Merge pull request 'feat(physics): chiral orientation sign-gate — sgn(Q)=const fail-closed (ADR-0241 §2.4C)' (#41) from feat/chiral-sign-preservation-gate into main
Some checks failed
full-pytest / fast pytest (-m "not quarantine and not slow" -n 2) (push) Failing after 12m19s
lane-shas / verify pinned lane SHAs (push) Has been cancelled

Reviewed-on: #41
This commit is contained in:
Joshua Matthew-Catudio Shay 2026-07-15 21:37:22 +00:00
commit 717aca9bab
6 changed files with 362 additions and 4 deletions

122
core/physics/chiral_gate.py Normal file
View file

@ -0,0 +1,122 @@
"""Chiral orientation gate — sgn(Q) preservation (ADR-0241 §2.4C / core_ha §5.2).
The topological spinor charge Q = ψ I₅ ψ̃ anchors the global orientation of
the cognitive manifold: "preserving the sign and magnitude of Q_top across all
scales prevent[s] mirror-image inversions" (ADR-0241 §2.4C), and the
core_ha unification memo pins the invariant safeguard as sgn(Q) = const
(§5.2). The substrate already carries a verified non-vacuous READOUT
(:meth:`WaveManifold.chiral_charge` conserved exactly under ψ ), but
its only consumer took ``abs(...)``, discarding the sign. This module is the
missing ENFORCEMENT: latch the orientation on the first non-vacuous reading,
fail closed on any materially re-emerging flip.
Honesty pins (#19 family):
* Even field-states have Q structurally ~0 no orientation is defined and
the gate never fabricates one (``vacuous``; the retired grade-5 gate is
not revived). Serve-path even versors therefore never latch: wiring this
gate into GoldTether is behaviorally inert for today's serve fields.
* |Q| may legitimately pass below the floor (superposition mixtures); the
latch persists and only a materially re-emerging opposite sign violates.
Tier-1 module: imported by ``goldtether`` (sanctioned serve substrate).
Algebra-native only; no teaching/vault imports.
"""
from __future__ import annotations
from dataclasses import dataclass, field
from typing import Any, Literal
import numpy as np
from core.physics.wave_manifold import WaveManifold
#: Below this |Q|, orientation is undefined (even-field honesty; #19 family).
DEFAULT_Q_FLOOR = 0.1
class ChiralOrientationError(ValueError):
"""Fail-closed chiral orientation violation (mirror inversion).
A sign flip of Q is unreachable under sanctioned rotor transport (Q is
strictly conserved under left unitary multiplication), so a materially
flipped sign is evidence of a non-unitary / corrupting transition and the
gate refuses it rather than averaging it away.
"""
def __init__(self, reason: str, **disclosure: Any) -> None:
self.reason = reason
self.disclosure = dict(disclosure)
super().__init__(f"chiral_orientation refused [{reason}]: {self.disclosure}")
@dataclass(frozen=True, slots=True)
class ChiralObservation:
"""One gate reading with an honest epistemic verdict."""
q: float
sign: int # 1 / +1, or 0 when |q| < q_floor (no orientation defined)
latched_sign: int # 0 until the first non-vacuous reading
verdict: Literal["vacuous", "latched", "conserved"]
def as_dict(self) -> dict[str, Any]:
return {
"q": self.q,
"sign": self.sign,
"latched_sign": self.latched_sign,
"verdict": self.verdict,
}
@dataclass(slots=True)
class ChiralOrientationGate:
"""Monitor-style latch for the global chiral orientation sign.
* First reading with |Q| ``q_floor`` latches ``sgn(Q)``.
* Later readings: same sign ``conserved``; |Q| < floor ``vacuous``
(latch persists); opposite sign at material |Q| raises
:class:`ChiralOrientationError` (fail-closed; never averages).
"""
q_floor: float = DEFAULT_Q_FLOOR
latched_sign: int = 0
_wave: WaveManifold = field(default_factory=WaveManifold)
def observe(self, psi: np.ndarray) -> ChiralObservation:
"""Read Q from the field and enforce sign preservation."""
return self.observe_q(self._wave.chiral_charge(psi))
def observe_q(self, q: float) -> ChiralObservation:
"""Enforce sign preservation on a precomputed charge readout."""
q = float(q)
sign = 0 if abs(q) < self.q_floor else (1 if q > 0.0 else -1)
if sign == 0:
# No orientation defined at this reading; latch (if any) persists.
return ChiralObservation(
q=q, sign=0, latched_sign=self.latched_sign, verdict="vacuous"
)
if self.latched_sign == 0:
self.latched_sign = sign
return ChiralObservation(
q=q, sign=sign, latched_sign=sign, verdict="latched"
)
if sign == self.latched_sign:
return ChiralObservation(
q=q, sign=sign, latched_sign=self.latched_sign, verdict="conserved"
)
raise ChiralOrientationError(
"orientation_flip",
q=q,
sign=sign,
latched_sign=self.latched_sign,
q_floor=self.q_floor,
)
__all__ = [
"DEFAULT_Q_FLOOR",
"ChiralObservation",
"ChiralOrientationError",
"ChiralOrientationGate",
]

