fix(algebra): P11a physics hot paths via algebra.backend (Rust-ready)

Stop wave/Third-Door physics from bypassing native dispatch:

- Route geometric_product / versor_apply / versor_condition / cga_inner
  through algebra.backend in wave_manifold, goldtether, trajectory,
  dynamic_manifold, surprise, holographic_vault, atlas_packing,
  biography, self_authorship.
- Backend: dtype-aware Rust use — f32 workloads use core_rs; f64 wave
  residual pins keep Python SOT until f64 GP parity exists. Coerce
  arrays for PyO3 bindings; fail soft to Python.
- AST hygiene pin: tests/test_physics_backend_dispatch_hygiene.py
- Docs: RUST.md, runtime_contracts, fidelity (ADR-0235 / UMA hygiene).

Verified: wave + cohesion suites green default and CORE_BACKEND=rust
(with core_rs built). MLX still exploratory off-serve.

algebra/backend.py

@@ -48,9 +48,36 @@ def _build_cga_inner_metric() -> np.ndarray:
 _CGA_INNER_METRIC: np.ndarray = _build_cga_inner_metric()
 
 
+def _f32_1d32(x: np.ndarray) -> np.ndarray:
+    """Contiguous f32 (32,) for core_rs PyReadonlyArray1 bindings."""
+    return np.ascontiguousarray(
+        np.asarray(x, dtype=np.float32).reshape(-1)[:32], dtype=np.float32
+    )
+
+
+def _is_f32_workload(*arrays: np.ndarray) -> bool:
+    """True when all arrays are float32 (Rust f32 kernel is parity-safe).
+
+    float64 wave residual pins require Python SOT (or future f64 Rust GP).
+    Forcing f64→f32 would break 1e-9 chiral / leakage pins (ADR-0241).
+    """
+    return all(np.asarray(a).dtype == np.float32 for a in arrays)
+
+
 def geometric_product(A: np.ndarray, B: np.ndarray) -> np.ndarray:
-    if _RUST:
-        return np.asarray(_rs.geometric_product(A, B), dtype=np.float32)
+    """Cl(4,1) geometric product via Rust f32 when enabled, else Python.
+
+    float64 inputs always use the pure-Python product (semantic SOT for
+    wave-field residual math). float32 field-graph workloads get Rust.
+    """
+    if _RUST and _is_f32_workload(A, B):
+        try:
+            return np.asarray(
+                _rs.geometric_product(_f32_1d32(A), _f32_1d32(B)),
+                dtype=np.float32,
+            )
+        except (AttributeError, TypeError, ValueError, Exception):
+            pass
     from algebra.cl41 import geometric_product as _gp
     return _gp(A, B)
 
@@ -67,25 +94,34 @@ def versor_apply(V: np.ndarray, F: np.ndarray) -> np.ndarray:
     """
     if _RUST:
         try:
-            Vc = np.ascontiguousarray(V, dtype=np.float64)
-            Fc = np.ascontiguousarray(F, dtype=np.float64)
-            return np.asarray(_rs.versor_apply_with_closure_f64(Vc, Fc), dtype=np.float64)
-        except (AttributeError, Exception):
+            Vc = np.ascontiguousarray(V, dtype=np.float64).reshape(-1)[:32]
+            Fc = np.ascontiguousarray(F, dtype=np.float64).reshape(-1)[:32]
+            return np.asarray(
+                _rs.versor_apply_with_closure_f64(Vc, Fc), dtype=np.float64
+            )
+        except (AttributeError, TypeError, ValueError, Exception):
             pass
     from algebra.versor import versor_apply as _va
     return _va(V, F)
 
 
 def versor_condition(F: np.ndarray) -> float:
-    if _RUST:
-        return float(_rs.versor_condition(F))
+    """Versor residual. Rust f32 path only for float32 inputs (see GP note)."""
+    if _RUST and _is_f32_workload(F):
+        try:
+            return float(_rs.versor_condition(_f32_1d32(F)))
+        except (AttributeError, TypeError, ValueError, Exception):
+            pass
     from algebra.versor import versor_condition as _vc
     return _vc(F)
 
