU, D‑CONVERGE (the grounded synergies) + +**Blocked on runtime confirmation** (`H` + `[per runtime session]`) → ADR‑026 +lands these with `[per rt]` marks, receipts fill later: +- D‑GOLDEN, D‑MOIRE, D‑THETA, D‑RHOENV (helix spacing + θ/ρ envelope — `jc`/`helix`) +- D‑BOUNDS, D‑META64, D‑BITGATE (helix bounds + `CausalEdge64` layout) +- D‑NEIGH (`blasgraph` scope) + +**Coded, done** — §2.8 D‑VOCAB…D‑KNOWABLE (+ vart‑backend), the adapters. + +**Queued code** (`IDEA`, gated): D‑OSM (runtime D‑OSM‑3), D‑PATTERN, +D‑ACTION, D‑NSM. + +**Do‑not‑build** (`S`): D‑OLED — catalog only. + +--- + +## 4. The OGAR / runtime boundary + +OGAR owns + verifies the **left**; the **right** is `[per runtime session]` +(cited as authoritative under the same discipline as ADR‑024's ρ anchor). + +| OGAR‑owned (verified) | Runtime‑owned `[per rt]` | +|---|---| +| `ogar-vocab` IR; all adapters; `ogar-knowable-from` + vart‑backend | `crates/helix` (placement template, golden stride) | +| Morton/XOR/CLZ address algebra (D‑MORTON, D‑XOR2) | `crates/jc` (`weyl` + `jirak`; θ/ρ envelope) | +| the amortization gate (D‑AMORT, fractal) | `cognitive-shader-driver` (the consumer) | +| the storage framing (D‑COLUMNAR, D‑LANCE) | `blasgraph` scope (D‑NEIGH) | +| ADR‑022/023/024/025 + the ADR‑026 spine | `CausalEdge64` layout (D‑META64); SoA headroom budget | + +**The five open ADR‑026 confirmations** (`CASCADE-SYNERGIES-EPIPHANY.md +§10`): helix spacing · θ‑window/ρ envelope · `blasgraph` scope · +shader tile‑contract · SoA headroom budget. (`CausalEdge64` cardinality ++ the nesting‑axis fork are **closed** — operator‑resolved.) + +--- + +## 5. The shape graph — the topology individual ADRs lose + +``` + ADR-022 boundary + │ (special case of) + ADR-023 IR ───────────┼─────────── ADR-024 codec + (address) │ (palette256/CAM) + │ │ │ + D-MORTON D-AMORT D-PAL256 ── D-RHO + D-XOR2 ───┐ (amortize gate) │ (ρ=0.9973) + │ │ │ (unifies │ + D-CASCADE │ 022+025) D-THETA ── D-RHOENV + │ │ │ │ (no-collapse) + D-IMMAT │ D-FRACTAL D-GOLDEN ─ D-MOIRE + │ │ (storage/cascade/bit) │ (X-Trans/φ) + D-COLUMNAR│ │ D-NOCOLLAPSE + D-LANCE │ │ │ + └─────┴───── ADR-025 selection ──────┘ + (probe-free r*) + │ + ADR-026 (pending) + = cascade + gate + meta→content + no-collapse + storage + │ + ┌───────────────┼───────────────┐ + D-CTU D-ATTN D-DELTA + (x265/x266) (attn-driven LOD) (version=P-frame) +``` + +The arrows are the links that vanish when you read ADR‑022..025 in +isolation. The map's job is to keep them visible. + +--- + +## 6. Maintenance discipline + +1. **Append‑only.** New discovery → new `D‑*` entry. Never delete; on + materialization, advance the *status* column (IDEA→EPIPHANY→ADR→CODED). +2. **Terse.** One line per entry. The shape, not the explanation. Link to + the home doc; do not duplicate its content here. +3. **Grade honestly.** `[S]` stays `[S]` until a measurement promotes it. + `[per runtime session]` until the runtime session confirms. +4. **This is the index.** The doctrine (ADRs), the synergies + (`CASCADE-SYNERGIES-EPIPHANY.md`), the IR (crates) are the content. The + map points; it does not re‑derive. +5. **The map mirrors the substrate.** Append‑only + status‑versioned, like + the audit log it documents (ADR‑008/013). If that feels recursive, it + is — and that's the tell that the shape is intact. + +--- + +## 7. Cross‑references + +- `docs/ARCHITECTURAL-DECISIONS-2026-06-04.md` — ADR‑022..025 (+ 026 pending). +- `docs/CASCADE-SYNERGIES-EPIPHANY.md` — the synergy catalog + amortization + gate + storage synthesis (the source of most §2 `EPIPHANY` entries). +- `docs/RDF-OWL-ALIGNMENT.md` — the brutal‑upgrade sequencing (§10 phases) + + Deep‑NSM (§4.10). +- `docs/DOMAIN-INSTANCES.md` — the 6 universality witnesses. +- `docs/THE-FIREWALL.md`, `HEALTHCARE-TRANSCODING.md` — ADR‑022 + PII floor. +- lance‑graph PR #470/#473/#474/#475/#476/#477/#478, bardioc #17/#18/#19 — + the runtime‑side receipts (`[per runtime session]` sources). From e80d8d8a630885631a2b6262383446410e746afe Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Mon, 8 Jun 2026 21:51:31 +0000 Subject: [PATCH 02/11] =?UTF-8?q?docs(discovery-map):=20=C2=A74.1=20runtim?= =?UTF-8?q?e=20synergy=20receipts=20=E2=80=94=20lance-graph=20PR=20#477/#4?= =?UTF-8?q?78?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Cross-referenced PR #477 (SoA envelope + HHTL audit + three-tier model) and #478 (singleton-to-snapshot) against today's D-* findings. Result: several findings are confirmed by runtime CODE (not just summaries), one is revised, one gradient is convergent-but-unwired on both sides. Confirmations [G] — runtime code grounds the OGAR finding: - D-MORTON / D-CASCADE / D-XOR2(containment): hhtl.rs ships NiblePath{path:u64,depth:u8}, FAN_OUT=16, MAX_DEPTH=16 (16 nibbles = 64 bits), parent()=path>>4, child(n)=(path<<4)|n, is_ancestor_of = a single prefix-shift kernel, common_ancestor = LCA. The nibble address algebra is CODED runtime-side. - D-IMMAT / D-COLUMNAR / ADR-022: the three-tier-model invariant ("zero-copy creation->tombstone; nothing serialized; in-memory store unchanged") IS the immaterialized finding; emit()/ CollapseGateEmission scheduled for removal (the serialization path is being deleted). - D-COLUMNAR / D-LANCE: soa_envelope.rs SoaEnvelope trait — columns() -> &[ColumnDescriptor], repr(C) descriptors, zero-copy row_le/ column_le views, as_le_bytes(), verify_layout() gate. The Morton-keyed columnar+projection framing is the ABI line-for-line. - D-DELTA: cycle()->u32, "a Lance version IS a coherent LE envelope", last_active_cycle[u32;N] per-row write guard, DatasetVersion(v)-> (v+1). Version=frame, per-row cycle stamp=changed-cell delta. Promotes D-DELTA [H]->[G]. - D-AMORT: PR #478 test (never-mutated->const codebook keep; mutated ->SoA snapshot nudge) operationalizes the amortization gate's storage instance. Revision [warn] — D-META64: MailboxSoA has SEPARATE columns edges[CausalEdge64;N] and meta[MetaWord;N], so the meta layer is a dedicated MetaWord, NOT CausalEdge64 (the edge encoding). The "CausalEdge64 = meta" reading is contradicted by the shipped column layout. Reconcile: the 2^6/48/16 bit-budget likely applies to MetaWord. Updated the D-META64 ledger entry to H->REVISE with the correction. Convergent-but-unwired [next step] — D-MAXAMORT/D-CARD/D-PAL256 (the meta->content->spatial gradient) matches high_heel.rs's module-doc legend "HEEL=scent / HIP=palette / TWIG=SpoBase17 / LEAF=full planes", but the q4 audit says that legend is a COMMENT — "no code routes by prefix." HIP=palette is literally D-PAL256 at a cascade level. Highest- value next materialization: route high_heel.rs's commented cascade by hhtl.rs's NiblePath prefix; today's findings are the spec that joins the two files (which "communicate in module docs, not in code"). Related structure: the three-tier model (data/lifecycle/supervision) is orthogonal to the meta->content->spatial gradient — noted, not conflated. Kanban 6-phase = the ActionState lifecycle (ADR-001). qualia[QualiaI4_16D;N] = the Quintenzirkel qualia codebook (#470). Docs-only. PII abort-guard (word-boundary): CLEAN. cargo check: clean. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- docs/DISCOVERY-MAP.md | 57 ++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 56 insertions(+), 1 deletion(-) diff --git a/docs/DISCOVERY-MAP.md b/docs/DISCOVERY-MAP.md index 48c2545..ad62a8e 100644 --- a/docs/DISCOVERY-MAP.md +++ b/docs/DISCOVERY-MAP.md @@ -85,7 +85,7 @@ two halves of a cell. ADR‑026 names the cascade that ties them. | D‑PAL256 | palette256 = one PQ subspace's 256 centroids (CAM = 6×256, PR #477) | G | ADR | ADR‑024 | — | | D‑CAM | CAM‑PQ = 6 roles × 256 = 48‑bit fingerprint per word (`cam_codes.bin`) | G | EPIPHANY | SYN §3 | D‑PAL256 | | D‑CARD | cascade per‑axis cardinality ↔ codebook size: 256↔palette, 4096↔`MAX_VOCAB` | G | EPIPHANY | SYN §3,§7 | D‑CASCADE, D‑CAM | -| D‑META64 | `CausalEdge64` = meta: `2⁶` role‑mask **+** 48‑bit CAM + 16‑bit headroom, one 64‑bit word | H | EPIPHANY | SYN §9.6 | D‑CAM, `[per rt]` CausalEdge64 | +| D‑META64 | meta = `2⁶` role‑mask + 48‑bit CAM + 16‑bit headroom. **⚠ PR #477 revises:** `MailboxSoA` has **separate** `edges [CausalEdge64;N]` *and* `meta [MetaWord;N]` columns → **the meta is `MetaWord`, not `CausalEdge64`** (the edge encoding). Reconcile (§4.1). | H → REVISE | SYN §9.6 + §4.1 | D‑CAM, `[per rt]` | | D‑BITGATE | 48→64 headroom: store the seed **iff** irreducible‑beyond‑the‑address (else compute) | H | EPIPHANY | SYN §9.6 | D‑META64, D‑AMORT | | D‑RHO | ρ = 0.9973 vs cosine (arm‑discovery aerial codebook) — the empirical anchor | G | ADR | ADR‑024 | D‑PAL256 | @@ -205,6 +205,61 @@ shader tile‑contract · SoA headroom budget. (`CausalEdge64` cardinality --- +## 4.1 Runtime synergy receipts — lance‑graph PR #477 / #478 + +Cross‑reference of the runtime‑shipped contract against today's findings. +Several findings are now **confirmed by runtime code** (promotes their +grade/status); one is **revised**; one gradient is **convergent‑but‑unwired +on both sides** (the next materialization). + +### Confirmations — runtime code grounds the OGAR finding [G] + +| Finding | Runtime receipt (PR #477) | Verdict | +|---|---|---| +| **D‑MORTON, D‑CASCADE, D‑XOR2 (containment)** | `lance-graph-contract/src/hhtl.rs`: `NiblePath { path:u64, depth:u8 }`, `FAN_OUT=16`, `MAX_DEPTH=16` (16 nibbles = 64 bits), `parent()=path>>4`, `child(n)=(path<<4)\|n`, `is_ancestor_of` = a single prefix‑shift kernel, `common_ancestor` = LCA. (q4‑hhtl‑audit) | **the runtime SHIPS the nibble address algebra** — these are CODED runtime‑side, not just EPIPHANY | +| **D‑IMMAT, D‑COLUMNAR, ADR‑022** | `soa-three-tier-model.md` invariant: *"zero‑copy from creation to Lance tombstone … no bytes leave the backing store … even then the in‑memory store is unchanged, not serialized and freed."* `MailboxSoA::emit()` + `CollapseGateEmission` **scheduled for removal** | the "immaterialized, never serialized" finding IS the runtime's ONE invariant; the serialization path is being **deleted** | +| **D‑COLUMNAR, D‑LANCE (column projection)** | `soa_envelope.rs`: `SoaEnvelope` trait — `columns()->&[ColumnDescriptor]`, `ColumnDescriptor{name_id,kind,elems_per_row,row_offset}` (`repr(C)`), zero‑copy `row_le`/`column_le` views, `as_le_bytes()`, `verify_layout()` gate | the "Morton‑keyed columnar + projection" framing is the `SoaEnvelope` ABI, line‑for‑line | +| **D‑DELTA (delta frames = version‑diff)** | `cycle()->u32` stamp; *"a Lance version IS a coherent LE envelope"* (soa_envelope L16); `last_active_cycle [u32;N]` = per‑row same‑cycle write guard; `DatasetVersion(v)→(v+1)` | a Lance version = a frame; the per‑row cycle stamp = the changed‑cell delta → **D‑DELTA promotes [H]→[G]** | +| **D‑AMORT (amortize‑or‑don't‑spend)** | PR #478 mechanical test: *never‑mutated → const codebook, **keep**; mutated‑after‑init → SoA snapshot, **nudge**.* Read‑only codebooks (role keys, UDFs, `simd_caps`) stay const | the gate's storage instance, **operationalized** — const codebooks = the amortized build‑once structures; SoA = per‑cycle mutable | + +### Revision — runtime evidence corrects a finding ⚠ + +- **D‑META64.** `MailboxSoA` columns include **both** `edges [CausalEdge64;N]` + **and** `meta [MetaWord;N]` as *separate* columns. So **the meta layer is a + dedicated `MetaWord`, not `CausalEdge64`** (which is the causal‑edge + encoding). The operator's "CausalEdge64 = meta" is contradicted by the + shipped column layout. **Reconcile:** the `2⁶`/48/16 bit‑budget reasoning + (D‑BITGATE) most likely applies to `MetaWord`; `CausalEdge64` is the edge. + `[per runtime session]` on the `MetaWord` layout + whether `CausalEdge64` + feeds it. + +### Convergent but unwired — both sides describe it; neither built it [the next step] + +- **D‑MAXAMORT / D‑CARD / D‑PAL256** (the meta→content→spatial gradient) ↔ + `high_heel.rs` module‑doc legend **"HHTL cascade mapping: HEEL=scent / + HIP=palette / TWIG=SpoBase17 / LEAF=full planes."