View file

@ -32,6 +32,7 @@ from algebra.backend import geometric_product, versor_condition
from algebra.cl41 import N_COMPONENTS, reverse
from algebra.rotor import rotor_power, word_transition_rotor
from algebra.versor import versor_unit_residual
from core.physics.chiral_gate import ChiralOrientationGate
from core.physics.wave_manifold import WaveManifold
_CLOSURE_TOL = 1e-6
@ -152,6 +153,13 @@ class GoldTetherMonitor:
r_floor: float = 0.1
r_critical: float = 1.0
gold_invariants: list = field(default_factory=_primal_gold_invariants, compare=False)
# Chiral orientation enforcement (ADR-0241 §2.4C / core_ha §5.2):
# sgn(Q) latches on the first non-vacuous spinor reading and a materially
# re-emerging flip fails closed. Even serve field-states are vacuous
# (Q ≈ 0), so the gate never latches on today's serve path — inert there.
chiral_gate: "ChiralOrientationGate" = field(
default_factory=lambda: ChiralOrientationGate(), compare=False
)
@property
def supervised_autonomy_level(self) -> float:
@ -227,7 +235,12 @@ class GoldTetherMonitor:
# path can move the residual without a second API.
wave = WaveManifold()
drift = wave.measure_unitary_residual(F_arr)
chiral = abs(float(wave.chiral_charge(F_arr)))
# Signed charge feeds the fail-closed orientation gate (sgn(Q)=const,
# ADR-0241 §2.4C / core_ha §5.2). The residual term keeps its
# magnitude-only semantics unchanged below.
q_signed = float(wave.chiral_charge(F_arr))
self.chiral_gate.observe_q(q_signed)
chiral = abs(q_signed)
drift_term = (
(drift + chiral) / self.epsilon_drift
if self.epsilon_drift > 0.0