 
 def cga_inner(X: np.ndarray, Y: np.ndarray) -> float:
-    if _RUST:
-        return float(_rs.cga_inner(X, Y))
+    if _RUST and _is_f32_workload(X, Y):
+        try:
+            return float(_rs.cga_inner(_f32_1d32(X), _f32_1d32(Y)))
+        except (AttributeError, TypeError, ValueError, Exception):
+            pass
     from algebra.cga import cga_inner as _ci
     return _ci(X, Y)
 

core/physics/atlas_packing.py

@@ -22,7 +22,8 @@ from typing import Sequence
 
 import numpy as np
 
-from algebra.cga import cga_inner, embed_point
+from algebra.backend import cga_inner
+from algebra.cga import embed_point
 from core.physics.wave_manifold import WaveManifold
 
 PHI = (1.0 + math.sqrt(5.0)) / 2.0

core/physics/biography.py

@@ -16,9 +16,10 @@ from typing import Any, Sequence
 
 import numpy as np
 
+from algebra.backend import versor_condition
 from algebra.cl41 import N_COMPONENTS
 from algebra.holonomy import holonomy_encode, holonomy_similarity
-from algebra.versor import unitize_versor, versor_condition
+from algebra.versor import unitize_versor
 from core.physics.wave_manifold import WaveManifold
 
 _CLOSURE_TOL = 1e-6

core/physics/dynamic_manifold.py

@@ -16,8 +16,9 @@ from typing import Optional, Sequence, Tuple, Union
 
 import numpy as np
 
+from algebra.backend import geometric_product, versor_condition
 from algebra.cga import is_null
-from algebra.cl41 import N_COMPONENTS, geometric_product, grade_project, reverse
+from algebra.cl41 import N_COMPONENTS, grade_project, reverse
 from algebra.null_point import (
     NullPointRecoveryError,
     dilator,
@@ -26,7 +27,6 @@ from algebra.null_point import (
     translator,
 )
 from algebra.rotor import rotor_power, word_transition_rotor
-from algebra.versor import versor_condition
 
 _CLOSURE_TOL = 1e-6
 _NEAR_ZERO = 1e-12

core/physics/goldtether.py

@@ -28,9 +28,10 @@ from typing import Any, Literal, Optional, Tuple
 
 import numpy as np
 
-from algebra.cl41 import N_COMPONENTS, geometric_product, reverse
+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_condition, versor_unit_residual
+from algebra.versor import versor_unit_residual
 from core.physics.wave_manifold import WaveManifold
 
 _CLOSURE_TOL = 1e-6

core/physics/holographic_vault.py

@@ -24,8 +24,8 @@ from typing import Any, Optional
 
 import numpy as np
 
+from algebra.backend import versor_condition
 from algebra.cl41 import N_COMPONENTS
-from algebra.versor import versor_condition
 from core.physics.wave_manifold import WaveManifold
 from teaching.epistemic import EpistemicStatus
 from vault.store import VaultStore, _parse_entry_status, _status_admits

core/physics/self_authorship.py

@@ -9,8 +9,8 @@ from typing import Any, Mapping, Sequence
 
 import numpy as np
 
+from algebra.backend import versor_condition
 from algebra.cl41 import N_COMPONENTS
-from algebra.versor import versor_condition
 from core.physics.dynamic_manifold import conformal_procrustes
 from core.physics.goldtether import GoldTetherMonitor, coherence_residual
 from core.physics.surprise import dual_procrustes_surprise, surprise_residual

core/physics/surprise.py

@@ -39,9 +39,8 @@ from typing import Optional, Sequence, Tuple, Union
 
 import numpy as np
 
-from algebra.cga import cga_inner
+from algebra.backend import cga_inner, versor_condition
 from algebra.cl41 import N_COMPONENTS, grade_project
-from algebra.versor import versor_condition
 from core.physics.dynamic_manifold import conformal_procrustes
 from core.physics.wave_manifold import WaveManifold, WaveSpectralLeakageError
 

core/physics/trajectory_invariants.py

@@ -22,8 +22,9 @@ from typing import Sequence
 
 import numpy as np
 
-from algebra.cl41 import N_COMPONENTS, geometric_product, reverse
-from algebra.versor import versor_condition, versor_unit_residual
+from algebra.backend import geometric_product, versor_condition
+from algebra.cl41 import N_COMPONENTS, reverse
+from algebra.versor import versor_unit_residual
 