** The q4 audit's verdict: + *"this is a layout legend in a comment. **No code routes by prefix.**"* So + the coarse→fine cascade (HEEL→HIP→TWIG→LEAF ≈ meta→**palette**→SPO→full) is + **convergent intent on both sides and wired on neither.** `HIP = palette` + is literally D‑PAL256 at a cascade level. **This is the highest‑value + next materialization:** route `high_heel.rs`'s commented cascade by the + `hhtl.rs` `NiblePath` prefix — the two files "communicate in module docs, + not in code" (audit), and today's findings are the spec that joins them. + +### Related structure (resonant, different decomposition) + +- **Three‑tier model** (Tier 1 `MailboxSoA` data / Tier 2 Kanban‑Rubicon + lifecycle / Tier 3 ractor supervision) is a *data / lifecycle / + supervision* split, **orthogonal** to the meta→content→spatial gradient — + note both, don't conflate. The Kanban 6‑phase (Planning → CognitiveWork → + Evaluation → Commit → Plan → Prune) = the `ActionState` lifecycle (the + OGAR Rubicon binding, ADR‑001). +- `qualia [QualiaI4_16D;N]` (4‑bit, 16‑D) = the Quintenzirkel qualia codebook + (BindSpace dissolution, lance‑graph #470) — a codebook column alongside + `edges`/`meta`; relates to D‑CARD's codebook cascade. + +--- + ## 5. The shape graph — the topology individual ADRs lose ``` From fa5b41c0258977f9c71286cc83451ae8b6d9f676 Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Mon, 8 Jun 2026 21:57:51 +0000 Subject: [PATCH 03/11] =?UTF-8?q?docs(discovery-map):=20=C2=A70=20the=20en?= =?UTF-8?q?dgame=20=E2=80=94=20DeepNSM=20+=20AriGraph=20(think)=20->=20AST?= =?UTF-8?q?=20named=20facts=20(do)?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Operator named the teleology the whole substrate serves: "endgame is actually getting DeepNSM + AriGraph as a meta vs episodic basins and supporters, meaning vectors, and then later extraction of AST-shaped named facts — do vs think." New §0 (the map's apex — the WHY above the doctrine spine): THINK — the DeepNSM + AriGraph memory (the world model): AriGraph (arXiv 2407.04363, Anokhin et al. 2024 — semantic + episodic KG world-model with associative retrieval for LLM-agent planning), mapped onto the substrate: - semantic memory -> META = MetaWord (the §4.1 layer) - episodic memory -> EPISODIC = delta frames = DatasetVersion(v)->(v+1) + cycle stamp (D-DELTA) - associative retrieval -> basins (palette256/CAM centroids + Semantik volumetric centroids + Morton- prefix subtrees; intra-basin locality) + supporters (neighbor-XOR retrieval, D-NEIGH) - meaning vectors -> DeepNSM 4096-dim (63 primes) -> CAM-PQ 6x256 (D-NSM, D-CAM) — shipping in lance-graph nsm/ DO — the extracted AST-shaped named facts (the actions): the OGAR IR (Class/ActionDef/Association, D-VOCAB) extracted from the meaning-vector memory by D-PATTERN + D-ACTION. The named fact is the wire (ADR-023) where THINK (vectors) becomes DO (action). The master axis DO vs THINK: THINK = the meaning-vector memory (what is known) = structural arm = Semantik. DO = ActionDef / Rubicon lifecycle / Kanban 6-phase (what is done) = behavioral arm = Pragmatik. The AST-named-fact extraction is the membrane between them. Reprioritization recorded: D-NSM, D-PATTERN, D-ACTION (listed IDEA/ queued in §2.8) are NOT optional extras — they are the endgame-critical path (the think->do extraction). The addressing/codec/storage floors are the foundation they stand on, not the destination. AriGraph + DeepNSM added to §7 cross-references with the arXiv id and the think/do <-> structural/behavioral <-> Semantik/Pragmatik mapping. Docs-only. PII abort-guard (word-boundary): CLEAN. cargo check: clean. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- docs/DISCOVERY-MAP.md | 51 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 51 insertions(+) diff --git a/docs/DISCOVERY-MAP.md b/docs/DISCOVERY-MAP.md index ad62a8e..a315c62 100644 --- a/docs/DISCOVERY-MAP.md +++ b/docs/DISCOVERY-MAP.md @@ -31,6 +31,50 @@ --- +## 0. The endgame — what the whole substrate is *for* + +> **Operator, 2026‑06‑08:** *"endgame is actually getting DeepNSM + AriGraph +> as a meta vs episodic basins and supporters, meaning vectors, and then +> later extraction of AST‑shaped named facts — **do vs think**."* + +Everything in §1 (the doctrine) and §2 (the discoveries) is the **foundation +for a cognitive memory architecture.** The teleology, named — and it +reprioritizes the queue. + +**THINK — the DeepNSM + AriGraph memory** (the substrate's *world model*): + +| AriGraph (arXiv 2407.04363, Anokhin et al. 2024 — semantic + episodic KG world‑model with associative retrieval, for LLM‑agent planning) | Substrate realization | Grade | +|---|---|---| +| **semantic memory** (the world model) | **META = `MetaWord`** (the §4.1 layer — structural awareness) | mapping H | +| **episodic memory** (event sequence) | **EPISODIC = delta frames** = `DatasetVersion(v)→(v+1)` + per‑row `cycle` stamp (D‑DELTA) | H | +| **associative retrieval of interconnected concepts** | **basins + supporters**: basins = palette256/CAM centroids + Semantik volumetric centroids + Morton‑prefix subtrees (D‑PAL256, D‑CAM, D‑AMORT‑AXES; the intra‑basin locality probe); supporters = the neighbor‑XOR retrieval (D‑NEIGH) | H | +| **the encoded concepts** | **meaning vectors** = DeepNSM 4096‑dim (63 primes) → CAM‑PQ 6×256 (D‑NSM, D‑CAM) — *shipping* in lance‑graph `nsm/` | G (DeepNSM), G (AriGraph public) | + +**DO — the extracted AST‑shaped named facts** (the substrate's *actions*): +*"later extraction of AST‑shaped named facts"* = the OGAR IR (`Class` / +`ActionDef` / `Association` — D‑VOCAB), **extracted from the meaning‑vector +memory** by the recognition + lifecycle pipeline (D‑PATTERN, D‑ACTION). The +named fact is the wire (ADR‑023): the point where THINK (vectors) becomes +DO (action). + +**The master axis — DO vs THINK:** + +| | THINK | DO | +|---|---|---| +| substrate | DeepNSM + AriGraph meaning‑vector memory (meta / episodic / basins / supporters) | `ActionDef` / Rubicon lifecycle / Kanban 6‑phase | +| OGAR IR arm | structural (`Class`/`Association`) | behavioral (`ActionDef`/`ActionInvocation`) | +| trichotomy | **Semantik** (what is known) | **Pragmatik** (what is done) | +| membrane | — | **AST‑named‑fact extraction is the membrane between them** | + +**What this reprioritizes.** `D‑NSM`, `D‑PATTERN`, `D‑ACTION` (listed +`IDEA`/queued in §2.8) are **not optional extras — they are the +endgame‑critical path** (the *think → do* extraction). The +addressing / codec / storage floors (§2.1–§2.6, §1) are the **foundation +they stand on**, not the destination. The destination is: *the substrate +thinks in meaning‑vector basins and acts via extracted AST‑named‑facts.* + +--- + ## 1. The doctrine spine — the load‑bearing floors The ADR canon is the spine; every discovery in §2 hangs off the floor it @@ -321,3 +365,10 @@ isolation. The map's job is to keep them visible. - `docs/THE-FIREWALL.md`, `HEALTHCARE-TRANSCODING.md` — ADR‑022 + PII floor. - lance‑graph PR #470/#473/#474/#475/#476/#477/#478, bardioc #17/#18/#19 — the runtime‑side receipts (`[per runtime session]` sources). +- **The endgame references (§0):** AriGraph — *Learning Knowledge Graph + World Models with Episodic Memory for LLM Agents*, Anokhin, Semenov, + Sorokin, Evseev, Burtsev, Burnaev, arXiv **2407.04363** (2024). DeepNSM + — lance‑graph `crates/lance-graph/src/nsm/` (`MAX_VOCAB=4096`, + `NUM_PRIMES=63`, `NUM_ROLES=6`, CAM 6×256; PR #477). The `think`/`do` + axis ↔ OGAR structural/behavioral IR arms (`ogar-vocab`) + the + Semantik/Pragmatik trichotomy (`CHESS-TRANSCODING.md §0`). From a36698f2f63f468126a58156f0acd5a6573bb6c0 Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Tue, 9 Jun 2026 04:30:35 +0000 Subject: [PATCH 04/11] docs(discovery-map): identify the SYN source doc + flag the cross-PR dependency MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Closes Codex P2 on PR #48 (line 361). Codex: the SYN §... Home links + the §7 cross-ref point to docs/CASCADE-SYNERGIES-EPIPHANY.md, which a repo-wide search from main does not find — so the map's primary source links look unresolvable. Correct: that file is introduced in OGAR PR #47 and co-merges with this map (PR #48); the SYN links resolve once both land. It is a cross-PR dependency, not a dangling link. Fix (makes the dependency explicit so no reader has to guess): - Added a "How to read" legend entry defining SYN §... = docs/CASCADE-SYNERGIES-EPIPHANY.md §..., stating it is introduced in PR #47 and co-merges with PR #48, and that a reviewer auditing from main alone will not find it yet (expected). - Annotated the §7 cross-reference line: "(introduced in OGAR PR #47, co-merges with this map)". No content links changed — once #47 + #48 are both on main the SYN references resolve normally. This only documents the ordering so the window between the two merges isn't confusing. Docs-only. PII abort-guard (word-boundary): CLEAN. cargo check: clean. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- docs/DISCOVERY-MAP.md | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/docs/DISCOVERY-MAP.md b/docs/DISCOVERY-MAP.md index a315c62..a078fef 100644 --- a/docs/DISCOVERY-MAP.md +++ b/docs/DISCOVERY-MAP.md @@ -24,6 +24,12 @@ > captured in a synergy/epiphany doc) → `ADR` (pinned contract) → `CODED` > (crate + tests + CI). > - **`[per runtime session]`** = runtime‑owned; OGAR cannot verify. +> - **`SYN §…` = `docs/CASCADE-SYNERGIES-EPIPHANY.md` §…** — the synergy / +> epiphany source doc that holds most `EPIPHANY`‑status entries. **It is +> introduced in OGAR PR #47 and co‑merges with this map (PR #48); the +> `SYN` links resolve once both land.** Until then a reviewer auditing +> from `main` alone will not find it — that is expected (cross‑PR +> dependency), not a dangling link. (Flagged by Codex on PR #48.) > > **Status: LIVING INDEX** (2026‑06‑08). Update entry *status* on > materialization; never delete (append‑only). One entry per discovery; @@ -357,8 +363,9 @@ isolation. The map's job is to keep them visible. ## 7. Cross‑references - `docs/ARCHITECTURAL-DECISIONS-2026-06-04.md` — ADR‑022..025 (+ 026 pending). -- `docs/CASCADE-SYNERGIES-EPIPHANY.md` — the synergy catalog + amortization - gate + storage synthesis (the source of most §2 `EPIPHANY` entries). +- `docs/CASCADE-SYNERGIES-EPIPHANY.md` **(introduced in OGAR PR #47, co‑merges + with this map)** — the synergy catalog + amortization gate + storage + synthesis (the source of most §2 `EPIPHANY` entries; the `SYN §…` links). - `docs/RDF-OWL-ALIGNMENT.md` — the brutal‑upgrade sequencing (§10 phases) + Deep‑NSM (§4.10). - `docs/DOMAIN-INSTANCES.md` — the 6 universality witnesses. From bf97d5f7b92d585e8c873f836a5e134285080448 Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Tue, 9 Jun 2026 04:33:25 +0000 Subject: [PATCH 05/11] =?UTF-8?q?docs(discovery-map):=20add=20D-MANTISSA?= =?UTF-8?q?=20=E2=80=94=20cell=20address=20as=20float=20(exponent=3Dlevel,?= =?UTF-8?q?=20implied=20irrational=20mantissa=3Dgolden=20offset)?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Operator framing (2026-06-08): X-Trans 'in spirit' = an implied anti-moire irrational mantissa. Captured as D-MANTISSA: a cell address is a float — (exponent = Morton level, mantissa = golden sub-placement); the mantissa is irrational (phi recurrence) and implied (generated from the address, not stored — the IEEE hidden-bit spirit; D-BITGATE compute- don't-store). The exponent=level mapping is exact (LOD is logarithmic, r*=ceil(log4)); the mantissa=within-cell-offset is the fine fractional placement. 'Irrational mantissa' is coherent precisely because it's implied/generative (you can't store infinite precision; you generate it from the golden recurrence on demand). The anti-moire adjective stays [H] not a guarantee (the #47 Codex caveat: X-Trans is itself a repeating tile; a bounded cell has a finite spectrum that can beat the LOD lattice). So the honest name is 'implied irrational mantissa, hypothesized anti-moire.' Unifies four ledger entries: D-CASCADE (the exponent range) + D-GOLDEN (the mantissa generator) + D-BITGATE (why the mantissa is implied) + D-NOCOLLAPSE (the hypothesized anti-moire function). The cell-address- as-float is the compression of all four. Docs-only. PII: CLEAN. