tnfr 3.0.3__py3-none-any.whl → 8.5.0__py3-none-any.whl

tnfr/operators/__init__.py

@@ -0,0 +1,1672 @@
+"""Network operators.
+
+Operator helpers interact with TNFR graphs adhering to
+:class:`tnfr.types.GraphLike`, relying on ``nodes``/``neighbors`` views,
+``number_of_nodes`` and the graph-level ``.graph`` metadata when applying
+structural transformations.
+"""
+
+from __future__ import annotations
+
+import heapq
+import math
+from collections.abc import Callable, Iterator
+from itertools import islice
+from statistics import StatisticsError, fmean
+from typing import TYPE_CHECKING, Any
+
+from tnfr import glyph_history
+
+from ..alias import get_attr
+from ..constants import DEFAULTS, get_param
+from ..constants.aliases import ALIAS_EPI, ALIAS_VF
+from ..utils import angle_diff
+from ..metrics.trig import neighbor_phase_mean
+from ..rng import make_rng
+from ..types import EPIValue, Glyph, NodeId, TNFRGraph
+from ..utils import get_nodenx
+from . import definitions as _definitions
+from .jitter import (
+    JitterCache,
+    JitterCacheManager,
+    get_jitter_manager,
+    random_jitter,
+    reset_jitter_manager,
+)
+from .registry import OPERATORS, discover_operators, get_operator_class
+from .remesh import (
+    apply_network_remesh,
+    apply_remesh_if_globally_stable,
+    apply_topological_remesh,
+)
+
+_remesh_doc = (
+    "Trigger a remesh once the stability window is satisfied.\n\n"
+    "Parameters\n----------\n"
+    "stable_step_window : int | None\n"
+    "    Number of consecutive stable steps required before remeshing.\n"
+    "    Only the English keyword 'stable_step_window' is supported."
+)
+if apply_remesh_if_globally_stable.__doc__:
+    apply_remesh_if_globally_stable.__doc__ += "\n\n" + _remesh_doc
+else:
+    apply_remesh_if_globally_stable.__doc__ = _remesh_doc
+
+discover_operators()
+
+_DEFINITION_EXPORTS = {
+    name: getattr(_definitions, name) for name in getattr(_definitions, "__all__", ())
+}
+globals().update(_DEFINITION_EXPORTS)
+
+if TYPE_CHECKING:  # pragma: no cover - type checking only
+    from ..node import NodeProtocol
+
+GlyphFactors = dict[str, Any]
+GlyphOperation = Callable[["NodeProtocol", GlyphFactors], None]
+
+from .grammar import apply_glyph_with_grammar  # noqa: E402
+from .health_analyzer import SequenceHealthAnalyzer, SequenceHealthMetrics  # noqa: E402
+from .hamiltonian import (
+    InternalHamiltonian,
+    build_H_coherence,
+    build_H_frequency,
+    build_H_coupling,
+)  # noqa: E402
+from .pattern_detection import (  # noqa: E402
+    PatternMatch,
+    UnifiedPatternDetector,
+    detect_pattern,
+    analyze_sequence,
+)
+
+__all__ = [
+    "JitterCache",
+    "JitterCacheManager",
+    "get_jitter_manager",
+    "reset_jitter_manager",
+    "random_jitter",
+    "get_neighbor_epi",
+    "get_glyph_factors",
+    "GLYPH_OPERATIONS",
+    "apply_glyph_obj",
+    "apply_glyph",
+    "apply_glyph_with_grammar",
+    "apply_network_remesh",
+    "apply_topological_remesh",
+    "apply_remesh_if_globally_stable",
+    "OPERATORS",
+    "discover_operators",
+    "get_operator_class",
+    "SequenceHealthMetrics",
+    "SequenceHealthAnalyzer",
+    "InternalHamiltonian",
+    "build_H_coherence",
+    "build_H_frequency",
+    "build_H_coupling",
+    # Pattern detection (unified module)
+    "PatternMatch",
+    "UnifiedPatternDetector",
+    "detect_pattern",
+    "analyze_sequence",
+]
+
+__all__.extend(_DEFINITION_EXPORTS.keys())
+
+
+def get_glyph_factors(node: NodeProtocol) -> GlyphFactors:
+    """Fetch glyph tuning factors for a node.
+
+    The glyph factors expose per-operator coefficients that modulate how an
+    operator reorganizes a node's Primary Information Structure (EPI),
+    structural frequency (νf), internal reorganization differential (ΔNFR), and
+    phase. Missing factors fall back to the canonical defaults stored at the
+    graph level.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        TNFR node providing a ``graph`` mapping where glyph factors may be
+        cached under ``"GLYPH_FACTORS"``.
+
+    Returns
+    -------
+    GlyphFactors
+        Mapping with operator-specific coefficients merged with the canonical
+        defaults. Mutating the returned mapping does not affect the graph.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self):
+    ...         self.graph = {"GLYPH_FACTORS": {"AL_boost": 0.2}}
+    >>> node = MockNode()
+    >>> factors = get_glyph_factors(node)
+    >>> factors["AL_boost"]
+    0.2
+    >>> factors["EN_mix"]  # Fallback to the default reception mix
+    0.25
+    """
+    return node.graph.get("GLYPH_FACTORS", DEFAULTS["GLYPH_FACTORS"].copy())
+
+
+def get_factor(gf: GlyphFactors, key: str, default: float) -> float:
+    """Return a glyph factor as ``float`` with a default fallback.
+
+    Parameters
+    ----------
+    gf : GlyphFactors
+        Mapping of glyph names to numeric factors.
+    key : str
+        Factor identifier to look up.
+    default : float
+        Value used when ``key`` is absent. This typically corresponds to the
+        canonical operator tuning and protects structural invariants.
+
+    Returns
+    -------
+    float
+        The resolved factor converted to ``float``.
+
+    Notes
+    -----
+    This function performs defensive validation to ensure numeric safety.
+    Invalid values (non-numeric, nan, inf) are silently replaced with the
+    default to prevent operator failures. For strict validation, use
+    ``validate_glyph_factors`` before passing factors to operators.
+
+    Examples
+    --------
+    >>> get_factor({"AL_boost": 0.3}, "AL_boost", 0.05)
+    0.3
+    >>> get_factor({}, "IL_dnfr_factor", 0.7)
+    0.7
+    """
+    value = gf.get(key, default)
+    # Defensive validation: ensure the value is numeric and finite
+    # Use default for invalid values to prevent operator failures
+    if not isinstance(value, (int, float, str)):
+        return default
+    try:
+        value = float(value)
+    except (ValueError, TypeError):
+        return default
+    if not math.isfinite(value):
+        return default
+    return value
+
+
+# -------------------------
+# Glyphs (local operators)
+# -------------------------
+
+
+def get_neighbor_epi(node: NodeProtocol) -> tuple[list[NodeProtocol], EPIValue]:
+    """Collect neighbour nodes and their mean EPI.
+
+    The neighbour EPI is used by reception-like glyphs (e.g., EN, RA) to
+    harmonise the node's EPI with the surrounding field without mutating νf,
+    ΔNFR, or phase. When a neighbour lacks a direct ``EPI`` attribute the
+    function resolves it from NetworkX metadata using known aliases.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose neighbours participate in the averaging.
+
+    Returns
+    -------
+    list of NodeProtocol
+        Concrete neighbour objects that expose TNFR attributes.