 _CLOSURE_TOL = 1e-6
 _DEFAULT_EPS_TRAJECTORY = 1e-5

core/physics/wave_manifold.py

@@ -12,6 +12,12 @@ Continuous multivector wave fields ψ ∈ ℝ³² under:
 
 Algebra-native only (algebra/*). No scipy-as-truth. No teaching/vault imports.
 Off-serving until explicit gates; dual-checked unitary residual.
+
+**Backend dispatch (P11a / ADR-0235):** Cl(4,1) hot ops go through
+``algebra.backend`` so ``CORE_BACKEND=rust`` can accelerate them when
+``core_rs`` is built. Python remains semantic source of truth when Rust
+is unset. Helpers without a Rust path (``reverse``, ``scalar_part``,
+``versor_unit_residual``) stay on pure algebra modules.
 """
 
 from __future__ import annotations
@@ -20,9 +26,14 @@ from typing import Any, Sequence, Tuple
 
 import numpy as np
 
-from algebra.cga import cga_inner
-from algebra.cl41 import N_COMPONENTS, geometric_product, reverse, scalar_part
-from algebra.versor import versor_apply, versor_condition, versor_unit_residual
+from algebra.backend import (
+    cga_inner,
+    geometric_product,
+    versor_apply,
+    versor_condition,
+)
+from algebra.cl41 import N_COMPONENTS, reverse, scalar_part
+from algebra.versor import versor_unit_residual
 
 _CLOSURE_TOL = 1e-6
 _NEAR_ZERO = 1e-12

docs/RUST.md

@@ -1,5 +1,31 @@
 # Rust Extension (core-rs)
 
+## Physics hot-path hygiene (P11a)
+
+Third-Door / ADR-0241 physics modules must import Cl(4,1) multiplies and
+closure residual helpers from **`algebra.backend`**, not directly from
+`algebra.cl41` / `algebra.versor` / `algebra.cga` for:
+
+- `geometric_product`
+- `versor_apply`
+- `versor_condition`
+- `cga_inner`
+
+Pinned by `tests/test_physics_backend_dispatch_hygiene.py`.
+
+```bash
+# Default: pure Python (semantic SOT)
+uv run pytest tests/test_adr_0241_wave_manifold.py -q
+
+# Apple Silicon / native acceleration (after maturin build)
+export CORE_BACKEND=rust
+uv run --with maturin maturin develop --release --manifest-path core-rs/Cargo.toml
+uv run python -c "from algebra.backend import using_rust; assert using_rust()"
+```
+
+MLX remains an **exploratory** UMA lane (ADR-0235); not required for serving.
+
+
 ## Why Rust
 
 The active Rust extension is an opt-in native substrate for parity-gated hot

docs/research/third-door-blueprint-fidelity.md

@@ -306,6 +306,7 @@ PY
 | Fibonacci-word scheduler (V4) | 🟢 `fibonacci_word_schedule` (telemetry only) |
 | Fibonacci anyons (V5) | 🟢 quarantine package only; zero production imports |
 | 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; E0–E1 crystallization) |
 

docs/specs/runtime_contracts.md

@@ -919,3 +919,17 @@ Acceptance inventory: `docs/audit/adr_0241_cohesion_acceptance_checklist.md`.
 