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- docs/DISCOVERY-MAP.md | 1 + 1 file changed, 1 insertion(+) diff --git a/docs/DISCOVERY-MAP.md b/docs/DISCOVERY-MAP.md index a078fef..23ff2d0 100644 --- a/docs/DISCOVERY-MAP.md +++ b/docs/DISCOVERY-MAP.md @@ -148,6 +148,7 @@ two halves of a cell. ADR‑026 names the cascade that ties them. | D‑GOLDEN | golden‑stride helix placement controls θ into the window **by construction** | H | EPIPHANY | SYN §2 | `[per rt]` helix | | D‑MOIRE | golden irrationality = X‑Trans‑grade anti‑moiré interlace ("x256 that can't collapse") | H | EPIPHANY | SYN §2 | D‑GOLDEN | | D‑NOCOLLAPSE | unify: near‑orthogonal **codebook** + aperiodic **lattice** = no degenerate beat in value‑ or position‑space | H | EPIPHANY → ADR‑026 §2 | SYN §9.1 | D‑THETA, D‑MOIRE | +| D‑MANTISSA | a cell address **is a float**: `(exponent = Morton level, mantissa = golden sub‑placement)`; the mantissa is **irrational** (φ recurrence) + **implied** (generated from the address, not stored — the hidden‑bit spirit). X‑Trans "in spirit" = an *implied irrational mantissa*; the **anti‑moiré** function stays `[H]` (§2 caveat). Unifies D‑CASCADE (exponent) + D‑GOLDEN (mantissa) + D‑BITGATE (implied) + D‑NOCOLLAPSE | H | EPIPHANY | SYN §2 | D‑CASCADE, D‑GOLDEN, D‑BITGATE | ### 2.5 Amortization gate From f8d3a1fbeddf0ba0150d2e1922ca96800bb1fef0 Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Tue, 9 Jun 2026 04:38:55 +0000 Subject: [PATCH 06/11] =?UTF-8?q?docs(discovery-map):=20add=20D-BGZ17=20?= =?UTF-8?q?=E2=80=94=20coprime-aperiodic=20(Base17)=20immutable=20anti-moi?= =?UTF-8?q?re=20quorum?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Operator proposal (2026-06-08): use bgz17 / HighHeelBGZ / SpoBase17 as an immutable irrational anti-moire quorum in the Morton Cascade. Captured as D-BGZ17 — the DISCRETE/IMMUTABLE counterpart to D-MANTISSA's continuous golden mantissa. Grounded in shipped lance-graph code (high_heel.rs, PR #477): - HighHeelBGZ = 'one family basin (TWIG centroid) + up to 240 causal edges' = literally the §0 AriGraph basin + supporters, in code. - SpoBase17 = three Base17 planes [i16;17] (S,P,O), ρ=0.965 vs full planes — a second measured exactness anchor (joins palette256 ρ=0.9973). - HEEL/HIP/TWIG/LEAF cascade legend real (doc-comment); TWIG=SpoBase17; LEAF=full planes 'computed on demand, not stored'. The sharpening (and the point): NOT 'irrational' but COPRIME-APERIODIC. 17 is prime, coprime to the base-16 Morton nibble: gcd(16,17)=1 => LCM(16,17)=272 => the moire beat is pushed to period 272 (the longest possible for the pair = lowest frequency = least visible). 272 = 16x17 = the BIT_WIDTH_PALETTE_BASE17 constant (MedCare F-2). This is the classic coprime-prime moire-spreading technique (halftoning/CFA), done DISCRETELY and EXACTLY (i16 planes, const, hashable) rather than via a continuous float irrational. Grade upgrade with caveat: anti-moire goes from [H]-unmeasured (golden) to [H] with a PROVABLE BOUND (number theory: beat pushed to period 272), but the #47 Codex caveat persists — coprimality SPREADS the beat, does not eliminate it. Honest claim: 'provably-longest-period (lowest-freq) moire,' not 'no moire.' Why it fits 'immutable' better than D-MANTISSA: discrete coprimality is exact/hashable/const; the golden mantissa is float/measured. For the immutable substrate, Base17 is the one you ship. They are the discrete and continuous routes to anti-aperiodicity (could be complementary: Base17 for the codebook/basin, golden for continuous sub-placement). Resolves the §4.1 unwired gap: HEEL/HIP/TWIG/LEAF exists as a container but 'no code routes by prefix' (q4 audit). This proposal IS that wiring — route HighHeelBGZ by the hhtl.rs NiblePath, Base17(TWIG) as the coprime anti-moire quorum layer. Realizes §0: HighHeelBGZ = basin+ supporters; SpoBase17 = meaning vector; scent() L1-prefilter = associative retrieval. [per runtime session] on bgz17/HighHeelBGZ/SpoBase17 internals + whether 17 was chosen for coprimality (the code uses 17; the anti-moire-via- coprimality is the recognized property, not a documented original intent). Docs-only. PII: CLEAN. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- docs/DISCOVERY-MAP.md | 1 + 1 file changed, 1 insertion(+) diff --git a/docs/DISCOVERY-MAP.md b/docs/DISCOVERY-MAP.md index 23ff2d0..6cc7391 100644 --- a/docs/DISCOVERY-MAP.md +++ b/docs/DISCOVERY-MAP.md @@ -149,6 +149,7 @@ two halves of a cell. ADR‑026 names the cascade that ties them. | D‑MOIRE | golden irrationality = X‑Trans‑grade anti‑moiré interlace ("x256 that can't collapse") | H | EPIPHANY | SYN §2 | D‑GOLDEN | | D‑NOCOLLAPSE | unify: near‑orthogonal **codebook** + aperiodic **lattice** = no degenerate beat in value‑ or position‑space | H | EPIPHANY → ADR‑026 §2 | SYN §9.1 | D‑THETA, D‑MOIRE | | D‑MANTISSA | a cell address **is a float**: `(exponent = Morton level, mantissa = golden sub‑placement)`; the mantissa is **irrational** (φ recurrence) + **implied** (generated from the address, not stored — the hidden‑bit spirit). X‑Trans "in spirit" = an *implied irrational mantissa*; the **anti‑moiré** function stays `[H]` (§2 caveat). Unifies D‑CASCADE (exponent) + D‑GOLDEN (mantissa) + D‑BITGATE (implied) + D‑NOCOLLAPSE | H | EPIPHANY | SYN §2 | D‑CASCADE, D‑GOLDEN, D‑BITGATE | +| D‑BGZ17 | the **discrete/immutable** counterpart to D‑MANTISSA: not "irrational" but **coprime‑aperiodic** — Base17 (`SpoBase17 = [i16;17]×3`, prime 17 ⊥ base‑16 Morton) ⟹ beat period `LCM(16,17)=272` (longest → lowest‑freq moiré). **Anti‑moiré upgrades `[H]`‑unmeasured → `[H]` with a *provable bound*** (number theory, not spectral) — still "longest‑period beat," not "no beat" (§2 caveat persists). `HighHeelBGZ` (basin + ≤240 edges) **realizes §0 AriGraph basin+supporters**; the "quorum" = the L1‑threshold basin‑merge consensus set. **Resolves the §4.1 unwired gap** (route the HEEL/HIP/TWIG/LEAF container by the `hhtl.rs` NiblePath; Base17=TWIG is the anti‑moiré layer). ρ=0.965 (TWIG) joins ρ=0.9973 (HIP) as measured anchors. For *immutable*, prefer this over D‑MANTISSA (exact/hashable vs float). | H | EPIPHANY | SYN §2 + §4.1 + §0 | D‑MANTISSA, D‑CAM, `[per rt]` `bgz17`/`HighHeelBGZ`/`SpoBase17` | ### 2.5 Amortization gate From 11ff3a405c1a72937cfd36676e8bae172e1821ab Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Tue, 9 Jun 2026 04:50:48 +0000 Subject: [PATCH 07/11] docs(discovery-map): add D-QUANTGATE + D-EXCITON MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Two operator inputs (2026-06-08): 1. D-QUANTGATE — 'whenever there's quantization, the irrational doesn't have the Morton tile cascade guarantee, but being an irrational muscle memory mantissa quorum.' This is a STRUCTURAL principle, not a hypothesis: continuous irrationals (golden mantissa, D-MANTISSA) lose their aperiodicity when quantized onto a lattice — the rounded value re-acquires the grid period. So at every quantized layer the substrate MUST use the discrete coprime route (D-BGZ17, Base17 x Morton) to preserve aperiodicity. The continuous irrational only survives as a 'muscle-memory' quorum: pre-computed at codebook build time, never re-quantized in the hot path (matches D-AMORT). Promotes ADR-026 §2 no-collapse from a single precondition to a LAYERED rule: discrete-coprime at quantized boundaries; muscle-memory irrationals at build-time / continuous layers. Reframes the §2 anti-moire caveat as ARCHITECTURAL, not just unmeasured. Grade [G] (the principle is structural). 2. D-EXCITON — operator: 'isn't OLED excitons also some irrational bundling?' Web-searched (2026-06-08): the standard 1:3 singlet: triplet ratio is quantum spin statistics (rational); but under bias the formation ratio DEVIATES from 1:3 (singlets scale with bias, triplets bias-independent — PMC4614446), giving a continuously- variable non-integer S:T ratio per operating point. Plus singlet/ triplet binding energies differ irreducibly (~0.5 eV vs up to 1.5 eV — ScienceDirect/noctiluca). So OLED ships, under bias, the SAME discrete coprime aperiodicity D-BGZ17 names: non-commensurate spin populations + non-commensurate binding energies + broken from rational 1:3 by a continuous control parameter. The leg I demoted to [S] in SYN §3 ('not sure what we can learn from excitons') was too harsh — promote to [H] shape-match. Companion SYN §3 revision lives on the #47 branch. Docs-only. PII: CLEAN. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- docs/DISCOVERY-MAP.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/docs/DISCOVERY-MAP.md b/docs/DISCOVERY-MAP.md index 6cc7391..c608c93 100644 --- a/docs/DISCOVERY-MAP.md +++ b/docs/DISCOVERY-MAP.md @@ -150,6 +150,8 @@ two halves of a cell. ADR‑026 names the cascade that ties them. | D‑NOCOLLAPSE | unify: near‑orthogonal **codebook** + aperiodic **lattice** = no degenerate beat in value‑ or position‑space | H | EPIPHANY → ADR‑026 §2 | SYN §9.1 | D‑THETA, D‑MOIRE | | D‑MANTISSA | a cell address **is a float**: `(exponent = Morton level, mantissa = golden sub‑placement)`; the mantissa is **irrational** (φ recurrence) + **implied** (generated from the address, not stored — the hidden‑bit spirit). X‑Trans "in spirit" = an *implied irrational mantissa*; the **anti‑moiré** function stays `[H]` (§2 caveat). Unifies D‑CASCADE (exponent) + D‑GOLDEN (mantissa) + D‑BITGATE (implied) + D‑NOCOLLAPSE | H | EPIPHANY | SYN §2 | D‑CASCADE, D‑GOLDEN, D‑BITGATE | | D‑BGZ17 | the **discrete/immutable** counterpart to D‑MANTISSA: not "irrational" but **coprime‑aperiodic** — Base17 (`SpoBase17 = [i16;17]×3`, prime 17 ⊥ base‑16 Morton) ⟹ beat period `LCM(16,17)=272` (longest → lowest‑freq moiré). **Anti‑moiré upgrades `[H]`‑unmeasured → `[H]` with a *provable bound*** (number theory, not spectral) — still "longest‑period beat," not "no beat" (§2 caveat persists). `HighHeelBGZ` (basin + ≤240 edges) **realizes §0 AriGraph basin+supporters**; the "quorum" = the L1‑threshold basin‑merge consensus set. **Resolves the §4.1 unwired gap** (route the HEEL/HIP/TWIG/LEAF container by the `hhtl.rs` NiblePath; Base17=TWIG is the anti‑moiré layer). ρ=0.965 (TWIG) joins ρ=0.9973 (HIP) as measured anchors. For *immutable*, prefer this over D‑MANTISSA (exact/hashable vs float). | H | EPIPHANY | SYN §2 + §4.1 + §0 | D‑MANTISSA, D‑CAM, `[per rt]` `bgz17`/`HighHeelBGZ`/`SpoBase17` | +| D‑QUANTGATE | **the quantization principle** (operator, 2026‑06‑08): **whenever there is quantization, the irrational does not have the Morton tile cascade guarantee.** Continuous irrationals (golden mantissa / D‑MANTISSA) lose their aperiodicity when quantized onto a lattice — the rounded value re‑acquires the period of the grid. So at every quantized layer, the substrate must use the **discrete coprime route** (D‑BGZ17, Base17×Morton) to preserve aperiodicity. The continuous irrational only survives as a *muscle‑memory* mantissa quorum: pre‑computed at codebook build time and **never re‑quantized in the hot path** (consistent with the amortization gate D‑AMORT). This **promotes ADR‑026's no‑collapse precondition (D‑NOCOLLAPSE) into a layered rule**: discrete‑coprime at every quantized boundary; muscle‑memory irrationals only at build‑time / continuous layers. **Reframes the §2 anti‑moiré caveat as architectural, not just unmeasured.