+    EPIValue
+        Arithmetic mean of the neighbouring EPIs. Equals the node EPI when no
+        valid neighbours are found, allowing glyphs to preserve the node state.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, epi, neighbors):
+    ...         self.EPI = epi
+    ...         self._neighbors = neighbors
+    ...         self.graph = {}
+    ...     def neighbors(self):
+    ...         return self._neighbors
+    >>> neigh_a = MockNode(1.0, [])
+    >>> neigh_b = MockNode(2.0, [])
+    >>> node = MockNode(0.5, [neigh_a, neigh_b])
+    >>> neighbors, epi_bar = get_neighbor_epi(node)
+    >>> len(neighbors), round(epi_bar, 2)
+    (2, 1.5)
+    """
+
+    epi = node.EPI
+    neigh = list(node.neighbors())
+    if not neigh:
+        return [], epi
+
+    if hasattr(node, "G"):
+        G = node.G
+        total = 0.0
+        count = 0
+        has_valid_neighbor = False
+        needs_conversion = False
+        for v in neigh:
+            if hasattr(v, "EPI"):
+                total += float(v.EPI)
+                has_valid_neighbor = True
+            else:
+                attr = get_attr(G.nodes[v], ALIAS_EPI, None)
+                if attr is not None:
+                    total += float(attr)
+                    has_valid_neighbor = True
+                else:
+                    total += float(epi)
+                needs_conversion = True
+            count += 1
+        if not has_valid_neighbor:
+            return [], epi
+        epi_bar = total / count if count else float(epi)
+        if needs_conversion:
+            NodeNX = get_nodenx()
+            if NodeNX is None:
+                raise ImportError("NodeNX is unavailable")
+            neigh = [
+                v if hasattr(v, "EPI") else NodeNX.from_graph(node.G, v) for v in neigh
+            ]
+    else:
+        try:
+            epi_bar = fmean(v.EPI for v in neigh)
+        except StatisticsError:
+            epi_bar = epi
+
+    return neigh, epi_bar
+
+
+def _determine_dominant(
+    neigh: list[NodeProtocol], default_kind: str
+) -> tuple[str, float]:
+    """Resolve the dominant ``epi_kind`` across neighbours.
+
+    The dominant kind guides glyphs that synchronise EPI, ensuring that
+    reshaping a node's EPI also maintains a coherent semantic label for the
+    structural phase space.
+
+    Parameters
+    ----------
+    neigh : list of NodeProtocol
+        Neighbouring nodes providing EPI magnitude and semantic kind.
+    default_kind : str
+        Fallback label when no neighbour exposes an ``epi_kind``.
+
+    Returns
+    -------
+    tuple of (str, float)
+        The dominant ``epi_kind`` together with the maximum absolute EPI. The
+        amplitude assists downstream logic when choosing between the node's own
+        label and the neighbour-driven kind.
+
+    Examples
+    --------
+    >>> class Mock:
+    ...     def __init__(self, epi, kind):
+    ...         self.EPI = epi
+    ...         self.epi_kind = kind
+    >>> _determine_dominant([Mock(0.2, "seed"), Mock(-1.0, "pulse")], "seed")
+    ('pulse', 1.0)
+    """
+    best_kind: str | None = None
+    best_abs = 0.0
+    for v in neigh:
+        abs_v = abs(v.EPI)
+        if abs_v > best_abs:
+            best_abs = abs_v
+            best_kind = v.epi_kind
+    if not best_kind:
+        return default_kind, 0.0
+    return best_kind, best_abs
+
+
+def _mix_epi_with_neighbors(
+    node: NodeProtocol, mix: float, default_glyph: Glyph | str
+) -> tuple[float, str]:
+    """Blend node EPI with the neighbour field and update its semantic label.
+
+    The routine is shared by reception-like glyphs. It interpolates between the
+    node EPI and the neighbour mean while selecting a dominant ``epi_kind``.
+    ΔNFR, νf, and phase remain untouched; the function focuses on reconciling
+    form.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node that exposes ``EPI`` and ``epi_kind`` attributes.
+    mix : float
+        Interpolation weight for the neighbour mean. ``mix = 0`` preserves the
+        current EPI, while ``mix = 1`` adopts the average neighbour field.
+    default_glyph : Glyph or str
+        Glyph driving the mix. Its value informs the fallback ``epi_kind``.
+
+    Returns
+    -------
+    tuple of (float, str)
+        The neighbour mean EPI and the resolved ``epi_kind`` after mixing.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, epi, kind, neighbors):
+    ...         self.EPI = epi
+    ...         self.epi_kind = kind
+    ...         self.graph = {}
+    ...         self._neighbors = neighbors
+    ...     def neighbors(self):
+    ...         return self._neighbors
+    >>> neigh = [MockNode(0.8, "wave", []), MockNode(1.2, "wave", [])]
+    >>> node = MockNode(0.0, "seed", neigh)
+    >>> _, kind = _mix_epi_with_neighbors(node, 0.5, Glyph.EN)
+    >>> round(node.EPI, 2), kind
+    (0.5, 'wave')
+    """
+    default_kind = (
+        default_glyph.value if isinstance(default_glyph, Glyph) else str(default_glyph)
+    )
+    epi = node.EPI
+    neigh, epi_bar = get_neighbor_epi(node)
+
+    if not neigh:
+        node.epi_kind = default_kind
+        return epi, default_kind
+
+    dominant, best_abs = _determine_dominant(neigh, default_kind)
+    new_epi = (1 - mix) * epi + mix * epi_bar
+    _set_epi_with_boundary_check(node, new_epi)
+    final = dominant if best_abs > abs(new_epi) else node.epi_kind
+    if not final:
+        final = default_kind
+    node.epi_kind = final
+    return epi_bar, final
+
+
+def _op_AL(node: NodeProtocol, gf: GlyphFactors) -> None:  # AL — Emission
+    """Amplify the node EPI via the Emission glyph.
+
+    Emission injects additional coherence into the node by boosting its EPI
+    without touching νf, ΔNFR, or phase. The boost amplitude is controlled by
+    ``AL_boost``.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose EPI is increased.
+    gf : GlyphFactors
+        Factor mapping used to resolve ``AL_boost``.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, epi):
+    ...         self.EPI = epi
+    ...         self.graph = {}
+    >>> node = MockNode(0.8)
+    >>> _op_AL(node, {"AL_boost": 0.2})
+    >>> node.EPI <= 1.0  # Bounded by structural_clip
+    True
+    """
+    f = get_factor(gf, "AL_boost", 0.05)
+    new_epi = node.EPI + f
+    _set_epi_with_boundary_check(node, new_epi)
+
+
+def _op_EN(node: NodeProtocol, gf: GlyphFactors) -> None:  # EN — Reception
+    """Mix the node EPI with the neighbour field via Reception.
+
+    Reception reorganizes the node's EPI towards the neighbourhood mean while
+    choosing a coherent ``epi_kind``. νf, ΔNFR, and phase remain unchanged.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose EPI is being reconciled.
+    gf : GlyphFactors
+        Source of the ``EN_mix`` blending coefficient.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, epi, neighbors):
+    ...         self.EPI = epi
+    ...         self.epi_kind = "seed"
+    ...         self.graph = {}
+    ...         self._neighbors = neighbors
+    ...     def neighbors(self):
+    ...         return self._neighbors
+    >>> neigh = [MockNode(1.0, []), MockNode(0.0, [])]
+    >>> node = MockNode(0.4, neigh)
+    >>> _op_EN(node, {"EN_mix": 0.5})
+    >>> round(node.EPI, 2)
+    0.7
+    """
+    mix = get_factor(gf, "EN_mix", 0.25)
+    _mix_epi_with_neighbors(node, mix, Glyph.EN)
+
+
+def _op_IL(node: NodeProtocol, gf: GlyphFactors) -> None:  # IL — Coherence
+    """Dampen ΔNFR magnitudes through the Coherence glyph.