View file

@ -0,0 +1,52 @@
# ADR-0241 / ADR-0242 — D10 Acceptance Packet for Joshua
**Date:** 2026-07-15 · **Prepared by:** Claude session (orchestration + adversarial verification) — **prepare-only; nothing here self-Accepts.**
**State under review:** `main @ 54228d45` (PR #36#40 merged) + PR #41 pending (chiral sign-gate).
**Companion documents:**
- `docs/audit/adr_0241_cohesion_acceptance_checklist.md` — the P12 checklist itself
- `docs/research/third-door-blueprint-fidelity.md` §12 — behavioral acceptance table (every 🟢 row names a failable pin)
- `docs/research/adr-0241-0242-adversarial-and-fidelity-findings.md` — hostile independent verification (W1)
- `docs/research/adr-0241-0242-blueprint-integration-plan.md` — all 4 authority blueprints dispositioned clause-by-clause
- `docs/plans/adr-0241-0242-session-completion-summary-2026-07-15.md` — living status surface (Plan A/B archived beside it)
---
## 1. Evidence summary (what is proven, and how)
| Claim | Evidence |
|---|---|
| Wave substrate behaves per ADR-0241 §2 | Dedicated suites green (wave_manifold / cohesion I-01…I-05 / Trace A/B); independently re-run this session |
| Chiral charge non-vacuous + conserved (P8) | **Executed adversarially**: Q = 2.5 on the mixed-parity fixture (grade-1 ⊕ grade-4), transport drift 0.0e+00 exactly, even-versor Q = 0 (no #19 revival) |
| Fibonacci cert discipline (ADR-0242 V1) | **Executed adversarially**: live func-call counter == cert.evaluations == budget across 8/15/20/21; dual-run digest byte-stable |
| P9 seal discipline | `seal_mode` SPECULATIVE-by-construction; contemplation seam defensively re-checks; COHERENT requires `seal_mode_reviewed(authorized=True)` |
| Serve containment (A-04) | W1 found a **live transitive breach** (5 off-serving modules loaded via the `core/physics` barrel) → **fixed in #40** (lazy Tier-2 exports + process-level `sys.modules` guard test) |
| Deterministic lanes | lane-shas 9/9 pinned SHAs + CLAIMS.md current — local at the exact merged tip AND server-side (runs #184/#185 green) |
| PR gates | smoke 175 (×3 runs), affected-file lane 179, goldtether-consumer regression 99 — all green |
## 2. Deviations submitted for ruling
1. **P7 true Clifford polar — demoted with ill-posedness proof.** Plan A's written bar was "landed, not demoted." The implementation *proves* `~C C` is non-scalar for multigrade fields (`test_true_clifford_polar_fails_on_multigrade_field`), so the analytic polar `R = C(~C C)^{-1/2}` is ill-posed there; sandwich conjugacy (SVD + Spin-GN) is the honest authority. **Recommend: accept-as-honest** per the #19 RETIRE precedent. Note the implementation is *more* faithful to the algebra than the blueprint's own §4 prototype (which used Euclidean `np.outer`/SVD proxies).
2. **Serve-boundary reconciliation (T1/T2) — ruled by Shay 2026-07-15, submitted for ratification.** `wave_manifold` is Tier-1 sanctioned serve substrate (goldtether/surprise/biography delegate to it; the subsumption map and the A-04 ban list could not both be true). Tier-2 (holographic_vault, fibonacci_search, wave_energy_boundary, multi_scale_energy, wave_seam, sensorium_wave_feed, atlas_packing) stays off-serving, now enforced **process-level** (transitive guard), not just direct-AST.
3. **Sensorium feed** — staged packets + real ρ algebra; real compilers remain open (W5).
4. **Continuous-field integrals** — progressive (mode samples), research track.
5. **Holographic recall precision** — float32-honest (I-02 tol 1e-6), not the blueprint's implied bit-exactness; already pinned in-suite.
## 3. R&D-memo claims deliberately filtered (Accept should be informed)
- **Fibonacci anyons** ("immune to float32 drift") — NOT BUILT; ledger row corrected this session from a prose-only 🟢 to ⚪ claim-quarantined (W1 Finding #5).
- **Fibonacci "avoids division" cost claim** — implementation uses the correct ratio subdivision; the no-division micro-claim only holds on integer lattices and was not chased.
## 4. Capability additions since the checklist was written
- **Chiral orientation sign-gate** (`sgn(Q)=const`, ADR-0241 §2.4C / core_ha §5.2 mirror-inversion safeguard) — the audit's one missing enforcement, built TDD, **PR #41 pending**: fail-closed `ChiralOrientationError` on materially re-emerging flips; signed charge wired in `goldtether_residual` with residual values pinned byte-identical; vacuous (inert) on today's even serve fields.
## 5. Known open items (explicitly NOT completion blockers)
1. **ADR-0242 R&D memo** ("deterministic-fibonacci-operators-and-evidence-gated-optimization") has no local export — its blueprint-fidelity slice is partial (implementation verified against its own contract + committed ADR).
2. **CRDT-vs-bit-exact determinism** — core_ha memo claims Delta-CRDT in `core/sync/`; actual mechanism is bit-exact codec + single-writer VaultStore. Needs a decision (build CRDT for multi-writer, or correct the doc).
3. W5 backlog: Rust f64 GP parity (D9), real sensorium compilers, V2 energy promotion (benchmark-gated), progressive continuous field.
4. GitHub Actions mirror CI — billing-locked; validation is local-first + Forgejo Act per the AGENTS.md protocol (merged in #40).
## 6. Requested action
Rule on §2 deviations (recommendations inline) and, if satisfied, flip **ADR-0241 and ADR-0242 → Accepted**. PR #41 (chiral gate) can be folded into the Accept or ruled separately.