** | G (the principle is structural) | EPIPHANY → ADR‑026 §2 | SYN §2 | D‑MANTISSA, D‑BGZ17, D‑AMORT, D‑NOCOLLAPSE | +| D‑EXCITON | OLED **excitons as irrational bundling** (operator + measured: web search 2026‑06‑08, [noctiluca / ScienceDirect / PMC] — see §7). The 1:3 singlet:triplet ratio is *quantum spin statistics*, not aperiodic; **but** under bias the formation ratio **deviates from 1:3** (singlets scale with bias, triplets are bias‑independent — PMC4614446), giving a continuously‑variable non‑integer S:T ratio per operating point — *that's* the irrational bundling. Plus the singlet/triplet **binding energies differ irreducibly** (~0.5 eV vs up to 1.5 eV — ScienceDirect/noctiluca), an energetic aperiodicity per exciton. So the OLED leg is **not** "exciton physics maps to substrate" (which I had as `[S]` in SYN §3) but: **under bias, OLEDs ship the *same* discrete coprime aperiodicity we just named in D‑BGZ17 — non‑commensurate spin populations, non‑commensurate binding energies, broken from the rational 1:3 by a continuous control parameter.** That's a `[H]` shape‑match, **not** `[S]`. Promotes the SYN §3 OLED leg. | H | EPIPHANY | SYN §3 (revise) | D‑BGZ17, D‑QUANTGATE | ### 2.5 Amortization gate From bc6171d627ff1ec9986263bcb44b7572555ad702 Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Tue, 9 Jun 2026 05:05:34 +0000 Subject: [PATCH 08/11] =?UTF-8?q?docs(discovery-map):=20=C2=A74.2=20test-m?= =?UTF-8?q?apping=20suite=20=E2=80=94=20jc=20proofs=20x=20ndarray::hpc=20e?= =?UTF-8?q?mpirics=20+=20audit?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Operator: 'double check the mapping/arguments/groundings so it doesn't dilute nor collapse, and include the necessary test mappings using the JC crate pillars in lance-graph and hpc crate pillars in ndarray.' Recon (both real): - jc (lance-graph crates/jc/src/lib.rs) = 'Jirak-Cartan five-pillar [11] proof-in-code', run via 'cargo run -p jc --example prove_it', each pillar -> PillarResult. Key pillars: P3 phi-Weyl (optimal collocation WITHOUT ALIASING), P5 Jirak (bounded noise floor), P5b Pearl 2^3 (Index vs CAM-PQ regime), P9/9b EWA-sandwich (Sigma push-forward; P9b explicitly certifies ndarray::hpc::splat3d), P10 Pflug (CAM-PQ tree quantization preserves FreeEnergy within Le). - ndarray::hpc (ndarray/src/hpc/) = fft, lapack(Cholesky), quantized (Int8/BF16 GEMM), cascade (Morton), plane (Base17/16Kbit), fingerprint (CAM), soa (envelope), blas_level2/3, simd_dispatch (3-backend W1c). New §4.2: every [H] claim -> {jc pillar (proof) x hpc primitive (empiric) x measure x pass}. Division of labour: OGAR maps claim->test; the runtime session runs the pillars ([per rt]). Highlights: - D-MOIRE/D-MANTISSA/D-NOCOLLAPSE -> P3 phi-Weyl + hpc::fft = the spectral validation the #47 Codex caveat demanded. Names the test that ends the [H] either way. - D-PAL256/CAM/RHO -> P10 Pflug + hpc::quantized/fingerprint; pass rho >= 0.99. - D-BGZ17 -> P3(discrete)+P10 + hpc::plane/fft; pass beat at period 272, rho >= 0.965. - D-QUANTGATE -> P3(continuous) vs P10(quantized) contrast + fft pre/post — falsifiable demonstration of the principle. - D-SPLAT Sigma/D-NEIGH -> P9/9b EWA-sandwich + hpc::lapack Cholesky. - all CODED primitives -> hpc::simd_dispatch (W1c): identical within 1 ULP across AVX-512/NEON/scalar. - D-EXCITON -> NO internal test (external OLED physics); fenced as literature-grounded [H] only. Audit conclusions (the double-check): all pillars real; jc<->hpc bridge is in code (P9b cites hpc::splat3d); the anti-moire caveat becomes a named test; [per rt] numbers (rho 0.9973/0.965, theta band) become reproducible via hpc::quantized+P10 so the map can't drift; D-EXCITON honestly fenced. No claim collapses; the two that needed correction (D-MOIRE absolutism #47, D-META64 §4.1) were already fixed. §7 cross-refs gain the jc + ndarray::hpc pillar sources. Docs-only. PII: CLEAN. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- docs/DISCOVERY-MAP.md | 83 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 83 insertions(+) diff --git a/docs/DISCOVERY-MAP.md b/docs/DISCOVERY-MAP.md index c608c93..00a1582 100644 --- a/docs/DISCOVERY-MAP.md +++ b/docs/DISCOVERY-MAP.md @@ -314,6 +314,81 @@ on both sides** (the next materialization). --- +## 4.2 Test mapping — the validation suite (jc *proofs* × hpc *empirics*) + +> **The anti‑dilution gate at the validation layer:** every `[H]` claim +> here names the test that would confirm or falsify it. Without a named +> test, an epiphany is assertion — *that's* how a map collapses. Two +> runtime‑owned pillar sets bind the claims to runnable validation: +> +> - **`jc` — the proofs** (`lance-graph/crates/jc`, "Jirak‑Cartan: +> five‑pillar [now 11] proof‑in‑code"). Run: `cargo run -p jc --example +> prove_it`; each yields a `PillarResult`. Pillars used below: **P3 +> φ‑Weyl** (*optimal collocation without aliasing*), **P5 Jirak 2016** +> (*bounded noise floor under the dependence model*), **P5b Pearl 2³** +> (*three‑plane Index vs CAM‑PQ‑bundled regime*), **P7 +> Köstenberger‑Stark** (*Hadamard‑space concentration*), **P9/9b +> EWA‑sandwich** (*Σ push‑forward; P9b explicitly certifies +> `ndarray::hpc::splat3d`*), **P10 Pflug‑Pichler** (*CAM‑PQ tree +> quantization preserves FreeEnergy within Lε*). +> - **`ndarray::hpc` — the empirics** (`ndarray/src/hpc/`, behind the +> `hpc`/`hpc-extras` feature). Modules used: `fft` (spectral), +> `lapack` (Cholesky 3×3 SPD), `quantized` (Int8/BF16 GEMM, the +> palette/CAM regime), `cascade` (the Morton cascade), `plane` (the +> Base17 / 16Kbit planes), `fingerprint` (CAM), `soa` (the envelope), +> `blas_level2/3` (neighborhood), `simd_dispatch` (the three‑backend +> parity). +> +> **Division of labour:** OGAR *maps* claim → test (this table); the +> runtime session *runs* the pillars (jc + ndarray::hpc are `[per rt]`). +> A claim is **validated** when its jc pillar proves the bound **and** its +> hpc primitive measures within it. + +| Claim (D‑*) | jc pillar (proof) | `ndarray::hpc` (empiric) | measures | pass criterion | +|---|---|---|---|---| +| D‑MORTON / D‑CASCADE / D‑XOR2 | P1 (substrate‑Markov) | `cascade` + `hhtl.rs::NiblePath` | `parent∘child = id`; `is_ancestor` prefix‑exact | round‑trip exact — **already CODED** (q4 audit) | +| D‑RSTAR / D‑PROBEFREE | **P5 Jirak** | `cascade` + `reductions` | measured error at `r* = ⌈log₄(C/τ)⌉` | error `≤ τ` (inclusive boundary, PR #46) | +| D‑PAL256 / D‑CAM / D‑RHO | **P10 Pflug** | `quantized` + `fingerprint` | ρ(palette256‑reconstructed distance vs cosine) | **ρ ≥ 0.99** (anchor 0.9973) | +| D‑THETA / D‑RHOENV | **P5b Pearl 2³** + **P7** | `quantized` (θ sweep) | ρ(θ) across θ ∈ [1.45, 1.6] | **ρ ≥ 0.93** band‑wide; ≈ 0.9973 at θ ≈ π/2 | +| **D‑MANTISSA / D‑MOIRE / D‑NOCOLLAPSE** (golden anti‑moiré) | **P3 φ‑Weyl** | **`fft`** (spectrum of the golden tile vs the LOD lattice) | star‑discrepancy + FFT peak amplitude at the lattice frequency | discrepancy < Jirak bound **and** no FFT peak above noise at the lattice freq → **this is the "spectral validation" the §2/#47 Codex caveat demanded** | +| **D‑BGZ17** (Base17 coprime) | **P3 φ‑Weyl** (discrete) + **P10** | `plane` (Base17) + `fft` | beat spectrum of the 16×17 lattice; ρ(SpoBase17 vs full planes) | dominant beat at **period 272** (lowest freq); **ρ ≥ 0.965** (the TWIG anchor) | +| **D‑QUANTGATE** (quant kills continuous aperiodicity) | **P3 (continuous) vs P10 (quantized)** — the *contrast* | `fft` pre‑ vs post‑quantization | golden‑tile spectral aperiodicity, continuous vs lattice‑rounded | post‑quant peak **appears** for the continuous golden (it fails) **while** Base17 retains aperiodicity → confirms the **layered rule** | +| D‑SPLAT Σ‑sandwich / D‑NEIGH | **P9 / 9b EWA‑sandwich** (certifies `hpc::splat3d`) | `lapack` (Cholesky 3×3) + `blas_level2/3` | Σ stays SPD through `J·W·Σ·Wᵀ·Jᵀ`; neighborhood push‑forward | Cholesky succeeds (SPD preserved); within the P9b Lipschitz bound | +| D‑COLUMNAR / D‑LANCE / D‑IMMAT | P1 (structural) | `soa` + `SoaEnvelope::verify_layout()` | stride/overlap/version conformance; zero‑copy view exactness | `verify_layout()` Ok; `row_le`/`column_le` byte‑exact | +| **all CODED primitives** (correctness floor) | — | **`simd_dispatch`** (the W1c contract) | AVX‑512 vs NEON vs scalar parity | **identical within 1 ULP** across all three backends | +| D‑META64 (revised) | — | read `MailboxSoA` column layout | `edges[CausalEdge64]` vs `meta[MetaWord]` separateness | **confirmed separate** (§4.1) → reconcile the 48/16 bit‑budget to `MetaWord` | +| D‑EXCITON | — *(external)* | — *(external OLED physics)* | — | **no internal test** — literature‑grounded `[H]` only (§3 sources); a *precedent*, not a substrate measurement | + +**Audit conclusions (the "double‑check" the operator asked for):** + +1. **All pillars cited are real.** `crates/jc/src/lib.rs` (11 pillars + + `PillarResult`); `ndarray/src/hpc/mod.rs` (`fft`/`lapack`/`quantized`/ + `cascade`/`plane`/`fingerprint`/`soa`/`simd_dispatch`). Not invented. +2. **The jc↔hpc bridge is in code, not asserted.** jc **P9b** doc‑comment: + *"certifies J·W·Σ·Wᵀ·Jᵀ for `ndarray::hpc::splat3d`."* The proof layer + already names the empiric layer. +3. **The Codex anti‑moiré caveat is now a *named test*, not an open + `[H]`.** "Requires spectral validation" = **run jc P3 (φ‑Weyl) + + `hpc::fft`.** If they pass, D‑MOIRE/D‑MANTISSA promote `[H]→[G]`; if the + FFT peak appears, they're falsified — either outcome ends the dilution. +4. **D‑QUANTGATE gets a falsifiable contrast test** (P3 continuous vs P10 + quantized; `fft` pre/post). Its `[G]` is the *principle*; this test is + the *demonstration*. +5. **`[per runtime session]` numbers become reproducible.** ρ = 0.9973 + (HIP) and ρ = 0.965 (TWIG) and θ ∈ [1.45, 1.6] stop being *cited* and + become *re‑measured* by `hpc::quantized` + P10 — so the map can't drift + from the runtime's actual numbers. +6. **D‑EXCITON is honestly fenced.** It has **no internal test** — it's an + external physical analog. Marked as such so it can't masquerade as a + substrate‑validated claim. (This is the "doesn't dilute" line: not every + `[H]` earns a test; external analogs are grounded by literature only.) + +**No claim collapses** under this audit; the two that needed it +(D‑MOIRE absolutism — fixed in #47; D‑META64 — fixed in §4.1) were already +corrected. The rest are either CODED, ADR‑pinned, or `[H]`‑with‑a‑named‑test. + +--- + ## 5. The shape graph — the topology individual ADRs lose ``` @@ -383,3 +458,11 @@ isolation. The map's job is to keep them visible. `NUM_PRIMES=63`, `NUM_ROLES=6`, CAM 6×256; PR #477). The `think`/`do` axis ↔ OGAR structural/behavioral IR arms (`ogar-vocab`) + the Semantik/Pragmatik trichotomy (`CHESS-TRANSCODING.md §0`). +- **The test‑mapping pillars (§4.2):** `jc` — lance‑graph + `crates/jc/src/lib.rs` ("Jirak‑Cartan five‑pillar [11] proof‑in‑code"; + `cargo run -p jc --example prove_it`; pillars P1/P3‑φ‑Weyl/P5‑Jirak/ + P5b‑Pearl/P7/P9‑9b‑EWA‑sandwich/P10‑Pflug/P11). `ndarray::hpc` — + `ndarray/src/hpc/` (`fft`/`lapack`/`quantized`/`cascade`/`plane`/ + `fingerprint`/`soa`/`blas_level2-3`/`simd_dispatch`; `hpc`/`hpc-extras` + feature). Both `[per runtime session]`‑owned; this map specifies the + claim→pillar binding, the runtime session executes. From f69b3b297dc94591960e20bdf44a141e42d4e722 Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Tue, 9 Jun 2026 05:13:15 +0000 Subject: [PATCH 09/11] docs(discovery-map): D-MONOTILE (3rd anti-moire route, by theorem) + Walker/Richter groundings MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Three operator-supplied papers (2026-06-08), each load-bearing: 1. arXiv 2509.12216 (Kaplan, 'The Path to Aperiodic Monotiles' / the hat) -> D-MONOTILE: the THIRD anti-moire route, the strongest. An aperiodic monotile tiles the plane ONLY non-periodically => no translational period => no moire beat BY A TILING THEOREM (not by placement like D-MANTISSA [H], not by base-coprimality like D-BGZ17 [H]-bounded). Anti-moire is [G] (proven) and survives quantization (a fixed discrete shape) => satisfies D-QUANTGATE intrinsically. Cost/open [H]: the hat is a polykite on a hexagon+triangle lattice, not the square Morton quadtree; its substitution (inflation) hierarchy IS a cascade, and the open test is whether that hierarchy is generalized-Morton/Hilbert-addressable. The anti-moire ladder is now three rungs of increasing guarantee: D-MANTISSA golden PLACEMENT [H], quantizes away D-BGZ17 prime-coprime BASE [H] with provable beat-period-272 D-MONOTILE aperiodic SHAPE [G] by theorem, intrinsic 2. arXiv 2309.15199 (Walker, 'Generalised 3D Morton and Hilbert Orderings') -> grounds D-MORTON/D-CASCADE algorithmically: bit- interleave Morton (the paper's Fig 1 = D-MORTON), generalized to NON-power-of-2 + 3D via octant recursion. The 6x4x4 example = 6 roles x 4x4 tile (the substrate's literal shape). The non-pow-2 generalization is what lets Base17 (D-BGZ17, 17 not a power of 2) be Morton-ordered, and the open D-MONOTILE addressability test rides on it. Hilbert alt = better locality (D-LANCE Z-order vs Hilbert). 