+
+    Coherence contracts the internal reorganization differential (ΔNFR) while
+    leaving EPI, νf, and phase untouched. The contraction preserves the sign of
+    ΔNFR, increasing structural stability.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose ΔNFR is being scaled.
+    gf : GlyphFactors
+        Provides ``IL_dnfr_factor`` controlling the contraction strength.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, dnfr):
+    ...         self.dnfr = dnfr
+    >>> node = MockNode(0.5)
+    >>> _op_IL(node, {"IL_dnfr_factor": 0.2})
+    >>> node.dnfr
+    0.1
+    """
+    factor = get_factor(gf, "IL_dnfr_factor", 0.7)
+    node.dnfr = factor * getattr(node, "dnfr", 0.0)
+
+
+def _op_OZ(node: NodeProtocol, gf: GlyphFactors) -> None:  # OZ — Dissonance
+    """Excite ΔNFR through the Dissonance glyph.
+
+    Dissonance amplifies ΔNFR or injects jitter, testing the node's stability.
+    EPI, νf, and phase remain unaffected while ΔNFR grows to trigger potential
+    bifurcations.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose ΔNFR is being stressed.
+    gf : GlyphFactors
+        Supplies ``OZ_dnfr_factor`` and optional noise parameters.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, dnfr):
+    ...         self.dnfr = dnfr
+    ...         self.graph = {}
+    >>> node = MockNode(0.2)
+    >>> _op_OZ(node, {"OZ_dnfr_factor": 2.0})
+    >>> node.dnfr
+    0.4
+    """
+    factor = get_factor(gf, "OZ_dnfr_factor", 1.3)
+    dnfr = getattr(node, "dnfr", 0.0)
+    if bool(node.graph.get("OZ_NOISE_MODE", False)):
+        sigma = float(node.graph.get("OZ_SIGMA", 0.1))
+        if sigma <= 0:
+            node.dnfr = dnfr
+            return
+        node.dnfr = dnfr + random_jitter(node, sigma)
+    else:
+        node.dnfr = factor * dnfr if abs(dnfr) > 1e-9 else 0.1
+
+
+def _um_candidate_iter(node: NodeProtocol) -> Iterator[NodeProtocol]:
+    sample_ids = node.graph.get("_node_sample")
+    if sample_ids is not None and hasattr(node, "G"):
+        NodeNX = get_nodenx()
+        if NodeNX is None:
+            raise ImportError("NodeNX is unavailable")
+        base = (NodeNX.from_graph(node.G, j) for j in sample_ids)
+    else:
+        base = node.all_nodes()
+    for j in base:
+        same = (j is node) or (getattr(node, "n", None) == getattr(j, "n", None))
+        if same or node.has_edge(j):
+            continue
+        yield j
+
+
+def _um_select_candidates(
+    node: NodeProtocol,
+    candidates: Iterator[NodeProtocol],
+    limit: int,
+    mode: str,
+    th: float,
+) -> list[NodeProtocol]:
+    """Select a subset of ``candidates`` for UM coupling."""
+    rng = make_rng(int(node.graph.get("RANDOM_SEED", 0)), node.offset(), node.G)
+
+    if limit <= 0:
+        return list(candidates)
+
+    if mode == "proximity":
+        return heapq.nsmallest(
+            limit, candidates, key=lambda j: abs(angle_diff(j.theta, th))
+        )
+
+    reservoir = list(islice(candidates, limit))
+    for i, cand in enumerate(candidates, start=limit):
+        j = rng.randint(0, i)
+        if j < limit:
+            reservoir[j] = cand
+
+    if mode == "sample":
+        rng.shuffle(reservoir)
+
+    return reservoir
+
+
+def compute_consensus_phase(phases: list[float]) -> float:
+    """Compute circular mean (consensus phase) from a list of phase angles.
+
+    This function calculates the consensus phase using the circular mean
+    formula: arctan2(mean(sin), mean(cos)). This ensures proper handling
+    of phase wrapping at ±π boundaries.
+
+    Parameters
+    ----------
+    phases : list[float]
+        List of phase angles in radians.
+
+    Returns
+    -------
+    float
+        Consensus phase angle in radians, in the range [-π, π).
+
+    Notes
+    -----
+    The consensus phase represents the central tendency of a set of angular
+    values, accounting for the circular nature of phase space. This is
+    critical for bidirectional phase synchronization in the UM operator.
+
+    Examples
+    --------
+    >>> import math
+    >>> phases = [0.0, math.pi/2, math.pi]
+    >>> result = compute_consensus_phase(phases)
+    >>> -math.pi <= result < math.pi
+    True
+    """
+    if not phases:
+        return 0.0
+
+    cos_sum = sum(math.cos(ph) for ph in phases)
+    sin_sum = sum(math.sin(ph) for ph in phases)
+    return math.atan2(sin_sum, cos_sum)
+
+
+def _op_UM(node: NodeProtocol, gf: GlyphFactors) -> None:  # UM — Coupling
+    """Align node phase and frequency with neighbours and optionally create links.
+
+    Coupling shifts the node phase ``theta`` towards the neighbour mean while
+    respecting νf and EPI. When bidirectional mode is enabled (default), both
+    the node and its neighbors synchronize their phases mutually. Additionally,
+    structural frequency (νf) synchronization causes coupled nodes to converge
+    their reorganization rates. Coupling also reduces ΔNFR through mutual
+    stabilization, decreasing reorganization pressure proportional to phase
+    alignment strength. When functional links are enabled it may add edges
+    based on combined phase, EPI, and sense-index similarity.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose phase and frequency are being synchronised.
+    gf : GlyphFactors
+        Provides ``UM_theta_push``, ``UM_vf_sync``, ``UM_dnfr_reduction`` and
+        optional selection parameters.
+
+    Notes
+    -----
+    Bidirectional synchronization (UM_BIDIRECTIONAL=True, default) implements
+    the canonical TNFR requirement φᵢ(t) ≈ φⱼ(t) by mutually adjusting phases
+    of both the node and its neighbors towards a consensus phase. This ensures
+    true coupling as defined in the theory.
+
+    Structural frequency synchronization (UM_SYNC_VF=True, default) implements
+    the TNFR requirement that coupling synchronizes not only phases but also
+    structural frequencies (νf). This enables coupled nodes to converge their
+    reorganization rates, which is essential for sustained resonance and coherent
+    network evolution as described by the nodal equation: ∂EPI/∂t = νf · ΔNFR(t).
+
+    ΔNFR stabilization (UM_STABILIZE_DNFR=True, default) implements the canonical
+    effect where coupling reduces reorganization pressure through mutual stabilization.
+    The reduction is proportional to phase alignment: well-coupled nodes (high phase
+    alignment) experience stronger ΔNFR reduction, promoting structural coherence.
+
+    Legacy unidirectional mode (UM_BIDIRECTIONAL=False) only adjusts the node's
+    phase towards its neighbors, preserving backward compatibility.