View file

@ -52,7 +52,7 @@ The recurring pattern across all four docs: the blueprints were written when the
### 🔨 MISSING PIECE — capability the vision specifies but the code does **not** yet enforce; **build anew** so we don't go backwards
| Missing invariant | Current state | Build |
|---|---|---|
| **Chiral SIGN-preservation gate** (ADR-0241 §2.4C + core_ha §5.2: `sgn(∫⟨ψIψ̃⟩₀)=const`, mirror-inversion protection) | `chiral_charge` is a verified non-vacuous **readout**, but its only consumer (`goldtether.py:230`) takes `abs(...)`**discarding the sign** — and it's ~0 on serve-path even-versors. No fail-closed `sgn(Q)=const` gate exists. | A `chiral_orientation_gate` that latches the initial sign and fails closed on flip during transport; wire the *signed* charge (not `abs`) where a non-vacuous spinor path exists. Preserves the "topologically secure alignment" the blueprint promises. |
| **Chiral SIGN-preservation gate** (ADR-0241 §2.4C + core_ha §5.2: `sgn(∫⟨ψIψ̃⟩₀)=const`, mirror-inversion protection) | **BUILT** (`feat/chiral-sign-preservation-gate`, stacked on PR #40): `core/physics/chiral_gate.py``ChiralOrientationGate` latches `sgn(Q)` on the first non-vacuous reading, fails closed (`ChiralOrientationError`) on a materially re-emerging flip; wired at the enforcement point (`goldtether_residual` now feeds the *signed* charge to the gate; residual keeps magnitude semantics byte-identical). Even serve fields stay vacuous → inert on today's serve path (no #19 revival). Pins: `tests/test_chiral_orientation_gate.py` (7). | — (was: latch initial sign + fail closed on flip; wire signed charge) |
| **CRDT delta-sync semilattice** (core_ha §2/§3 tombstone → "commutative/associative/idempotent Delta-CRDT in `core/sync/`") | `core/sync/` is a journal/object-store (no `merge`/idempotent/semilattice semantics found). Determinism today = bit-exact `array_codec` + single-writer `VaultStore` (Shape B+), **not** multi-writer CRDT merge. | Decide: either build the CRDT (needed for genuine multi-agent/multi-writer "one continuous life") **or** correct the doc to state determinism is single-writer bit-exact. Don't leave the claim floating. |
### ⛔ SCRAP / DO-NOT-BUILD — correctly rejected over-claims
@ -83,7 +83,7 @@ The recurring pattern across all four docs: the blueprints were written when the
1. **DONE** — Finding #2 serve-boundary reconciliation (de-barrel + transitive guard + A-04/ledger correction). Green.
2. **Next fix-forward (LOW, batch):** downgrade anyon ledger row; `EpistemicStatus` boundary note (or relocate to shared kernel); Fibonacci `minimizer` docstring; `ClosureViolationException` naming note.
3. **Build-anew (capability):** chiral **sign-preservation gate** (§5.2) — the one true missing safeguard; TDD a `sgn(Q)=const` fail-closed gate + signed wiring.
3. **DONE** — chiral **sign-preservation gate** (§5.2) built TDD (RED→GREEN) on `feat/chiral-sign-preservation-gate`: gate primitive + goldtether wiring + 7 pins; 99 goldtether-consumer tests green.
4. **Decision + build/doc:** CRDT-vs-bit-exact determinism story for `core/sync` (affects multi-agent "one continuous life").
5. **Get ADR-0242 memo export** → close its fidelity slice (evidence-gated-optimization vs the cert impl; confirm no further deviations).
6. Staged/gated items (Rust parity D9, sensorium compilers, continuous integrals) remain post-Accept backlog.