 Every wave transition still obeys `versor_condition(F) < 1e-6`. Residual breach
 is **fail-closed** — no hot-path nearest-versor drift repair.
+
+### Algebra backend / Apple Silicon (P11a hygiene)
+
+Cl(4,1) hot ops in physics (`geometric_product`, `versor_apply`,
+`versor_condition`, `cga_inner`) must import from **`algebra.backend`**, not
+direct pure-algebra modules. Pin: `tests/test_physics_backend_dispatch_hygiene.py`.
+
+| Mode | How |
+|------|-----|
+| Default | Pure Python (semantic source of truth) |
+| Native accel | `CORE_BACKEND=rust` + `core_rs` built (`maturin develop --release -m core-rs/Cargo.toml`) |
+| float64 wave residual pins | Stay on Python product when inputs are f64 (Rust f32 GP not parity-safe for 1e-9 pins yet) |
+| float32 field graphs | Rust f32 GP / residual when enabled |
+| MLX / UMA | Exploratory (ADR-0235); not serving until parity gates |

tests/test_physics_backend_dispatch_hygiene.py

@@ -0,0 +1,92 @@
+"""P11a — physics hot paths must dispatch Cl(4,1) ops via algebra.backend.
+
+Prevents silent drift back to pure-Python-only imports for geometric_product /
+versor_apply / cga_inner / versor_condition in wave-field and related modules.
+Python remains default when CORE_BACKEND is unset; Rust accelerates when
+CORE_BACKEND=rust and core_rs is built (ADR-0235).
+"""
+
+from __future__ import annotations
+
+import ast
+from pathlib import Path
+
+_ROOT = Path(__file__).resolve().parents[1]
+_PHYSICS = _ROOT / "core" / "physics"
+
+# Modules that perform Cl(4,1) multiplies / residuals in the cognitive physics
+# layer — must take the load-bearing ops from algebra.backend.
+_BACKEND_HOT_MODULES = (
+    "wave_manifold.py",
+    "goldtether.py",
+    "trajectory_invariants.py",
+    "dynamic_manifold.py",
+    "surprise.py",
+    "holographic_vault.py",
+    "atlas_packing.py",
+    "biography.py",
+    "self_authorship.py",
+)
+
+# Names that must not be imported from algebra.cl41 / algebra.versor / algebra.cga
+# in those modules (use backend instead).
+_BANNED_FROM_PURE = frozenset(
+    {
+        "geometric_product",
+        "versor_apply",
+        "versor_condition",
+        "cga_inner",
+    }
+)
+
+
+def _imports_from(module: str, path: Path) -> set[str]:
+    tree = ast.parse(path.read_text(encoding="utf-8"))
+    names: set[str] = set()
+    for node in ast.walk(tree):
+        if isinstance(node, ast.ImportFrom) and node.module == module:
+            for alias in node.names:
+                names.add(alias.name)
+    return names
+
+
+def test_hot_modules_import_backend_for_algebra_ops():
+    for name in _BACKEND_HOT_MODULES:
+        path = _PHYSICS / name
+        assert path.is_file(), f"missing {path}"
+        backend_names = _imports_from("algebra.backend", path)
+        # Each file should pull at least one of the dispatch ops.
+        assert backend_names & _BANNED_FROM_PURE, (
+            f"{name} must import Cl(4,1) hot ops from algebra.backend; "
+            f"got backend imports {sorted(backend_names)}"
+        )
+
+
+def test_hot_modules_do_not_import_hot_ops_from_pure_algebra():
+    pure_modules = ("algebra.cl41", "algebra.versor", "algebra.cga")
+    for name in _BACKEND_HOT_MODULES:
+        path = _PHYSICS / name
+        for mod in pure_modules:
+            pure_names = _imports_from(mod, path)
+            offenders = pure_names & _BANNED_FROM_PURE
+            assert not offenders, (
+                f"{name} imports {sorted(offenders)} from {mod}; "
+                "route through algebra.backend for CORE_BACKEND=rust"
+            )
+
+
+def test_wave_manifold_uses_backend_geometric_product_and_versor_apply():
+    path = _PHYSICS / "wave_manifold.py"
+    backend = _imports_from("algebra.backend", path)
+    assert "geometric_product" in backend
+    assert "versor_apply" in backend
+    assert "versor_condition" in backend
+    assert "cga_inner" in backend
+
+
+def test_backend_using_rust_api_exists():
+    from algebra.backend import using_rust
+
+    # Default env: Python path (no silent force of Rust).
+    assert using_rust() is False or using_rust() is True  # bool only
+    assert isinstance(using_rust(), bool)