core/field/state.py
Shay b61e79353a feat: manifold field topology, graph diffusion operator, vertical pulse
Add ManifoldState (N,32) versor field over graph edges, GraphDiffusionOperator
with damped convergence via construction_seed_versor closure, deterministic
hash-to-versor stub, and run_pulse.py end-to-end script proving injection →
propagation → vault recall → token output. 24 new tests, zero regressions
on architectural invariants.
2026-05-15 16:02:48 -07:00

122 lines
4.7 KiB
Python

"""
FieldState — the complete cognitive field at one moment.
Invariant: versor_condition(F) < 1e-6 always.
This is checked at injection and maintained structurally by versor_apply().
FieldState is immutable by design (frozen=True, slots=True).
The np.ndarray F is copied and validated at construction — the copy() call
is the explicit contract boundary. Callers must not retain a mutable
reference to the array passed in and expect coherence.
"""
from __future__ import annotations
from dataclasses import dataclass
from typing import TYPE_CHECKING
import numpy as np
if TYPE_CHECKING:
from core.physics.energy import EnergyProfile
from core.physics.valence import ValenceBundle
_EXPECTED_COMPONENTS = 32
@dataclass(frozen=True, slots=True)
class FieldState:
F: np.ndarray # shape (32,) float32/float64 — Cl(4,1) multivector on the versor manifold
node: int = 0 # current node index in the vocabulary manifold
step: int = 0 # number of propagation steps taken
holonomy: np.ndarray | None = None
energy: EnergyProfile | None = None
valence: ValenceBundle | None = None
def __post_init__(self) -> None:
# Enforce copy + dtype + shape at the construction boundary.
# frozen=True prevents reassignment, but ndarray contents are still
# mutable via the array object; copy() here is the defence.
# slots=True closes __dict__ so no incidental attributes can be added.
f_dtype = np.asarray(self.F).dtype
if f_dtype not in (np.dtype(np.float32), np.dtype(np.float64)):
f_dtype = np.dtype(np.float32)
F = np.array(self.F, dtype=f_dtype).copy()
if F.shape != (_EXPECTED_COMPONENTS,):
raise ValueError(
f"FieldState.F must have shape ({_EXPECTED_COMPONENTS},), "
f"got {F.shape}."
)
# Bypass frozen to store the validated copy.
object.__setattr__(self, "F", F)
if self.holonomy is not None:
h_dtype = np.asarray(self.holonomy).dtype
if h_dtype not in (np.dtype(np.float32), np.dtype(np.float64)):
h_dtype = np.dtype(np.float32)
H = np.array(self.holonomy, dtype=h_dtype).copy()
if H.shape != (_EXPECTED_COMPONENTS,):
raise ValueError(
f"FieldState.holonomy must have shape ({_EXPECTED_COMPONENTS},), "
f"got {H.shape}."
)
object.__setattr__(self, "holonomy", H)
def advance(self, new_F: np.ndarray, new_node: int) -> FieldState:
"""Return a new FieldState after one propagation step."""
return FieldState(
F=new_F,
node=new_node,
step=self.step + 1,
holonomy=self.holonomy,
energy=self.energy,
valence=self.valence,
)
@dataclass(frozen=True, slots=True)
class ManifoldState:
"""Field over a graph topology — one versor per node, with edge connectivity.
Invariant: versor_condition(fields[i]) < 1e-6 for every node i.
"""
fields: np.ndarray # (N, 32) float32 — one Cl(4,1) versor per node
edges: np.ndarray # (E, 2) int32 — directed edge list
step: int = 0
def __post_init__(self) -> None:
from algebra.backend import versor_condition
F = np.array(self.fields, dtype=np.float32).copy()
if F.ndim != 2 or F.shape[1] != _EXPECTED_COMPONENTS:
raise ValueError(
f"ManifoldState.fields must have shape (N, {_EXPECTED_COMPONENTS}), "
f"got {F.shape}."
)
object.__setattr__(self, "fields", F)
E = np.array(self.edges, dtype=np.int32).copy()
if E.ndim != 2 or E.shape[1] != 2:
raise ValueError(
f"ManifoldState.edges must have shape (E, 2), got {E.shape}."
)
n_nodes = F.shape[0]
if E.size > 0 and (E.min() < 0 or E.max() >= n_nodes):
raise ValueError(
f"Edge indices must be in [0, {n_nodes}), "
f"got range [{E.min()}, {E.max()}]."
)
object.__setattr__(self, "edges", E)
for i in range(n_nodes):
vc = versor_condition(F[i])
if vc >= 1e-6:
raise ValueError(
f"ManifoldState.fields[{i}] violates versor_condition: {vc:.2e} >= 1e-6."
)
def with_fields(self, new_fields: np.ndarray) -> ManifoldState:
"""Return a new ManifoldState with updated field values."""
return ManifoldState(fields=new_fields, edges=self.edges, step=self.step)
def advance(self) -> ManifoldState:
"""Return a new ManifoldState one step forward."""
return ManifoldState(fields=self.fields, edges=self.edges, step=self.step + 1)