View file

@ -304,11 +304,12 @@ PY
| κ cert gate fail → baseline 1.0 (V1b) | 🟢 `propose_kappa_from_search` / `goldtether.propose_kappa_line_search` |
| Multi-scale \(\tau_n=F_n\tau_0\) + \(E_n(t)\) helpers (V2) | 🟢 research API (`multi_scale_energy`); production energy default unchanged |
| Fibonacci-word scheduler (V4) | 🟢 `fibonacci_word_schedule` (telemetry only) |
| Fibonacci anyons (V5) | 🟢 quarantine package only; zero production imports |
| Fibonacci anyons (V5) | ⚪ NOT BUILT — claim-quarantined (no package, no tests; W1 Finding #5 caught this row as prose-only green). The R&D memo's "immune to float32 drift" claim is deliberately unimplemented; revisit only if non-abelian braiding is explicitly greenlit. |
| Sensorium → ψ feed (I-04) | 🟢 `sensorium_wave_feed` (fake packets + real \(\rho\)) |
| Physics Cl(4,1) via `algebra.backend` (P11a) | 🟢 wave/goldtether/trajectory/procrustes/surprise/vault/packing; AST pin; Rust when `CORE_BACKEND=rust` |
| Contemplation Trace A SPECULATIVE holographic seal (P9) | 🟢 `core/contemplation/wave_seam.py` (hypothesis vs COHERENT evidence) |
| Energy boundary + multi-scale τ (P10 Trace B) | 🟢 `wave_energy_boundary` (wave residual → energy/trajectory; τ_n=F_n·τ_0; E0E1 crystallization) |
| Chiral orientation sign-gate \(\mathrm{sgn}(Q)=\text{const}\) (§2.4C / core_ha §5.2 mirror-inversion safeguard) | 🟢 `chiral_gate.ChiralOrientationGate` — latch on first non-vacuous \(Q\), fail-closed `ChiralOrientationError` on flip; wired via signed charge in `goldtether_residual` (residual value unchanged; even fields vacuous → serve-inert, no #19 revival). Pins: `test_chiral_orientation_gate.py` |
### Subsumption map (Slice 23)
| Operator | Delegation |