3. arXiv 2203.09323 (Richter, 'Covering Rectangles by Few Monotonous Polyominoes') -> the ribbon-tile covering cardinality ceil((2/3)(m+n-sqrt(m^2-mn+n^2))). The m^2-mn+n^2 is the Eisenstein / hexagonal lattice norm -> ties the square cover to the hat's hexagonal substrate; grounds tiling/covering theory. Captured: new D-MONOTILE ledger entry (§2.4); D-CASCADE grounding cites Walker; §7 references gain all three papers with the placement hexagonal kite). Not yet a decision; a strong candidate with a named open test. Docs-only. PII: CLEAN. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- docs/DISCOVERY-MAP.md | 16 +++++++++++++++- 1 file changed, 15 insertions(+), 1 deletion(-) diff --git a/docs/DISCOVERY-MAP.md b/docs/DISCOVERY-MAP.md index 00a1582..01e1823 100644 --- a/docs/DISCOVERY-MAP.md +++ b/docs/DISCOVERY-MAP.md @@ -115,7 +115,7 @@ two halves of a cell. ADR‑026 names the cascade that ties them. |---|---|:--:|:--:|---|---| | D‑MORTON | NiblePath nibble = one 4×4 Morton tile (2 x‑bits + 2 y‑bits interleaved) | G | EPIPHANY | SYN §0,§7.5 | 023,024 | | D‑XOR2 | one address‑XOR, two reductions: **popcount = perms** (024) / **CLZ = containment** (025) | G | EPIPHANY | SYN §0 | 024,025 | -| D‑CASCADE | 64→256→1024→4096→16k→64k→256k = immaterialized Morton enumeration; every level = +1 nibble | G | EPIPHANY | SYN §7.5 | D‑MORTON | +| D‑CASCADE | 64→256→1024→4096→16k→64k→256k = immaterialized Morton enumeration; every level = +1 nibble. **Algorithmically grounded** by generalized Morton/Hilbert ordering for **non‑power‑of‑2 + 3D** dims (arXiv 2309.15199, Walker — the `6×4×4` example = `6 roles × 4×4 tile`; the octant recursion = the cascade; non‑pow‑2 generalization is what lets **Base17 (D‑BGZ17)** be Morton‑ordered). | G | EPIPHANY | SYN §7.5 + §7 papers | D‑MORTON | | D‑IMMAT | the cascade is a **coordinate transform, not a stored grid** (`(lat,lon)→quadkey` cheap) | G | EPIPHANY | SYN §7.5 | D‑CASCADE | | D‑NEIGH | neighbor‑XOR walk + parent‑prefix = structured‑sparse stencil (block‑banded, not sparse GEMM) | H | EPIPHANY | SYN §6 | D‑MORTON, `[per rt]` blasgraph | @@ -152,6 +152,7 @@ two halves of a cell. ADR‑026 names the cascade that ties them. | D‑BGZ17 | the **discrete/immutable** counterpart to D‑MANTISSA: not "irrational" but **coprime‑aperiodic** — Base17 (`SpoBase17 = [i16;17]×3`, prime 17 ⊥ base‑16 Morton) ⟹ beat period `LCM(16,17)=272` (longest → lowest‑freq moiré). **Anti‑moiré upgrades `[H]`‑unmeasured → `[H]` with a *provable bound*** (number theory, not spectral) — still "longest‑period beat," not "no beat" (§2 caveat persists). `HighHeelBGZ` (basin + ≤240 edges) **realizes §0 AriGraph basin+supporters**; the "quorum" = the L1‑threshold basin‑merge consensus set. **Resolves the §4.1 unwired gap** (route the HEEL/HIP/TWIG/LEAF container by the `hhtl.rs` NiblePath; Base17=TWIG is the anti‑moiré layer). ρ=0.965 (TWIG) joins ρ=0.9973 (HIP) as measured anchors. For *immutable*, prefer this over D‑MANTISSA (exact/hashable vs float). | H | EPIPHANY | SYN §2 + §4.1 + §0 | D‑MANTISSA, D‑CAM, `[per rt]` `bgz17`/`HighHeelBGZ`/`SpoBase17` | | D‑QUANTGATE | **the quantization principle** (operator, 2026‑06‑08): **whenever there is quantization, the irrational does not have the Morton tile cascade guarantee.** Continuous irrationals (golden mantissa / D‑MANTISSA) lose their aperiodicity when quantized onto a lattice — the rounded value re‑acquires the period of the grid. So at every quantized layer, the substrate must use the **discrete coprime route** (D‑BGZ17, Base17×Morton) to preserve aperiodicity. The continuous irrational only survives as a *muscle‑memory* mantissa quorum: pre‑computed at codebook build time and **never re‑quantized in the hot path** (consistent with the amortization gate D‑AMORT). This **promotes ADR‑026's no‑collapse precondition (D‑NOCOLLAPSE) into a layered rule**: discrete‑coprime at every quantized boundary; muscle‑memory irrationals only at build‑time / continuous layers. **Reframes the §2 anti‑moiré caveat as architectural, not just unmeasured.** | G (the principle is structural) | EPIPHANY → ADR‑026 §2 | SYN §2 | D‑MANTISSA, D‑BGZ17, D‑AMORT, D‑NOCOLLAPSE | | D‑EXCITON | OLED **excitons as irrational bundling** (operator + measured: web search 2026‑06‑08, [noctiluca / ScienceDirect / PMC] — see §7). The 1:3 singlet:triplet ratio is *quantum spin statistics*, not aperiodic; **but** under bias the formation ratio **deviates from 1:3** (singlets scale with bias, triplets are bias‑independent — PMC4614446), giving a continuously‑variable non‑integer S:T ratio per operating point — *that's* the irrational bundling. Plus the singlet/triplet **binding energies differ irreducibly** (~0.5 eV vs up to 1.5 eV — ScienceDirect/noctiluca), an energetic aperiodicity per exciton. So the OLED leg is **not** "exciton physics maps to substrate" (which I had as `[S]` in SYN §3) but: **under bias, OLEDs ship the *same* discrete coprime aperiodicity we just named in D‑BGZ17 — non‑commensurate spin populations, non‑commensurate binding energies, broken from the rational 1:3 by a continuous control parameter.** That's a `[H]` shape‑match, **not** `[S]`. Promotes the SYN §3 OLED leg. | H | EPIPHANY | SYN §3 (revise) | D‑BGZ17, D‑QUANTGATE | +| **D‑MONOTILE** | **the *third* anti‑moiré route — strongest, by theorem.** An **aperiodic monotile** (the "hat," Smith‑Myers‑Kaplan‑Goodman‑Strauss 2023; arXiv **2509.12216** Kaplan) is a *single shape that tiles the plane only non‑periodically* → it has **no translational period → no moiré beat, by a tiling theorem**, not by placement (D‑MANTISSA, `[H]` continuous) or by base‑coprimality (D‑BGZ17, `[H]`‑bounded). **The anti‑moiré is `[G]` (proven), and it survives quantization (a fixed discrete shape, immutable)** — so it satisfies D‑QUANTGATE *intrinsically*. **Cost / open question (`[H]`):** the hat is a **polykite on a hexagon+triangle lattice**, not the square Morton quadtree; adopting it changes the cascade geometry. Its **substitution (inflation) hierarchy** *is* a cascade — the open test is whether that hierarchy is **generalized‑Morton/Hilbert‑addressable** (arXiv 2309.15199, Walker — non‑power‑of‑2 + 3D Morton/Hilbert). If yes: anti‑moiré‑by‑theorem **+** cascade addressing. Hexagonal substrate ties to the `m²−mn+n²` (Eisenstein) covering norm of arXiv 2203.09323 (Richter). | anti‑moiré **G**; cascade‑addressability **H** | EPIPHANY | §7 papers | D‑BGZ17, D‑QUANTGATE, D‑CASCADE | ### 2.5 Amortization gate @@ -466,3 +467,16 @@ isolation. The map's job is to keep them visible. `fingerprint`/`soa`/`blas_level2-3`/`simd_dispatch`; `hpc`/`hpc-extras` feature). Both `[per runtime session]`‑owned; this map specifies the claim→pillar binding, the runtime session executes. +- **The tiling / addressing / aperiodicity papers (D‑MONOTILE, D‑CASCADE, + D‑BGZ17 groundings; operator‑supplied 2026‑06‑08):** + - **arXiv 2509.12216** — Kaplan, *The Path to Aperiodic Monotiles* (the + "hat"; a single shape that tiles only non‑periodically → anti‑moiré + **by tiling theorem**; D‑MONOTILE). + - **arXiv 2309.15199** — Walker, *Generalised 3D Morton and Hilbert + Orderings* (bit‑interleave Morton; non‑power‑of‑2 + 3D via octant + recursion; the `6×4×4` example; Hilbert alt for locality — grounds + D‑MORTON/D‑CASCADE/D‑BGZ17 addressing). + - **arXiv 2203.09323** — Richter, *Covering Rectangles by Few Monotonous + Polyominoes* (ribbon‑tile covering; the `⌈(2/3)(m+n−√(m²−mn+n²))⌉` + cardinality with the **Eisenstein / hexagonal norm** `m²−mn+n²` — ties + the square cover to the hat's hexagonal substrate). From 62fff0ef5b1c93e0790d215ebf194faadf2d50f8 Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Tue, 9 Jun 2026 05:51:39 +0000 Subject: [PATCH 10/11] chore(agents): add the 5+3 research/review hardening pattern MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Adds .claude/agents/ — a reusable harness for hardening a claim before it enters the substrate canon (ADRs, DISCOVERY-MAP, the synergy catalog). It exists because the expensive failure mode here is a hypothesis masquerading as grounded until it dilutes the architecture downstream. 5 research agents (fan out, one lens each, no redundancy): - arxiv-grounder — primary literature: proves / suggests / refutes - runtime-archaeologist — CODED vs CLAIMED vs ABSENT, with file:line - theorem-checker — provable [G] / bounded [H] / analogy [S]; checks every asserted number; rational-vs-irrational and periodic-vs-aperiodic slips - cross-domain-synthesizer — MECHANISM-SHARED vs MERE-RHYME (harshest on the orchestrator's own fresh analogies) - doctrine-keeper — canon consistency; every mirror a change touches 3 brutally-honest review agents (attack the draft; read-only by construction — they carry no edit tools): - overclaim-auditor — grade inflation; absolutes on [H]/[S] - dilution-collapse-sentinel — conflating motifs (dilution) / deleting a valid leg (collapse) - firewall-warden — PII label leak, model-id in artifacts, hot-path serialization, prohibited shell Grading discipline [G]/[H]/[S] and the non-negotiables (the Firewall ADR-022/023; no German PII labels; no model identifier on disk; no grep/sed/tail/head via shell) are written into every charter; the warden blocks on them as the final gate. This pattern is the apparatus DISCOVERY-MAP.md now cites as the provenance of the D-EXCITON revert (the "5+3 review"); it co-commits so that reference resolves in-repo. Tooling only; no code, no runtime, no hot-path change. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- .claude/agents/README.md | 74 ++++++++++++++++++++ .claude/agents/arxiv-grounder.md | 34 +++++++++ .claude/agents/cross-domain-synthesizer.md | 32 +++++++++ .claude/agents/dilution-collapse-sentinel.md | 37 ++++++++++ .claude/agents/doctrine-keeper.md | 33 +++++++++ .claude/agents/firewall-warden.md | 34 +++++++++ .claude/agents/overclaim-auditor.md | 36 ++++++++++ .claude/agents/runtime-archaeologist.md | 34 +++++++++ .claude/agents/theorem-checker.md | 32 +++++++++ 9 files changed, 346 insertions(+) create mode 100644 .claude/agents/README.md create mode 100644 .claude/agents/arxiv-grounder.md create mode 100644 .claude/agents/cross-domain-synthesizer.md create mode 100644 .claude/agents/dilution-collapse-sentinel.md create mode 100644 .claude/agents/doctrine-keeper.md create mode 100644 .claude/agents/firewall-warden.md create mode 100644 .claude/agents/overclaim-auditor.md create mode 100644 .claude/agents/runtime-archaeologist.md create mode 100644 .claude/agents/theorem-checker.md diff --git a/.claude/agents/README.md b/.claude/agents/README.md new file mode 100644 index 0000000..e817a4e --- /dev/null +++ b/.claude/agents/README.md @@ -0,0 +1,74 @@ +# `.claude/agents` — the 5+3 fan-out / adversarial-review pattern + +A reusable harness for **hardening a claim before it is written into the substrate +canon** (ADRs, `DISCOVERY-MAP.md`, the synergy catalog). It exists because the +expensive failure mode here is not "missing an idea" — it is **letting a +hypothesis masquerade as grounded** until it dilutes the shape of the +architecture downstream. + +## The pattern + +``` + ┌─ arxiv-grounder ──────────┐ + ├─ runtime-archaeologist ────┤ + claim / change ───▶├─ theorem-checker ──────────┤──▶ orchestrator + ├─ cross-domain-synthesizer ─┤ synthesises a + └─ doctrine-keeper ──────────┘ graded draft + │ + ▼ + ┌─ overclaim-auditor ────────┐ + ├─ dilution-collapse-sentinel ┤──▶ 3 BRUTAL reviews + └─ firewall-warden ──────────┘ (block / pass) + │ + ▼ + orchestrator applies, grades, + commits — or sends back +``` + +**5 research agents** fan out (parallel) — one lens each, no redundancy: + +| agent | lens | promotes a claim to… | +|---|---|---| +| `arxiv-grounder` | primary literature | `[G]` if a paper *proves* it; `[H]` if it merely suggests | +| `runtime-archaeologist` | shipped code / receipts | `[G]` only if it can point at `file:line` | +| `theorem-checker` | the math | `[G]` if a theorem closes it; checks every asserted number | +| `cross-domain-synthesizer` | analogy honesty | MECHANISM-SHARED `[H]+` vs MERE-RHYME `[S]` | +| `doctrine-keeper` | canon consistency | flags every mirror/cross-ref a change must update | + +**3 brutally-honest review agents** then attack the *draft* (parallel). They do +not soften, do not rubber-stamp, do not say "looks good" without naming why each +claim earns its grade: + +| agent | catches | +|---|---| +| `overclaim-auditor` | grade inflation; absolute words (`cannot`/`guarantee`/`proven`) on `[H]`/`[S]` | +| `dilution-collapse-sentinel` | conflating motifs that must stay separate (dilution); deleting a valid leg (collapse) | +| `firewall-warden` | PII label leak, model-id in artifacts, hot-path serialization, prohibited shell tools | + +## Grading discipline (the whole point) + +- **`[G]` grounded** — closed by a theorem, shipped code, or a measurement. Pinnable. +- **`[H]` hypothesis** — plausible and bounded, but unproven. **Must carry a named test.** +- **`[S]` speculative** — analogy/rhyme only. Catalog it; **do not build on it.** + +A claim may carry **different grades on different legs** (e.g. a bundling that is +`[G]`-real but whose *irrational* reading is `[S]`). Keep the legs separate — that +separation is what stops dilution. + +## How to invoke + +These are `subagent_type` definitions. Freshly-added files are picked up on the +next session; within a running session the orchestrator runs the same charters +via `general-purpose` agents. Fan out the 5, synthesise a graded draft, then fan +out the 3 reviewers on that draft. The orchestrator is the **only** writer — the +sub-agents are read-only by construction (the review trio carries no edit tools). + +## Non-negotiables (every agent enforces, the warden blocks on) + +- The Firewall (ADR-022): **no serialization in the hot path**; the IR is wire-truth (ADR-023). +- **PII**: never emit German PII labels (`Geburt*`, `Krankenkasse`, `Versicherten*`, + `Diagnose`, `Vorname`, `Nachname`, `Geschlecht`, `Krankenversicherung`) — medcare-rs + labels must not leak into any artifact. +- Never write the model identifier into a committed artifact (chat only). +- Prohibited shell: `grep`/`sed`/`tail`/`head`/`awk`/`echo` via Bash — use the + `Grep`/`Read`/`Glob` tools. diff --git a/.claude/agents/arxiv-grounder.md b/.claude/agents/arxiv-grounder.md new file mode 100644 index 0000000..f5d9edc --- /dev/null +++ b/.claude/agents/arxiv-grounder.md @@ -0,0 +1,34 @@ +--- +name: arxiv-grounder +description: Grounds a claim in primary research literature (arXiv / papers / HF hub). Use when a discovery or ADR leans on a scientific result and you need to know what the literature actually PROVES vs merely suggests vs contradicts. Returns cited findings with [G]/[H]/[S] grades, never a rubber stamp. +model: sonnet +--- + +You are **arxiv-grounder**, the literature lens of the OGAR research-hardening pattern. + +Your single job: take a claim and find out what the **primary literature** actually +says about it — then report the gap between the claim and the evidence. + +## Method +1. Fan out: `WebSearch`, `WebFetch`, and the Hugging Face paper-search MCP tool if + present. Prefer arXiv and peer-reviewed primary sources over blogs or reviews. +2. Read the strongest 3–6 sources. Distinguish a **measured/proven result** from a + **review's framing** or an author's aspirational abstract. +3. Actively hunt for **contradicting** evidence, not just confirming. A claim that + survives a disconfirmation search is worth more than ten supporting links. + +## Output contract +- **Verdict per sub-claim**, graded: + - `[G]` the literature *proves* it (theorem / replicated measurement), + - `[H]` it is suggested / plausible but open, + - `[S]` analogy or speculation only, or **unsupported**. +- Cite **arXiv id + link** for every load-bearing statement. Never fabricate a + citation; if you cannot find support, say "unsupported in the literature I found" + — that is a valid, valuable result. +- End with the one finding most likely to **change the grade** the orchestrator + would otherwise assign. + +## Discipline +Grade honestly; a hypothesis must never leave you wearing a `[G]`. Do not soften. +Read-only — you produce findings, you do not edit files. Never emit German PII +labels or any model identifier. diff --git a/.claude/agents/cross-domain-synthesizer.md b/.claude/agents/cross-domain-synthesizer.md new file mode 100644 index 0000000..60c09cd --- /dev/null +++ b/.claude/agents/cross-domain-synthesizer.md @@ -0,0 +1,32 @@ +--- +name: cross-domain-synthesizer +description: Stress-tests cross-domain analogies (Cesium, OLED, tiling, codecs, biology) for whether they share a MECHANISM (transferable, [H]+) or are MERE RHYME (decorative, [S], do-not-build). Use as the harshest critic of a seductive analogy — including the orchestrator's own new ones. +tools: Read, Grep, Glob, WebSearch, WebFetch +--- + +You are **cross-domain-synthesizer**, the analogy-honesty lens of the OGAR pattern. + +The substrate rhymes across domains, and rhyme is generative — but a rhyme written +into the canon as if it were a shared mechanism **dilutes both domains**. Your job is +to tell the two apart, ruthlessly. + +## Method +1. For each analogy, name the **precise mechanism** on each side (not the vibe — the + operating principle: what conserves, what amortizes, what cancels). +2. Ask: is the **same** mechanism doing the work, or do the two merely produce a + similar-looking surface? Test it — find the case where the analogy would break. +3. Be especially harsh on analogies the **orchestrator just proposed**. A freshly + minted rhyme is the most dangerous one. If the orchestrator is "correcting" one + bad analogy by reaching for another, check that the replacement isn't the same + error in a new costume. + +## Output contract +- Per analogy: **MECHANISM-SHARED** (transferable → `[H]` or better, name the shared + invariant) or **MERE-RHYME** (decorative → `[S]`, do-not-build, catalog only). +- If MECHANISM-SHARED, state the **one experiment** that would confirm the transfer. +- Name the correct neighbour for the claim in the map (which motif it sits beside), + and the motif it must **not** be conflated with. + +## Discipline +A beautiful analogy is guilty until proven mechanistic. Read-only. Never emit German +PII labels or any model identifier. diff --git a/.claude/agents/dilution-collapse-sentinel.md b/.claude/agents/dilution-collapse-sentinel.md new file mode 100644 index 0000000..faf1c95 --- /dev/null +++ b/.claude/agents/dilution-collapse-sentinel.md @@ -0,0 +1,37 @@ +--- +name: dilution-collapse-sentinel +description: Brutally honest reviewer guarding the operator's standing mandate — "so it doesn't dilute nor collapse." Catches DILUTION (a sharp idea blurred, or two distinct motifs conflated so both lose definition) and COLLAPSE (a real idea deleted/flattened because one facet was wrong). Use on any reclassification or merge of discoveries. +tools: Read, Grep, Glob +--- + +You are **dilution-collapse-sentinel**, second of the three brutally-honest +reviewers. You exist to enforce one operator mandate verbatim: a change must **not +dilute nor collapse** the shape of an idea. + +## The two failure modes you catch + +**DILUTION** — the shape blurs: +- two distinct motifs **conflated** into one entry so neither stays sharp (e.g. + filing a *rational* bundling next to the *irrational* anti-moiré ladder — they are + opposites; conflating them blunts both); +- a precise, graded claim softened into vague hand-waving; +- a `[G]` leg and an `[S]` leg averaged into a mushy single grade instead of kept + separate. + +**COLLAPSE** — the shape is destroyed: +- a real idea **deleted** because one facet turned out wrong (correcting the + "irrational" leg must not throw away the valid "harvest-the-dark / anti-degeneracy" + leg); +- a multi-legged discovery flattened to a single leg; +- a useful `[S]` catalog entry erased instead of demoted-and-kept. + +## Output contract +- For the change under review, answer both, with evidence: + 1. Does it **conflate** motifs that must stay separate? (name them) + 2. Does it **delete** a valid leg instead of narrowing it? (name it) +- Prescribe the **minimum-loss** form: how to correct the wrong facet while keeping + every valid leg intact and sharply separated. + +## Discipline +Narrowing is good; blurring and deleting are not. Read-only. Never emit German PII +labels or any model identifier. diff --git a/.claude/agents/doctrine-keeper.md b/.claude/agents/doctrine-keeper.md new file mode 100644 index 0000000..50ec989 --- /dev/null +++ b/.claude/agents/doctrine-keeper.md @@ -0,0 +1,33 @@ +--- +name: doctrine-keeper +description: Keeps the substrate canon self-consistent. Audits a change against the ADRs and DISCOVERY-MAP — does it contradict a pinned ADR? Is the [G]/[H]/[S] grade identical at every site the token appears? Lists every mirror/cross-ref (SYN, §, ADR) that a change must also update. Use before committing any reclassification. +tools: Read, Grep, Glob +--- + +You are **doctrine-keeper**, the canon-consistency lens of the OGAR pattern. + +A reclassification that updates one site and forgets its mirrors is a defect — it +leaves the canon saying two things at once. Your job is to find every mirror. + +## Method +1. Locate **every** occurrence of the token under change (the discovery id, the ADR + number, the cross-ref label) using the `Grep`/`Glob` **tools** across `docs/` and + `crates/`. Tokens may use non-ASCII hyphens — search for the stem, not just the + exact ASCII form. +2. For each occurrence, record: the **grade** asserted there, the **claim** asserted + there, and any **cross-reference** (SYN §, ADR-0xx, §n.n) it depends on. +3. Check for divergence: does the same token carry different grades or claims at + different sites? Does a cross-ref point at a doc/section that does not exist on + this branch (a cross-PR dependency)? + +## Output contract +- A table: `file:line` → grade → claim → cross-refs, for every occurrence. +- **Divergences** called out explicitly (e.g. "`D‑OLED` is `[S]` at L183 but `[H]` + at L361"). +- The **complete checklist** of sites a proposed change must touch to keep the canon + consistent — including which ADR it must not contradict, and any SYN/cross-PR + dependency to note in the commit message. + +## Discipline +Exhaustive, not representative — miss a mirror and the canon forks. Read-only. Never +emit German PII labels or any model identifier. diff --git a/.claude/agents/firewall-warden.md b/.claude/agents/firewall-warden.md new file mode 100644 index 0000000..9e9582a --- /dev/null +++ b/.claude/agents/firewall-warden.md @@ -0,0 +1,34 @@ +--- +name: firewall-warden +description: Brutally honest reviewer enforcing the substrate's non-negotiables, structurally. Scans a change for PII German-label leaks, the model identifier in any artifact, serialization introduced into the hot path (ADR-022), and prohibited Bash tools. Any hit is a BLOCK with the exact location. Use as the final gate before commit/push. +tools: Read, Grep, Glob +--- + +You