+
+    Examples
+    --------
+    >>> import math
+    >>> class MockNode:
+    ...     def __init__(self, theta, neighbors):
+    ...         self.theta = theta
+    ...         self.EPI = 1.0
+    ...         self.Si = 0.5
+    ...         self.graph = {}
+    ...         self._neighbors = neighbors
+    ...     def neighbors(self):
+    ...         return self._neighbors
+    ...     def offset(self):
+    ...         return 0
+    ...     def all_nodes(self):
+    ...         return []
+    ...     def has_edge(self, _):
+    ...         return False
+    ...     def add_edge(self, *_):
+    ...         raise AssertionError("not used in example")
+    >>> neighbor = MockNode(math.pi / 2, [])
+    >>> node = MockNode(0.0, [neighbor])
+    >>> _op_UM(node, {"UM_theta_push": 0.5})
+    >>> round(node.theta, 2)
+    0.79
+    """
+    k = get_factor(gf, "UM_theta_push", 0.25)
+    k_vf = get_factor(gf, "UM_vf_sync", 0.10)
+    th_i = node.theta
+
+    # Check if bidirectional synchronization is enabled (default: True)
+    bidirectional = bool(node.graph.get("UM_BIDIRECTIONAL", True))
+
+    if bidirectional:
+        # Bidirectional mode: mutually synchronize node and neighbors
+        neighbor_ids = list(node.neighbors())
+        if neighbor_ids:
+            # Get NodeNX wrapper for accessing neighbor attributes
+            NodeNX = get_nodenx()
+            if NodeNX is None or not hasattr(node, "G"):
+                # Fallback to unidirectional if NodeNX unavailable
+                thL = neighbor_phase_mean(node)
+                d = angle_diff(thL, th_i)
+                node.theta = th_i + k * d
+            else:
+                # Wrap neighbor IDs to access theta attribute
+                neighbors = [NodeNX.from_graph(node.G, nid) for nid in neighbor_ids]
+
+                # Collect all phases (node + neighbors)
+                phases = [th_i] + [n.theta for n in neighbors]
+                target_phase = compute_consensus_phase(phases)
+
+                # Adjust node phase towards consensus
+                node.theta = th_i + k * angle_diff(target_phase, th_i)
+
+                # Adjust neighbor phases towards consensus
+                for neighbor in neighbors:
+                    th_j = neighbor.theta
+                    neighbor.theta = th_j + k * angle_diff(target_phase, th_j)
+    else:
+        # Legacy unidirectional mode: only adjust node towards neighbors
+        thL = neighbor_phase_mean(node)
+        d = angle_diff(thL, th_i)
+        node.theta = th_i + k * d
+
+    # Structural frequency (νf) synchronization
+    # According to TNFR theory, coupling synchronizes both phase and frequency
+    sync_vf = bool(node.graph.get("UM_SYNC_VF", True))
+    if sync_vf:
+        neighbor_ids = list(node.neighbors())
+        if neighbor_ids and hasattr(node, "G"):
+            # Canonical access to vf through alias system
+            vf_i = node.vf
+            vf_neighbors = [
+                get_attr(node.G.nodes[nid], ALIAS_VF, 0.0) for nid in neighbor_ids
+            ]
+
+            if vf_neighbors:
+                vf_mean = sum(vf_neighbors) / len(vf_neighbors)
+
+                # Gradual convergence towards mean (similar to phase sync)
+                node.vf = vf_i + k_vf * (vf_mean - vf_i)
+
+    # ΔNFR reduction by mutual stabilization
+    # Coupling produces a stabilizing effect that reduces reorganization pressure
+    stabilize_dnfr = bool(node.graph.get("UM_STABILIZE_DNFR", True))
+
+    if stabilize_dnfr:
+        k_dnfr = get_factor(gf, "UM_dnfr_reduction", 0.15)
+
+        # Calculate compatibility with neighbors based on phase alignment
+        neighbor_ids = list(node.neighbors())
+        if neighbor_ids:
+            # Get NodeNX wrapper for accessing neighbor attributes
+            NodeNX = get_nodenx()
+            if NodeNX is not None and hasattr(node, "G"):
+                neighbors = [NodeNX.from_graph(node.G, nid) for nid in neighbor_ids]
+
+                # Compute phase alignments with each neighbor
+                phase_alignments = []
+                # Compute phase alignment using canonical formula
+                from ..metrics.phase_compatibility import compute_phase_coupling_strength
+                
+                for neighbor in neighbors:
+                    alignment = compute_phase_coupling_strength(node.theta, neighbor.theta)
+                    phase_alignments.append(alignment)
+
+                # Mean alignment represents coupling strength
+                mean_alignment = sum(phase_alignments) / len(phase_alignments)
+
+                # Reduce ΔNFR proportionally to coupling strength
+                # reduction_factor < 1.0 when well-coupled (high alignment)
+                reduction_factor = 1.0 - (k_dnfr * mean_alignment)
+                node.dnfr = node.dnfr * reduction_factor
+
+    if bool(node.graph.get("UM_FUNCTIONAL_LINKS", True)):
+        thr = float(
+            node.graph.get(
+                "UM_COMPAT_THRESHOLD",
+                DEFAULTS.get("UM_COMPAT_THRESHOLD", 0.75),
+            )
+        )
+        epi_i = node.EPI
+        si_i = node.Si
+
+        limit = int(node.graph.get("UM_CANDIDATE_COUNT", 0))
+        mode = str(node.graph.get("UM_CANDIDATE_MODE", "sample")).lower()
+        candidates = _um_select_candidates(
+            node, _um_candidate_iter(node), limit, mode, th_i
+        )
+
+        for j in candidates:
+            # Use canonical phase coupling strength formula
+            from ..metrics.phase_compatibility import compute_phase_coupling_strength
+            
+            phase_coupling = compute_phase_coupling_strength(th_i, j.theta)
+            
+            epi_j = j.EPI
+            si_j = j.Si
+            epi_sim = 1.0 - abs(epi_i - epi_j) / (abs(epi_i) + abs(epi_j) + 1e-9)
+            si_sim = 1.0 - abs(si_i - si_j)
+            # Compatibility combines phase coupling (50%), EPI similarity (25%), Si similarity (25%)
+            compat = phase_coupling * 0.5 + 0.25 * epi_sim + 0.25 * si_sim
+            if compat >= thr:
+                node.add_edge(j, compat)
+
+
+def _op_RA(node: NodeProtocol, gf: GlyphFactors) -> None:  # RA — Resonance
+    """Propagate coherence through resonance with νf amplification.
+
+    Resonance (RA) propagates EPI along existing couplings while amplifying
+    the structural frequency (νf) to reflect network coherence propagation.
+    According to TNFR theory, RA creates "resonant cascades" where coherence
+    amplifies across the network, increasing collective νf and global C(t).
+
+    **Canonical Effects (always active):**
+
+    - **EPI Propagation**: Diffuses EPI to neighbors (identity-preserving)
+    - **νf Amplification**: Increases structural frequency when propagating coherence
+    - **Phase Alignment**: Strengthens phase synchrony across propagation path
+    - **Network C(t)**: Contributes to global coherence increase
+    - **Identity Preservation**: Maintains structural identity during propagation
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node harmonising with its neighbourhood.
+    gf : GlyphFactors
+        Provides ``RA_epi_diff`` (mixing coefficient, default 0.15),
+        ``RA_vf_amplification`` (νf boost factor, default 0.05), and
+        ``RA_phase_coupling`` (phase alignment factor, default 0.10).
+
+    Notes
+    -----
+    **νf Amplification (Canonical)**: When neighbors have coherence (|epi_bar| > 1e-9),
+    node.vf is multiplied by (1.0 + RA_vf_amplification). This reflects
+    the canonical TNFR property that resonance amplifies collective νf.
+    This is NOT optional - it is a fundamental property of resonance per TNFR theory.
+
+    **Phase Alignment Strengthening (Canonical)**: RA strengthens phase alignment
+    with neighbors by applying a small phase correction toward the network mean.