View file

@ -0,0 +1,170 @@
"""ADR-0241 §2.4C / core_ha §5.2 — chiral orientation gate (sgn(Q) = const).
RED until core/physics/chiral_gate.py lands.
The blueprint's mirror-inversion safeguard: the global chiral anomaly sign of
the spinor field must stay constant "preserving the sign … anchors the
global orientation of the cognitive manifold, preventing mirror-image
inversions." The substrate has a verified non-vacuous READOUT
(WaveManifold.chiral_charge; Q = ψ I₅ ψ̃), but its only consumer
(GoldTetherMonitor.goldtether_residual) takes abs(), discarding the sign
so orientation was measured, never enforced (integration-plan missing piece).
Fixture algebra (grade-1 v, central I₅ with I₅² = 1, Ĩ₅ = I₅):
Q(v + v·I₅) = 2 and Q(v v·I₅) = +2
so the mirror pair is a genuine sign flip and it is unreachable by any
rotor transport (Q is strictly conserved under ψ ), i.e. it models
exactly the corruption class the gate exists to catch.
"""
from __future__ import annotations
import numpy as np
import pytest
from algebra.cl41 import N_COMPONENTS, geometric_product
from algebra.rotor import make_rotor_from_angle
from core.physics.chiral_gate import (
ChiralObservation,
ChiralOrientationError,
ChiralOrientationGate,
)
from core.physics.goldtether import GoldTetherMonitor
from core.physics.wave_manifold import WaveManifold, _I5
def _e(i: int, val: float = 1.0) -> np.ndarray:
v = np.zeros(N_COMPONENTS, dtype=np.float64)
v[i] = val
return v
def _spinor_pair() -> tuple[np.ndarray, np.ndarray]:
"""(ψ, mirror-ψ) with Q(ψ) = 2.5 and Q(mirror) = +2.5."""
v = _e(1) + 0.5 * _e(3)
psi = v + geometric_product(v, _I5)
mirror = v - geometric_product(v, _I5)
return psi, mirror
# --- Latch + conservation ----------------------------------------------------
def test_gate_latches_then_reports_conserved_under_left_spinor_transport():
gate = ChiralOrientationGate()
M = WaveManifold()
psi, _ = _spinor_pair()
first = gate.observe(psi)
assert isinstance(first, ChiralObservation)
assert first.verdict == "latched"
assert first.sign == -1
assert first.latched_sign == -1
assert abs(first.q) > 0.1
R = make_rotor_from_angle(0.4, bivector_idx=7)
psi_next = M.left_spinor_step(psi, R)
second = gate.observe(psi_next)
assert second.verdict == "conserved"
assert second.sign == -1
assert abs(second.q - first.q) < 1e-9 # Q strictly conserved
def test_gate_fails_closed_on_mirror_inversion():
gate = ChiralOrientationGate()
psi, mirror = _spinor_pair()
gate.observe(psi)
with pytest.raises(ChiralOrientationError, match="orientation_flip"):
gate.observe(mirror)
# --- Even-field honesty (#19 family) ------------------------------------------
def test_gate_honest_on_even_versor_never_latches():
"""Even field-states have Q ≈ 0: no orientation is defined, the gate must
not fabricate one, and repeated vacuous observations are not violations
(does not revive the retired #19 gate)."""
gate = ChiralOrientationGate()
for angle, plane in ((0.9, 11), (0.3, 6), (0.55, 8)):
obs = gate.observe(make_rotor_from_angle(angle, bivector_idx=plane))
assert obs.verdict == "vacuous"
assert obs.sign == 0
assert obs.latched_sign == 0
def test_gate_vacuous_after_latch_is_not_violation_but_flip_reemergence_is():
"""|Q| may legitimately pass below the floor (superposition states); the
latch persists, and a materially re-emerging FLIPPED sign still fails."""
gate = ChiralOrientationGate()
psi, mirror = _spinor_pair()
gate.observe(psi)
even = make_rotor_from_angle(0.7, bivector_idx=9)
mid = gate.observe(even)
assert mid.verdict == "vacuous"
assert mid.latched_sign == -1 # latch persists
same = gate.observe(psi)
assert same.verdict == "conserved"
with pytest.raises(ChiralOrientationError, match="orientation_flip"):
gate.observe(mirror)
# --- Determinism ---------------------------------------------------------------
def test_gate_determinism():
psi, _ = _spinor_pair()
even = make_rotor_from_angle(0.2, bivector_idx=6)
seq = (psi, even, psi)
def run() -> tuple[str, ...]:
gate = ChiralOrientationGate()
return tuple(gate.observe(x).verdict for x in seq)
assert run() == run() == ("latched", "vacuous", "conserved")
# --- GoldTether wiring (the enforcement point) ---------------------------------
def test_goldtether_residual_latches_and_fails_closed_on_flip():
monitor = GoldTetherMonitor()
psi, mirror = _spinor_pair()
r = monitor.goldtether_residual(psi)
assert np.isfinite(r)
assert monitor.chiral_gate.latched_sign == -1
with pytest.raises(ChiralOrientationError, match="orientation_flip"):
monitor.goldtether_residual(mirror)
def test_goldtether_residual_unchanged_and_inert_on_even_serve_fields():
"""Serve-path even field-states are vacuous: the gate never latches and
the residual VALUE is byte-identical to the pre-gate computation
(drift + abs(chiral) magnitude semantics untouched)."""
monitor = GoldTetherMonitor()
wave = WaveManifold()
F = make_rotor_from_angle(0.42, bivector_idx=8)
r = monitor.goldtether_residual(F)
assert monitor.chiral_gate.latched_sign == 0
# Replicate the pre-gate residual formula exactly (ADR-0238 §2.3): the
# gate must not perturb the value. Fresh monitors seed primal gold
# invariants, so the geometric term is live.
drift = wave.measure_unitary_residual(F)
chiral = abs(float(wave.chiral_charge(F)))
drift_term = (drift + chiral) / monitor.epsilon_drift
scale = float(np.linalg.norm(F))
min_dist = min(
float(np.linalg.norm(F - np.asarray(inv, dtype=np.float64)))
for inv in monitor.gold_invariants
)
geo_term = min_dist / scale
expected = monitor.w_drift * drift_term + (1.0 - monitor.w_drift) * geo_term
assert abs(r - expected) < 1e-15