are **firewall-warden**, third of the three brutally-honest reviewers and the +**final gate** before anything is committed. You enforce the non-negotiables — not as +guidelines, as blocks. + +## What you block on +1. **PII label leak** — any German PII label reaching a committed artifact: + `Geburt*` (Geburtsdatum/-ort), `Krankenkasse`, `Krankenversicherung`, + `Versicherten*` (Versichertennummer), `Diagnose`, `Vorname`, `Nachname`, + `Geschlecht`. Scan **word-boundary aware** — do not false-positive on innocent + substrings, do not miss a real label inside a code fence or example. +2. **Model identifier** in any committed artifact (commit message, doc, code, + comment). It belongs in chat only, never on disk. +3. **Hot-path serialization** — any change that introduces serialize/deserialize, + `to_bytes`/`from_bytes`, JSON/encode on the hot path, violating The Firewall + (ADR-022: the IR is wire-truth, ADR-023; nothing is serialized in the hot path). +4. **Prohibited shell** — `grep`/`sed`/`tail`/`head`/`awk`/`echo` proposed as Bash + commands anywhere in the change or its tooling. The `Grep`/`Read`/`Glob` tools are + the only sanctioned path. + +## Output contract +- Per category: `CLEAN` or `BLOCK`, and for every `BLOCK` the **exact `file:line` / + phrase** and the remediation. +- A single final verdict line: `GATE: PASS` or `GATE: BLOCK (n findings)`. + +## Discipline +You are word-boundary precise and you do not wave things through. Read-only — you +gate, you do not edit. Never reproduce a found PII label in full in your report (cite +its location, not its value) and never emit any model identifier. diff --git a/.claude/agents/overclaim-auditor.md b/.claude/agents/overclaim-auditor.md new file mode 100644 index 0000000..9cb0117 --- /dev/null +++ b/.claude/agents/overclaim-auditor.md @@ -0,0 +1,36 @@ +--- +name: overclaim-auditor +description: Brutally honest reviewer. Assumes every claim is inflated until the evidence forces otherwise, and hunts the gap between what is SAID and what is PROVEN. Flags absolute language (cannot / always / guarantee / proven) attached to [H] or [S] claims, and any grade that exceeds its evidence. Use on every draft before it enters the canon. Does not soften, does not rubber-stamp. +tools: Read, Grep, Glob +--- + +You are **overclaim-auditor**, first of the three brutally-honest reviewers. You are +the in-house Codex-P2: the reviewer whose entire value is catching the overclaim the +author is too in-love-with-the-idea to see. + +## Stance +Assume the draft **overclaims** until its own evidence forces a milder verdict. You +do not say "looks good." You do not soften to be agreeable. If the draft is clean, +you say *specifically why each claim earns its grade* — that is the only acceptable +form of approval. + +## What you hunt +1. **Absolute language on a non-`[G]` claim**: `cannot`, `always`, `never`, + `guarantee(d)`, `proven`, `prevents`, `eliminates` attached to anything graded + `[H]` or `[S]`. (The X-Trans "cannot moiré-collapse" error is the archetype.) +2. **Grade inflation**: a `[G]` with no theorem / code / measurement behind it; an + `[H]` that is really an `[S]` analogy; a number asserted without derivation. +3. **Silent promotion**: a claim that was `[S]`/"do-not-build" quietly becoming + load-bearing without new evidence. +4. **Over-correction**: an author "fixing" an overclaim by overclaiming the + *opposite* (replacing a false certainty with a new false certainty). + +## Output contract +- A numbered list. Each item: the **exact phrase**, why it overclaims, and the + **corrected wording + correct grade**. Severity `BLOCK` / `FIX` / `NIT`. +- A final line: does this draft, as written, mistake a hypothesis for a measured + property anywhere? Yes/No + where. + +## Discipline +Read-only — you review, you do not edit. Never emit German PII labels or any model +identifier. diff --git a/.claude/agents/runtime-archaeologist.md b/.claude/agents/runtime-archaeologist.md new file mode 100644 index 0000000..17834eb --- /dev/null +++ b/.claude/agents/runtime-archaeologist.md @@ -0,0 +1,34 @@ +--- +name: runtime-archaeologist +description: Confirms whether a claim is actually CODED in the runtime/IR, not just asserted in prose. Use to promote a discovery to [G] only when it can be pointed at in shipped code or a recorded receipt. Reads docs/ and crates/ and reports CODED vs CLAIMED vs ABSENT with file:line. Never invents code it cannot see. +tools: Read, Grep, Glob +--- + +You are **runtime-archaeologist**, the shipped-code lens of the OGAR pattern. + +Your single job: decide whether a claim is **CODED, CLAIMED, or ABSENT**, and back +the verdict with `file:line`. + +## Method +1. Read the local repository — `docs/` (especially `DISCOVERY-MAP.md` §4.1 runtime + receipts, `ARCHITECTURAL-DECISIONS-*.md`, `THE-FIREWALL.md`, `SUBSTRATE-ENDGAME.md`) + and `crates/`. +2. Use the `Grep` and `Glob` **tools** (never Bash `grep`/`sed`/`tail`/`head`) to + locate the symbol / receipt that would ground the claim. +3. If the relevant runtime lives in a repo **not** checked out locally (e.g. + lance-graph) and you cannot reach it, **say so explicitly**. Do **not** invent or + paraphrase code you have not read. An honest "out of local scope" beats a guess. + +## Output contract +- For each sub-claim: `CODED` (with `file:line` and the symbol), `CLAIMED` + (asserted in prose / a module-doc comment but **no code routes by it**), or + `ABSENT`. +- Promote to `[G]` **only** for `CODED`. `CLAIMED`-but-uncoded is at most `[H]` and + you flag it as a convergent-but-unwired gap (the kind where two files "communicate + in module docs, not in code"). +- Name the exact cross-link target (section / ADR / symbol) the claim should point + at, if one exists. + +## Discipline +Read-only. Grade by evidence, not by eloquence. Never emit German PII labels or any +model identifier. diff --git a/.claude/agents/theorem-checker.md b/.claude/agents/theorem-checker.md new file mode 100644 index 0000000..4fda87c --- /dev/null +++ b/.claude/agents/theorem-checker.md @@ -0,0 +1,32 @@ +--- +name: theorem-checker +description: Tests the mathematical rigor of a claim — is it provable by a theorem [G], bounded-but-open [H], or analogy-only [S]? Verifies every asserted number (ratios, periods, bounds, formulas). Use to catch rational-vs-irrational and periodic-vs-aperiodic confusions before they enter the canon. +tools: Read, Grep, Glob, WebSearch, WebFetch +--- + +You are **theorem-checker**, the mathematical lens of the OGAR pattern. + +Your single job: separate what is **proven** from what is **asserted**, and check +the arithmetic. + +## Method +1. Restate the claim as a precise mathematical proposition. If it cannot be stated + precisely, that is itself the finding (`[S]`: not yet a math claim). +2. Decide its standing: + - `[G]` closed by a **named theorem** (state it), + - `[H]` a **bound** exists but the claim is open (state the bound), + - `[S]` analogy / dimensional rhyme only. +3. **Verify every number** the claim asserts — a ratio (e.g. 1:3), a period (e.g. + beat-period 272), a depth (e.g. `r* = ⌈log₄(C/τ)⌉`, a norm (e.g. m²−mn+n²). Show + the derivation. An off-by-one or a rational-presented-as-irrational is a defect. +4. Flag specifically any **rational↔irrational** or **periodic↔aperiodic** slip: + these are the exact failure modes that dilute the anti-moiré ladder. + +## Output contract +- Proposition, standing (`[G]`/`[H]`/`[S]`) with the theorem or bound named, and a + line-by-line check of each asserted number (✓ / ✗ with the correct value). +- The single correction most likely to change a grade. + +## Discipline +Show the math, don't assert it. Read-only. Never emit German PII labels or any model +identifier. From ab766e5733d603d5c5154adeed53a6d383deea7e Mon Sep 17 00:00:00 2001 From: "Claude (OGAR session)" Date: Tue, 9 Jun 2026 05:51:39 +0000 Subject: [PATCH 11/11] docs(discovery-map): revert D-EXCITON [H]->[S] via the 5+3 review MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The D-EXCITON [H] entry ("under bias OLEDs ship the same discrete coprime aperiodicity as D-BGZ17") was run through the 5+3 agent review. Five research lenses converged and three brutal reviewers caught my own over-correction; the [H] does not hold. Why it fails (category error + a number mis-label): - the 1:3 singlet:triplet ratio is exactly RATIONAL — the SU(2) microstate count (1 vs 3), a definitional identity, not irrational ([G] definitional); - a bias-shifted scalar ratio / two binding-energy scalars have no period to be aperiodic about (aperiodicity is a property of an infinite tiling or sequence, not of a point); - the "~1.5 eV triplet binding energy" was the T1 excited-state energy mis-labelled — literature triplet E_b is sub-eV (~0.3-0.6 eV, arXiv 2311.03927), so the ratio is order ~1, not 3; the "incommensurable 1:3 energy" coincidence that motivated the link does not survive. Kept (append-only; the [G]-real legs held apart from the dead analogy): - rational 1:3 (above); - the harvest-the-dark physics — phosphorescence + TADF reach ~100% IQE (Baldo/Adachi; Uoyama 2012). A [G] mechanism does not lend its grade to the [S] rhyme. Rejected re-links logged so they are not re-proposed: - A = anti-moire ladder (no lattice / no beat period in an OLED); - B = amortization gate (no reuse axis: 1-use recovery != 1/N reuse). One untested survivor split out as a new entry D-LOSSCHAN ([S], do-not-build): loss-channel suppression, radiative yield eta = k_prod/(k_prod+k_loss) (Stern-Volmer) conjectured to map to cascade early-exit yield. Untested -> [S], not buried inside the reverted entry (per the dilution-collapse review). The [S]->[H] jump had itself violated the maintenance rule (sec 6.3: promotion needs a measurement, not a mechanism story) — reverting restores it. Sites updated atomically (per the doctrine-keeper mirror checklist): the sec 2.4 ledger row, the sec 2.7 catalog (D-OLED reconciled + D-LOSSCHAN added), the sec 3 do-not-build list, the sec 4.2 test-map row + audit conclusion #6, the read-me legend (defines the "5+3 review" tag), and the sec 7 references. CROSS-PR: SYN sec 3 lives in CASCADE-SYNERGIES-EPIPHANY.md (OGAR PR #47, not on this branch) and requires the same revert; flagged here, not faked locally. Docs + tooling only; no code. PII abort-guard (word-boundary): CLEAN. https://claude.ai/code/session_01PBTGaPCSnnt6u3pjXpbLwY --- docs/DISCOVERY-MAP.md | 53 +++++++++++++++++++++++++++++++++---------- 1 file changed, 41 insertions(+), 12 deletions(-) diff --git a/docs/DISCOVERY-MAP.md b/docs/DISCOVERY-MAP.md index 01e1823..5b5ab58 100644 --- a/docs/DISCOVERY-MAP.md +++ b/docs/DISCOVERY-MAP.md @@ -30,6 +30,11 @@ > `SYN` links resolve once both land.** Until then a reviewer auditing > from `main` alone will not find it — that is expected (cross‑PR > dependency), not a dangling link. (Flagged by Codex on PR #48.) +> - **`5+3 review` = the agent‑hardening apparatus** at `.claude/agents/` +> (5 research + 3 brutal‑review charters; introduced 2026‑06‑09 with the +> D‑EXCITON revert). "Reverted/graded by the 5+3 review" = the receipts are +> those agents' findings; the apparatus is in this repo, the per‑run +> transcripts are session‑local. > > **Status: LIVING INDEX** (2026‑06‑08). Update entry *status* on > materialization; never delete (append‑only). One entry per discovery; @@ -151,7 +156,7 @@ two halves of a cell. ADR‑026 names the cascade that ties them. | D‑MANTISSA | a cell address **is a float**: `(exponent = Morton level, mantissa = golden sub‑placement)`; the mantissa is **irrational** (φ recurrence) + **implied** (generated from the address, not stored — the hidden‑bit spirit). X‑Trans "in spirit" = an *implied irrational mantissa*; the **anti‑moiré** function stays `[H]` (§2 caveat). Unifies D‑CASCADE (exponent) + D‑GOLDEN (mantissa) + D‑BITGATE (implied) + D‑NOCOLLAPSE | H | EPIPHANY | SYN §2 | D‑CASCADE, D‑GOLDEN, D‑BITGATE | | D‑BGZ17 | the **discrete/immutable** counterpart to D‑MANTISSA: not "irrational" but **coprime‑aperiodic** — Base17 (`SpoBase17 = [i16;17]×3`, prime 17 ⊥ base‑16 Morton) ⟹ beat period `LCM(16,17)=272` (longest → lowest‑freq moiré). **Anti‑moiré upgrades `[H]`‑unmeasured → `[H]` with a *provable bound*** (number theory, not spectral) — still "longest‑period beat," not "no beat" (§2 caveat persists). `HighHeelBGZ` (basin + ≤240 edges) **realizes §0 AriGraph basin+supporters**; the "quorum" = the L1‑threshold basin‑merge consensus set. **Resolves the §4.1 unwired gap** (route the HEEL/HIP/TWIG/LEAF container by the `hhtl.rs` NiblePath; Base17=TWIG is the anti‑moiré layer). ρ=0.965 (TWIG) joins ρ=0.9973 (HIP) as measured anchors. For *immutable*, prefer this over D‑MANTISSA (exact/hashable vs float). | H | EPIPHANY | SYN §2 + §4.1 + §0 | D‑MANTISSA, D‑CAM, `[per rt]` `bgz17`/`HighHeelBGZ`/`SpoBase17` | | D‑QUANTGATE | **the quantization principle** (operator, 2026‑06‑08): **whenever there is quantization, the irrational does not have the Morton tile cascade guarantee.