+    This ensures that "Phase alignment: Strengthens across propagation path" as
+    stated in the theoretical foundations. Uses existing phase utility functions
+    to avoid code duplication.
+
+    **Network Coherence Tracking (Optional)**: If ``TRACK_NETWORK_COHERENCE`` is enabled,
+    global C(t) is measured before/after RA application to quantify network-level
+    coherence increase.
+
+    **Identity Preservation (Canonical)**: EPI structure (kind and sign) are preserved
+    during propagation to ensure structural identity is maintained as required by theory.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, epi, neighbors):
+    ...         self.EPI = epi
+    ...         self.epi_kind = "seed"
+    ...         self.vf = 1.0
+    ...         self.theta = 0.0
+    ...         self.graph = {}
+    ...         self._neighbors = neighbors
+    ...     def neighbors(self):
+    ...         return self._neighbors
+    >>> neighbor = MockNode(1.0, [])
+    >>> neighbor.theta = 0.1
+    >>> node = MockNode(0.2, [neighbor])
+    >>> _op_RA(node, {"RA_epi_diff": 0.25, "RA_vf_amplification": 0.05})
+    >>> round(node.EPI, 2)
+    0.4
+    >>> node.vf  # Amplified due to neighbor coherence (canonical effect)
+    1.05
+    """
+    # Get configuration factors
+    diff = get_factor(gf, "RA_epi_diff", 0.15)
+    vf_boost = get_factor(gf, "RA_vf_amplification", 0.05)
+    phase_coupling = get_factor(
+        gf, "RA_phase_coupling", 0.10
+    )  # Canonical phase strengthening
+
+    # Track network C(t) before RA if enabled (optional telemetry)
+    track_coherence = bool(node.graph.get("TRACK_NETWORK_COHERENCE", False))
+    c_before = None
+    if track_coherence and hasattr(node, "G"):
+        try:
+            from ..metrics.coherence import compute_network_coherence
+
+            c_before = compute_network_coherence(node.G)
+            if "_ra_c_tracking" not in node.graph:
+                node.graph["_ra_c_tracking"] = []
+        except ImportError:
+            pass  # Metrics module not available
+
+    # Capture state before for metrics
+    vf_before = node.vf
+    epi_before = node.EPI
+    kind_before = node.epi_kind
+    theta_before = node.theta if hasattr(node, "theta") else None
+
+    # EPI diffusion (existing behavior)
+    neigh, epi_bar = get_neighbor_epi(node)
+    epi_bar_result, kind_result = _mix_epi_with_neighbors(node, diff, Glyph.RA)
+
+    # CANONICAL EFFECT 1: νf amplification through resonance
+    # This is always active - it's a fundamental property of resonance per TNFR theory
+    # Only amplify if neighbors have coherence to propagate
+    if abs(epi_bar_result) > 1e-9 and len(neigh) > 0:
+        node.vf *= 1.0 + vf_boost
+
+    # CANONICAL EFFECT 2: Phase alignment strengthening
+    # Per theory: "Phase alignment: Strengthens across propagation path"
+    # Uses existing phase locking logic from IL operator (avoid duplication)
+    phase_strengthened = False
+    if len(neigh) > 0 and hasattr(node, "theta") and hasattr(node, "G"):
+        try:
+            # Use existing phase locking utility from IL operator
+            from ..alias import get_attr
+            from ..constants.aliases import ALIAS_THETA
+            import cmath
+            import math
+
+            # Get neighbor phases using existing utilities
+            neighbor_phases = []
+            for n in neigh:
+                try:
+                    theta_n = float(get_attr(n, ALIAS_THETA, 0.0))
+                    neighbor_phases.append(theta_n)
+                except (KeyError, ValueError, TypeError):
+                    continue
+
+            if neighbor_phases:
+                # Circular mean using the same method as in phase_coherence.py
+                complex_phases = [cmath.exp(1j * theta) for theta in neighbor_phases]
+                mean_real = sum(z.real for z in complex_phases) / len(complex_phases)
+                mean_imag = sum(z.imag for z in complex_phases) / len(complex_phases)
+                mean_complex = complex(mean_real, mean_imag)
+                mean_phase = cmath.phase(mean_complex)
+
+                # Ensure positive phase [0, 2π]
+                if mean_phase < 0:
+                    mean_phase += 2 * math.pi
+
+                # Calculate phase difference (shortest arc)
+                delta_theta = mean_phase - node.theta
+                if delta_theta > math.pi:
+                    delta_theta -= 2 * math.pi
+                elif delta_theta < -math.pi:
+                    delta_theta += 2 * math.pi
+
+                # Apply phase strengthening (move toward network mean)
+                # Same approach as IL operator phase locking
+                node.theta = node.theta + phase_coupling * delta_theta
+
+                # Normalize to [0, 2π]
+                node.theta = node.theta % (2 * math.pi)
+                phase_strengthened = True
+        except (AttributeError, ImportError):
+            pass  # Phase alignment not possible in this context
+
+    # Track identity preservation (canonical validation)
+    identity_preserved = (
+        kind_result == kind_before or kind_result == Glyph.RA.value
+    ) and (
+        float(epi_before) * float(node.EPI) >= 0
+    )  # Sign preserved
+
+    # Collect propagation metrics if enabled (optional telemetry)
+    collect_metrics = bool(node.graph.get("COLLECT_RA_METRICS", False))
+    if collect_metrics:
+        metrics = {
+            "operator": "RA",
+            "epi_propagated": epi_bar_result,
+            "vf_amplification": node.vf / vf_before if vf_before > 0 else 1.0,
+            "neighbors_influenced": len(neigh),
+            "identity_preserved": identity_preserved,
+            "epi_before": epi_before,
+            "epi_after": float(node.EPI),
+            "vf_before": vf_before,
+            "vf_after": node.vf,
+            "phase_before": theta_before,
+            "phase_after": node.theta if hasattr(node, "theta") else None,
+            "phase_alignment_strengthened": phase_strengthened,
+        }
+        if "ra_metrics" not in node.graph:
+            node.graph["ra_metrics"] = []
+        node.graph["ra_metrics"].append(metrics)
+
+    # Track network C(t) after RA if enabled (optional telemetry)
+    if track_coherence and c_before is not None and hasattr(node, "G"):
+        try:
+            from ..metrics.coherence import compute_network_coherence
+
+            c_after = compute_network_coherence(node.G)
+            node.graph["_ra_c_tracking"].append(
+                {
+                    "node": getattr(node, "n", None),
+                    "c_before": c_before,
+                    "c_after": c_after,
+                    "c_delta": c_after - c_before,
+                }
+            )
+        except ImportError:
+            pass
+
+
+def _op_SHA(node: NodeProtocol, gf: GlyphFactors) -> None:  # SHA — Silence
+    """Reduce νf while preserving EPI, ΔNFR, and phase.
+
+    Silence decelerates a node by scaling νf (structural frequency) towards
+    stillness. EPI, ΔNFR, and phase remain unchanged, signalling a temporary
+    suspension of structural evolution.
+
+    **TNFR Canonical Behavior:**
+
+    According to the nodal equation ∂EPI/∂t = νf · ΔNFR(t), reducing νf → νf_min ≈ 0
+    causes structural evolution to freeze (∂EPI/∂t → 0) regardless of ΔNFR magnitude.
+    This implements **structural silence** - a state where the node's form (EPI) is
+    preserved intact despite external pressures, enabling memory consolidation and
+    protective latency.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose νf is being attenuated.
+    gf : GlyphFactors
+        Provides ``SHA_vf_factor`` to scale νf (default 0.85 for gradual reduction).
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, vf):
+    ...         self.vf = vf
+    >>> node = MockNode(1.0)
+    >>> _op_SHA(node, {"SHA_vf_factor": 0.5})
+    >>> node.vf
+    0.5
+    """
+    factor = get_factor(gf, "SHA_vf_factor", 0.85)
+    # Canonical SHA effect: reduce structural frequency toward zero
+    # This implements: νf → νf_min ≈ 0 ⇒ ∂EPI/∂t → 0 (structural preservation)
+    node.vf = factor * node.vf
+
+
+factor_val = 1.05  # Conservative scale prevents EPI overflow near boundaries
+factor_nul = 0.85
+_SCALE_FACTORS = {Glyph.VAL: factor_val, Glyph.NUL: factor_nul}
+
+
+def _set_epi_with_boundary_check(
+    node: NodeProtocol, new_epi: float, *, apply_clip: bool = True
+) -> None:
+    """Canonical EPI assignment with structural boundary preservation.
+
+    This is the unified function all operators should use when modifying EPI
+    to ensure structural boundaries are respected. Provides single point of
+    enforcement for TNFR canonical invariant: EPI ∈ [EPI_MIN, EPI_MAX].
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose EPI is being updated
+    new_epi : float
+        New EPI value to assign
+    apply_clip : bool, default True
+        If True, applies structural_clip to enforce boundaries.
+        If False, assigns value directly (use only when boundaries
+        are known to be satisfied, e.g., from edge-aware pre-computation).
+
+    Notes
+    -----
+    TNFR Principle: This function embodies the canonical invariant that EPI
+    must remain within structural boundaries. All operator EPI modifications
+    should flow through this function to maintain coherence.
+
+    The function uses the graph-level configuration for EPI_MIN, EPI_MAX,
+    and CLIP_MODE to ensure consistent boundary enforcement across all operators.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, epi):
+    ...         self.EPI = epi
+    ...         self.graph = {"EPI_MAX": 1.0, "EPI_MIN": -1.0}
+    >>> node = MockNode(0.5)
+    >>> _set_epi_with_boundary_check(node, 1.2)  # Will be clipped to 1.0
+    >>> float(node.EPI)
+    1.0
+    """
+    from ..dynamics.structural_clip import structural_clip
+
+    if not apply_clip:
+        node.EPI = new_epi
+        return
+
+    # Ensure new_epi is float (in case it's a BEPI or other structure)
+    new_epi_float = float(new_epi)
+
+    # Get boundary configuration from graph (with defensive fallback)
+    graph_attrs = getattr(node, "graph", {})
+    epi_min = float(graph_attrs.get("EPI_MIN", DEFAULTS.get("EPI_MIN", -1.0)))
+    epi_max = float(graph_attrs.get("EPI_MAX", DEFAULTS.get("EPI_MAX", 1.0)))
+    clip_mode_str = str(graph_attrs.get("CLIP_MODE", "hard"))
+
+    # Validate clip mode
+    if clip_mode_str not in ("hard", "soft"):
+        clip_mode_str = "hard"
+
+    # Apply structural boundary preservation
+    clipped_epi = structural_clip(
+        new_epi_float,
+        lo=epi_min,
+        hi=epi_max,
+        mode=clip_mode_str,  # type: ignore[arg-type]
+        record_stats=False,
+    )
+
+    node.EPI = clipped_epi
+
+
+def _compute_val_edge_aware_scale(
+    epi_current: float, scale: float, epi_max: float, epsilon: float
+) -> float:
+    """Compute edge-aware scale factor for VAL (Expansion) operator.
+
+    Adapts the expansion scale to prevent EPI overflow beyond EPI_MAX.
+    When EPI is near the upper boundary, the effective scale is reduced
+    to ensure EPI * scale_eff <= EPI_MAX.
+
+    Parameters
+    ----------
+    epi_current : float
+        Current EPI value
+    scale : float
+        Desired expansion scale factor (e.g., VAL_scale = 1.05)
+    epi_max : float
+        Upper EPI boundary (typically 1.0)
+    epsilon : float
+        Small value to prevent division by zero (e.g., 1e-12)
+
+    Returns
+    -------
+    float
+        Effective scale factor, adapted to respect EPI_MAX boundary
+
+    Notes
+    -----
+    TNFR Principle: This implements "resonance to the edge" - expansion
+    scales adaptively to explore volume while respecting structural envelope.
+    The adaptation is a dynamic compatibility check, not a fixed constant.
+
+    Examples
+    --------
+    >>> # Normal case: EPI far from boundary
+    >>> _compute_val_edge_aware_scale(0.5, 1.05, 1.0, 1e-12)
+    1.05
+
+    >>> # Edge case: EPI near boundary, scale adapts
+    >>> scale = _compute_val_edge_aware_scale(0.96, 1.05, 1.0, 1e-12)
+    >>> abs(scale - 1.0417) < 0.001  # Roughly 1.0/0.96
+    True
+    """
+    abs_epi = abs(epi_current)
+    if abs_epi < epsilon:
+        # EPI near zero, full scale can be applied safely
+        return scale
+
+    # Compute maximum safe scale that keeps EPI within bounds
+    max_safe_scale = epi_max / abs_epi
+
+    # Return the minimum of desired scale and safe scale
+    return min(scale, max_safe_scale)
+
+
+def _compute_nul_edge_aware_scale(
+    epi_current: float, scale: float, epi_min: float, epsilon: float
+) -> float:
+    """Compute edge-aware scale factor for NUL (Contraction) operator.
+
+    Adapts the contraction scale to prevent EPI underflow below EPI_MIN.
+
+    Parameters
+    ----------
+    epi_current : float
+        Current EPI value
+    scale : float
+        Desired contraction scale factor (e.g., NUL_scale = 0.85)
+    epi_min : float
+        Lower EPI boundary (typically -1.0)
+    epsilon : float
+        Small value to prevent division by zero (e.g., 1e-12)
+
+    Returns
+    -------
+    float
+        Effective scale factor, adapted to respect EPI_MIN boundary
+
+    Notes
+    -----
+    TNFR Principle: Contraction concentrates structure toward core while
+    maintaining coherence.
+
+    For typical NUL_scale < 1.0, contraction naturally moves EPI toward zero
+    (the center), which is always safe regardless of whether EPI is positive
+    or negative. Edge-awareness is only needed if scale could somehow push
+    EPI beyond boundaries.
+
+    In practice, with NUL_scale = 0.85 < 1.0:
+    - Positive EPI contracts toward zero: safe
+    - Negative EPI contracts toward zero: safe
+
+    Edge-awareness is provided for completeness and future extensibility.
+
+    Examples
+    --------
+    >>> # Normal contraction (always safe with scale < 1.0)
+    >>> _compute_nul_edge_aware_scale(0.5, 0.85, -1.0, 1e-12)
+    0.85
+    >>> _compute_nul_edge_aware_scale(-0.5, 0.85, -1.0, 1e-12)
+    0.85
+    """
+    # With NUL_scale < 1.0, contraction moves toward zero (always safe)
+    # No adaptation needed in typical case
+    return scale
+
+
+def _op_scale(node: NodeProtocol, factor: float) -> None:
+    """Scale νf with the provided factor.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose νf is being updated.
+    factor : float
+        Multiplicative change applied to νf.