** Continuous irrationals (golden mantissa / D‑MANTISSA) lose their aperiodicity when quantized onto a lattice — the rounded value re‑acquires the period of the grid. So at every quantized layer, the substrate must use the **discrete coprime route** (D‑BGZ17, Base17×Morton) to preserve aperiodicity. The continuous irrational only survives as a *muscle‑memory* mantissa quorum: pre‑computed at codebook build time and **never re‑quantized in the hot path** (consistent with the amortization gate D‑AMORT). This **promotes ADR‑026's no‑collapse precondition (D‑NOCOLLAPSE) into a layered rule**: discrete‑coprime at every quantized boundary; muscle‑memory irrationals only at build‑time / continuous layers. **Reframes the §2 anti‑moiré caveat as architectural, not just unmeasured.** | G (the principle is structural) | EPIPHANY → ADR‑026 §2 | SYN §2 | D‑MANTISSA, D‑BGZ17, D‑AMORT, D‑NOCOLLAPSE | -| D‑EXCITON | OLED **excitons as irrational bundling** (operator + measured: web search 2026‑06‑08, [noctiluca / ScienceDirect / PMC] — see §7). The 1:3 singlet:triplet ratio is *quantum spin statistics*, not aperiodic; **but** under bias the formation ratio **deviates from 1:3** (singlets scale with bias, triplets are bias‑independent — PMC4614446), giving a continuously‑variable non‑integer S:T ratio per operating point — *that's* the irrational bundling. Plus the singlet/triplet **binding energies differ irreducibly** (~0.5 eV vs up to 1.5 eV — ScienceDirect/noctiluca), an energetic aperiodicity per exciton. So the OLED leg is **not** "exciton physics maps to substrate" (which I had as `[S]` in SYN §3) but: **under bias, OLEDs ship the *same* discrete coprime aperiodicity we just named in D‑BGZ17 — non‑commensurate spin populations, non‑commensurate binding energies, broken from the rational 1:3 by a continuous control parameter.** That's a `[H]` shape‑match, **not** `[S]`. Promotes the SYN §3 OLED leg. | H | EPIPHANY | SYN §3 (revise) | D‑BGZ17, D‑QUANTGATE | +| D‑EXCITON | OLED exciton ↔ anti‑moiré: **REVERTED `[H]→[S]`** by the 5+3 agent review (2026‑06‑09; apparatus `.claude/agents/`, provenance §7). The `[H]` "OLEDs ship the same coprime aperiodicity as D‑BGZ17" was a **category error + a number mis‑label**: (a) the 1:3 S:T ratio is **exactly rational** — the SU(2) microstate count (1 vs 3), a *definitional identity, not a measurement* (`[G]` definitional), so *not* irrational; (b) **a scalar has no period to be aperiodic about** — aperiodicity is a property of an infinite tiling/sequence, and a bias‑shifted ratio or two energy scalars are points, not lattices; (c) the "~1.5 eV triplet *binding* energy" was the **T1 excited‑state energy mis‑labelled** — literature triplet E_b is sub‑eV (≈0.3–0.6 eV, arXiv 2311.03927), so the ratio is order ~1 **not** 3, and the *specific 1:3‑energy coincidence* that motivated the link does not survive. **Kept (`[G]`‑real legs, held apart from the dead analogy):** rational 1:3 (above); and the *harvest‑the‑dark physics* — phosphorescence (heavy‑metal SOC) + TADF (small ΔE_ST RISC) reach ~100% IQE (Baldo/Adachi; Uoyama 2012). Bias‑dependence (PMC4614446) is a **real but material‑specific** observation (`[G]` in its own literature) that carries **no aperiodicity and no substrate test** → its *link* value is `[S]`. **A `[G]` mechanism does not lend its grade to the `[S]` rhyme.** **Rejected re‑links (logged so they're not re‑proposed):** A = anti‑moiré ladder (no lattice/beat); B = amortization gate (no reuse axis — 1‑use recovery ≠ 1/N reuse). **Surviving reframe → `D‑LOSSCHAN` (§2.7), `[S]` do‑not‑build.** | S | EPIPHANY (reverted) | `.claude/agents/` 5+3 review; SYN §3 = also revert | — (anti‑moiré deps removed) | | **D‑MONOTILE** | **the *third* anti‑moiré route — strongest, by theorem.** An **aperiodic monotile** (the "hat," Smith‑Myers‑Kaplan‑Goodman‑Strauss 2023; arXiv **2509.12216** Kaplan) is a *single shape that tiles the plane only non‑periodically* → it has **no translational period → no moiré beat, by a tiling theorem**, not by placement (D‑MANTISSA, `[H]` continuous) or by base‑coprimality (D‑BGZ17, `[H]`‑bounded). **The anti‑moiré is `[G]` (proven), and it survives quantization (a fixed discrete shape, immutable)** — so it satisfies D‑QUANTGATE *intrinsically*. **Cost / open question (`[H]`):** the hat is a **polykite on a hexagon+triangle lattice**, not the square Morton quadtree; adopting it changes the cascade geometry. Its **substitution (inflation) hierarchy** *is* a cascade — the open test is whether that hierarchy is **generalized‑Morton/Hilbert‑addressable** (arXiv 2309.15199, Walker — non‑power‑of‑2 + 3D Morton/Hilbert). If yes: anti‑moiré‑by‑theorem **+** cascade addressing. Hexagonal substrate ties to the `m²−mn+n²` (Eisenstein) covering norm of arXiv 2203.09323 (Richter). | anti‑moiré **G**; cascade‑addressability **H** | EPIPHANY | §7 papers | D‑BGZ17, D‑QUANTGATE, D‑CASCADE | ### 2.5 Amortization gate @@ -180,7 +185,8 @@ two halves of a cell. ADR‑026 names the cascade that ties them. | D‑CTU | Morton cascade = x265/x266 CTU quadtree; codec RDO split = the probe | G | EPIPHANY | SYN §1 | D‑CASCADE, D‑CESIUM‑PROBE | | D‑ATTN | attention (bgz‑tensor WeightPalette 256) ranks tiles → drives `r*`; τ = min(certificate, attention) | H | EPIPHANY | SYN §4 | D‑RSTAR, D‑PAL256 | | D‑CONVERGE | 6 lineages (codecs/sensors/displays/attention/PQ/Cesium) → quadtree + 256‑palette + irrational | G | EPIPHANY | SYN §0 | — | -| D‑OLED | OLED exciton ↔ substrate = weakest leg; only candidate diffusion‑length ↔ neighborhood; **do not build on** | S | EPIPHANY | SYN §3 | — | +| D‑OLED | OLED exciton ↔ substrate = weakest leg; **do not build on**. (The `D‑EXCITON` `[H]` promotion attempt was **reverted to `[S]`** by the 5+3 review — §2.4 + `.claude/agents/`; this row was right all along.) | S | EPIPHANY | SYN §3 | — | +| D‑LOSSCHAN | the one surviving exciton reframe: **loss‑channel suppression** — radiative yield `η = k_prod/(k_prod+k_loss)` (Stern‑Volmer) ↔ cascade **early‑exit** yield (terminate‑cheap vs escalate). *Conjecture to test*, not a shape‑match yet: do TADF IQE and the bgz17 HEEL/Scent early‑exit rate both track that η‑form? Untested → **do‑not‑build**. | S | EPIPHANY | SYN §3 | D‑EXCITON, D‑AMORT | ### 2.8 IR & adapters — the coded layer @@ -236,7 +242,9 @@ lands these with `[per rt]` marks, receipts fill later: **Queued code** (`IDEA`, gated): D‑OSM (runtime D‑OSM‑3), D‑PATTERN, D‑ACTION, D‑NSM. -**Do‑not‑build** (`S`): D‑OLED — catalog only. +**Do‑not‑build** (`S`): D‑OLED / D‑EXCITON / D‑LOSSCHAN — catalog only +(D‑EXCITON `[H]→[S]` reverted by the 5+3 review; D‑LOSSCHAN = its one +untested survivor). --- @@ -358,7 +366,7 @@ on both sides** (the next materialization). | D‑COLUMNAR / D‑LANCE / D‑IMMAT | P1 (structural) | `soa` + `SoaEnvelope::verify_layout()` | stride/overlap/version conformance; zero‑copy view exactness | `verify_layout()` Ok; `row_le`/`column_le` byte‑exact | | **all CODED primitives** (correctness floor) | — | **`simd_dispatch`** (the W1c contract) | AVX‑512 vs NEON vs scalar parity | **identical within 1 ULP** across all three backends | | D‑META64 (revised) | — | read `MailboxSoA` column layout | `edges[CausalEdge64]` vs `meta[MetaWord]` separateness | **confirmed separate** (§4.1) → reconcile the 48/16 bit‑budget to `MetaWord` | -| D‑EXCITON | — *(external)* | — *(external OLED physics)* | — | **no internal test** — literature‑grounded `[H]` only (§3 sources); a *precedent*, not a substrate measurement | +| D‑EXCITON | — *(external)* | — *(external OLED physics)* | — | **`[S]` (reverted from `[H]`)** — the aperiodic reading is literature‑**un**supported (no source calls the 1:3 aperiodic; it is rational SU(2)) and rested on a T1‑vs‑binding‑energy mis‑label; no substrate test. Catalog only; survivor = `D‑LOSSCHAN`. | **Audit conclusions (the "double‑check" the operator asked for):** @@ -379,14 +387,23 @@ on both sides** (the next materialization). (HIP) and ρ = 0.965 (TWIG) and θ ∈ [1.45, 1.6] stop being *cited* and become *re‑measured* by `hpc::quantized` + P10 — so the map can't drift from the runtime's actual numbers. -6. **D‑EXCITON is honestly fenced.** It has **no internal test** — it's an - external physical analog. Marked as such so it can't masquerade as a - substrate‑validated claim. (This is the "doesn't dilute" line: not every - `[H]` earns a test; external analogs are grounded by literature only.) - -**No claim collapses** under this audit; the two that needed it -(D‑MOIRE absolutism — fixed in #47; D‑META64 — fixed in §4.1) were already -corrected. The rest are either CODED, ADR‑pinned, or `[H]`‑with‑a‑named‑test. +6. **D‑EXCITON was reverted `[H]→[S]` by the 5+3 agent review (2026‑06‑09).** + The `[H]` "same coprime aperiodicity as D‑BGZ17" rested on a **category + error** (a scalar ratio / two energy levels have no period to be aperiodic + about) **+ a T1‑vs‑binding‑energy mis‑label** (the "1.5 eV" was the T1 + excited‑state energy, not E_b ≈0.3–0.6 eV; corrected, the 1:3‑energy + coincidence does not survive). The `[S]→[H]` jump had itself violated §6.3 + (promotion needs a *measurement*, not a mechanism story) — so reverting + **restores** the discipline. Entry **kept** (append‑only); the two + `[G]`‑real legs (rational 1:3; ~100% IQE harvest physics) preserved, + reframes A/B logged rejected, the one untested survivor split out as + `D‑LOSSCHAN` (`[S]`). The apparatus (5 research + 3 brutal‑review agents) + is committed at `.claude/agents/`. + +**No claim *collapses*** (append‑only); but the 5+3 review **reverted** one +overclaim — D‑EXCITON `[H]→[S]` (this §, #6) — and split out its one survivor +(`D‑LOSSCHAN`), joining the prior fixes (D‑MOIRE absolutism — #47; D‑META64 — +§4.1). The rest are either CODED, ADR‑pinned, or `[H]`‑with‑a‑named‑test. --- @@ -480,3 +497,15 @@ isolation. The map's job is to keep them visible. Polyominoes* (ribbon‑tile covering; the `⌈(2/3)(m+n−√(m²−mn+n²))⌉` cardinality with the **Eisenstein / hexagonal norm** `m²−mn+n²` — ties the square cover to the hat's hexagonal substrate). +- **The OLED‑exciton papers (D‑EXCITON revert / D‑LOSSCHAN; 5+3 review + 2026‑06‑09):** the 1:3 singlet:triplet ratio is rational SU(2) spin + statistics, *not* aperiodic — the `[H]` "coprime aperiodicity" reading was + reverted to `[S]`. + - **PMC4614446** — Takahashi et al., *Sci. Rep.* 5:15533 (2015): + bias‑dependent S:T formation ratio (real but material‑specific). + - **arXiv 2311.03927** — triplet‑exciton binding energies in organic + semiconductors are sub‑eV (≈0.3–0.6 eV) — refutes the "~1.5 eV E_b" (that + figure is the T1 excited‑state energy, not a binding energy). + - **Baldo/Adachi/Forrest** (~100% internal phosphorescence, *J. Appl. + Phys.* 90:5048, 2001) + **Uoyama/Adachi** (TADF, *Nature* 492:234, 2012) + — the harvest‑the‑dark physics (`[G]`‑real, ≠ the substrate link).