+    """
+    node.vf *= factor
+
+
+def _make_scale_op(glyph: Glyph) -> GlyphOperation:
+    def _op(node: NodeProtocol, gf: GlyphFactors) -> None:
+        key = "VAL_scale" if glyph is Glyph.VAL else "NUL_scale"
+        default = _SCALE_FACTORS[glyph]
+        factor = get_factor(gf, key, default)
+
+        # Always scale νf (existing behavior)
+        _op_scale(node, factor)
+
+        # NUL canonical ΔNFR densification (implements structural pressure concentration)
+        if glyph is Glyph.NUL:
+            # Volume reduction: V' = V · scale_factor (where scale_factor < 1.0)
+            # Density increase: ρ_ΔNFR = ΔNFR / V' = ΔNFR / (V · scale_factor)
+            # Result: ΔNFR' = ΔNFR · densification_factor
+            #
+            # Physics: When volume contracts by factor λ < 1, structural pressure
+            # concentrates by factor 1/λ > 1. For NUL_scale = 0.85, densification ≈ 1.176
+            #
+            # Default densification_factor from config (typically 1.3-1.5) provides
+            # additional canonical amplification beyond geometric 1/λ to account for
+            # nonlinear structural effects at smaller scales.
+            densification_key = "NUL_densification_factor"
+            densification_default = 1.35  # Canonical default: moderate amplification
+            densification_factor = get_factor(gf, densification_key, densification_default)
+            
+            # Apply densification to ΔNFR (use lowercase dnfr for NodeProtocol)
+            current_dnfr = node.dnfr
+            node.dnfr = current_dnfr * densification_factor
+            
+            # Record densification telemetry for traceability
+            telemetry = node.graph.setdefault("nul_densification_log", [])
+            telemetry.append(
+                {
+                    "dnfr_before": current_dnfr,
+                    "dnfr_after": float(node.dnfr),
+                    "densification_factor": densification_factor,
+                    "contraction_scale": factor,
+                }
+            )
+
+        # Edge-aware EPI scaling (new behavior) if enabled
+        edge_aware_enabled = bool(
+            node.graph.get(
+                "EDGE_AWARE_ENABLED", DEFAULTS.get("EDGE_AWARE_ENABLED", True)
+            )
+        )
+
+        if edge_aware_enabled:
+            epsilon = float(
+                node.graph.get(
+                    "EDGE_AWARE_EPSILON", DEFAULTS.get("EDGE_AWARE_EPSILON", 1e-12)
+                )
+            )
+            epi_min = float(node.graph.get("EPI_MIN", DEFAULTS.get("EPI_MIN", -1.0)))
+            epi_max = float(node.graph.get("EPI_MAX", DEFAULTS.get("EPI_MAX", 1.0)))
+
+            epi_current = node.EPI
+
+            # Compute edge-aware scale factor
+            if glyph is Glyph.VAL:
+                scale_eff = _compute_val_edge_aware_scale(
+                    epi_current, factor, epi_max, epsilon
+                )
+            else:  # Glyph.NUL
+                scale_eff = _compute_nul_edge_aware_scale(
+                    epi_current, factor, epi_min, epsilon
+                )
+
+            # Apply edge-aware EPI scaling with boundary check
+            # Edge-aware already computed safe scale, but use unified function
+            # for consistency (with apply_clip=True as safety net)
+            new_epi = epi_current * scale_eff
+            _set_epi_with_boundary_check(node, new_epi, apply_clip=True)
+
+            # Record telemetry if scale was adapted
+            if abs(scale_eff - factor) > epsilon:
+                telemetry = node.graph.setdefault("edge_aware_interventions", [])
+                telemetry.append(
+                    {
+                        "glyph": glyph.name if hasattr(glyph, "name") else str(glyph),
+                        "epi_before": epi_current,
+                        "epi_after": float(
+                            node.EPI
+                        ),  # Get actual value after boundary check
+                        "scale_requested": factor,
+                        "scale_effective": scale_eff,
+                        "adapted": True,
+                    }
+                )
+
+    _op.__doc__ = """{} glyph scales νf and EPI with edge-aware adaptation.
+
+        VAL (expansion) increases νf and EPI, whereas NUL (contraction) decreases them.
+        Edge-aware scaling adapts the scale factor near EPI boundaries to prevent
+        overflow/underflow, maintaining structural coherence within [-1.0, 1.0].
+
+        When EDGE_AWARE_ENABLED is True (default), the effective scale is computed as:
+        - VAL: scale_eff = min(VAL_scale, EPI_MAX / |EPI_current|)
+        - NUL: scale_eff = min(NUL_scale, |EPI_MIN| / |EPI_current|) for negative EPI
+
+        This implements TNFR principle: "resonance to the edge" without breaking
+        the structural envelope. Telemetry records adaptation events.
+
+        Parameters
+        ----------
+        node : NodeProtocol
+            Node whose νf and EPI are updated.
+        gf : GlyphFactors
+            Provides the respective scale factor (``VAL_scale`` or
+            ``NUL_scale``).
+
+        Examples
+        --------
+        >>> class MockNode:
+        ...     def __init__(self, vf, epi):
+        ...         self.vf = vf
+        ...         self.EPI = epi
+        ...         self.graph = {{"EDGE_AWARE_ENABLED": True, "EPI_MAX": 1.0}}
+        >>> node = MockNode(1.0, 0.96)
+        >>> op = _make_scale_op(Glyph.VAL)
+        >>> op(node, {{"VAL_scale": 1.05}})
+        >>> node.vf  # νf scaled normally
+        1.05
+        >>> node.EPI <= 1.0  # EPI kept within bounds
+        True
+        """.format(
+        glyph.name
+    )
+    return _op
+
+
+def _op_THOL(node: NodeProtocol, gf: GlyphFactors) -> None:  # THOL — Self-organization
+    """Inject curvature from ``d2EPI`` into ΔNFR to trigger self-organization.
+
+    The glyph keeps EPI, νf, and phase fixed while increasing ΔNFR according to
+    the second derivative of EPI, accelerating structural rearrangement.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node contributing ``d2EPI`` to ΔNFR.
+    gf : GlyphFactors
+        Source of the ``THOL_accel`` multiplier.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, dnfr, curvature):
+    ...         self.dnfr = dnfr
+    ...         self.d2EPI = curvature
+    >>> node = MockNode(0.1, 0.5)
+    >>> _op_THOL(node, {"THOL_accel": 0.2})
+    >>> node.dnfr
+    0.2
+    """
+    a = get_factor(gf, "THOL_accel", 0.10)
+    node.dnfr = node.dnfr + a * getattr(node, "d2EPI", 0.0)
+
+
+def _op_ZHIR(node: NodeProtocol, gf: GlyphFactors) -> None:  # ZHIR — Mutation
+    """Apply canonical phase transformation θ → θ' based on structural dynamics.
+
+    ZHIR (Mutation) implements the canonical TNFR phase transformation that depends on
+    the node's reorganization state (ΔNFR). Unlike a fixed rotation, the transformation
+    magnitude and direction are determined by structural pressure, implementing the
+    physics: θ → θ' when ΔEPI/Δt > ξ (AGENTS.md §11, TNFR.pdf §2.2.11).
+
+    **Canonical Behavior**:
+    - Direction: Based on ΔNFR sign (positive → forward phase, negative → backward)
+    - Magnitude: Proportional to theta_shift_factor and |ΔNFR|
+    - Regime detection: Identifies quadrant crossings (π/2 boundaries)
+    - Deterministic: Same seed produces same transformation
+
+    The transformation preserves structural identity (epi_kind) while shifting the
+    operational regime, enabling adaptation without losing coherence.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose phase is transformed based on its structural state.
+    gf : GlyphFactors
+        Supplies ``ZHIR_theta_shift_factor`` (default: 0.3) controlling transformation
+        magnitude. Can override with explicit ``ZHIR_theta_shift`` for fixed rotation.
+
+    Examples
+    --------
+    >>> import math
+    >>> class MockNode:
+    ...     def __init__(self, theta, dnfr):
+    ...         self.theta = theta
+    ...         self.dnfr = dnfr
+    ...         self.graph = {}
+    >>> # Positive ΔNFR → forward phase shift
+    >>> node = MockNode(0.0, 0.5)
+    >>> _op_ZHIR(node, {"ZHIR_theta_shift_factor": 0.3})
+    >>> 0.2 < node.theta < 0.3  # ~π/4 * 0.3 ≈ 0.24
+    True
+    >>> # Negative ΔNFR → backward phase shift
+    >>> node2 = MockNode(math.pi, -0.5)
+    >>> _op_ZHIR(node2, {"ZHIR_theta_shift_factor": 0.3})
+    >>> 2.9 < node2.theta < 3.0  # π - 0.24 ≈ 2.90
+    True
+    >>> # Fixed shift overrides dynamic behavior
+    >>> node3 = MockNode(0.0, 0.5)
+    >>> _op_ZHIR(node3, {"ZHIR_theta_shift": math.pi / 2})
+    >>> round(node3.theta, 2)
+    1.57
+    """
+    # Check for explicit fixed shift (backward compatibility)
+    if "ZHIR_theta_shift" in gf:
+        shift = get_factor(gf, "ZHIR_theta_shift", math.pi / 2)
+        node.theta = node.theta + shift
+        # Store telemetry for fixed shift mode
+        storage = node._glyph_storage()
+        storage["_zhir_theta_shift"] = shift
+        storage["_zhir_fixed_mode"] = True
+        return
+
+    # Canonical transformation: θ → θ' based on ΔNFR
+    theta_before = node.theta
+    dnfr = node.dnfr
+    
+    # Transformation magnitude controlled by factor
+    theta_shift_factor = get_factor(gf, "ZHIR_theta_shift_factor", 0.3)
+    
+    # Direction based on ΔNFR sign (coherent with structural pressure)
+    # Magnitude based on |ΔNFR| (stronger pressure → larger shift)
+    # Base shift is π/4, scaled by factor and ΔNFR
+    base_shift = math.pi / 4
+    shift = theta_shift_factor * math.copysign(1.0, dnfr) * base_shift
+    
+    # Apply transformation with phase wrapping [0, 2π)
+    theta_new = (theta_before + shift) % (2 * math.pi)
+    node.theta = theta_new
+    
+    # Detect regime change (crossing quadrant boundaries)
+    regime_before = int(theta_before // (math.pi / 2))
+    regime_after = int(theta_new // (math.pi / 2))
+    regime_changed = regime_before != regime_after
+    
+    # Store telemetry for metrics collection
+    storage = node._glyph_storage()
+    storage["_zhir_theta_shift"] = shift
+    storage["_zhir_theta_before"] = theta_before
+    storage["_zhir_theta_after"] = theta_new
+    storage["_zhir_regime_changed"] = regime_changed
+    storage["_zhir_regime_before"] = regime_before
+    storage["_zhir_regime_after"] = regime_after
+    storage["_zhir_fixed_mode"] = False
+
+
+def _op_NAV(node: NodeProtocol, gf: GlyphFactors) -> None:  # NAV — Transition
+    """Rebalance ΔNFR towards νf while permitting jitter.
+
+    Transition pulls ΔNFR towards a νf-aligned target, optionally adding jitter
+    to explore nearby states. EPI and phase remain untouched; νf may be used as
+    a reference but is not directly changed.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose ΔNFR is redirected.
+    gf : GlyphFactors
+        Supplies ``NAV_eta`` and ``NAV_jitter`` tuning parameters.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self, dnfr, vf):
+    ...         self.dnfr = dnfr
+    ...         self.vf = vf
+    ...         self.graph = {"NAV_RANDOM": False}
+    >>> node = MockNode(-0.6, 0.4)
+    >>> _op_NAV(node, {"NAV_eta": 0.5, "NAV_jitter": 0.0})
+    >>> round(node.dnfr, 2)
+    -0.1
+    """
+    dnfr = node.dnfr
+    vf = node.vf
+    eta = get_factor(gf, "NAV_eta", 0.5)
+    strict = bool(node.graph.get("NAV_STRICT", False))
+    if strict:
+        base = vf
+    else:
+        sign = 1.0 if dnfr >= 0 else -1.0
+        target = sign * vf
+        base = (1.0 - eta) * dnfr + eta * target
+    j = get_factor(gf, "NAV_jitter", 0.05)
+    if bool(node.graph.get("NAV_RANDOM", True)):
+        jitter = random_jitter(node, j)
+    else:
+        jitter = j * (1 if base >= 0 else -1)
+    node.dnfr = base + jitter
+
+
+def _op_REMESH(
+    node: NodeProtocol, gf: GlyphFactors | None = None
+) -> None:  # REMESH — advisory
+    """Record an advisory requesting network-scale remeshing.
+
+    REMESH does not change node-level EPI, νf, ΔNFR, or phase. Instead it
+    annotates the glyph history so orchestrators can trigger global remesh
+    procedures once the stability conditions are met.
+
+    Parameters
+    ----------
+    node : NodeProtocol
+        Node whose history records the advisory.
+    gf : GlyphFactors, optional
+        Unused but accepted for API symmetry.
+
+    Examples
+    --------
+    >>> class MockNode:
+    ...     def __init__(self):
+    ...         self.graph = {}
+    >>> node = MockNode()
+    >>> _op_REMESH(node)
+    >>> "_remesh_warn_step" in node.graph
+    True
+    """
+    step_idx = glyph_history.current_step_idx(node)
+    last_warn = node.graph.get("_remesh_warn_step", None)
+    if last_warn != step_idx:
+        msg = (
+            "REMESH operates at network scale. Use apply_remesh_if_globally_"
+            "stable(G) or apply_network_remesh(G)."
+        )
+        hist = glyph_history.ensure_history(node)
+        glyph_history.append_metric(
+            hist,
+            "events",
+            ("warn", {"step": step_idx, "node": None, "msg": msg}),
+        )
+        node.graph["_remesh_warn_step"] = step_idx
+    return
+
+
+# -------------------------
+# Dispatcher
+# -------------------------
+
+GLYPH_OPERATIONS: dict[Glyph, GlyphOperation] = {
+    Glyph.AL: _op_AL,
+    Glyph.EN: _op_EN,
+    Glyph.IL: _op_IL,
+    Glyph.OZ: _op_OZ,
+    Glyph.UM: _op_UM,
+    Glyph.RA: _op_RA,
+    Glyph.SHA: _op_SHA,
+    Glyph.VAL: _make_scale_op(Glyph.VAL),
+    Glyph.NUL: _make_scale_op(Glyph.NUL),
+    Glyph.THOL: _op_THOL,
+    Glyph.ZHIR: _op_ZHIR,
+    Glyph.NAV: _op_NAV,
+    Glyph.REMESH: _op_REMESH,
+}
+
+
+def apply_glyph_obj(
+    node: NodeProtocol, glyph: Glyph | str, *, window: int | None = None
+) -> None:
+    """Apply ``glyph`` to an object satisfying :class:`NodeProtocol`."""
+
+    from .grammar import function_name_to_glyph
+    from ..validation.input_validation import ValidationError, validate_glyph
+
+    # Validate glyph parameter
+    try:
+        if not isinstance(glyph, Glyph):
+            validated_glyph = validate_glyph(glyph)
+            glyph = (
+                validated_glyph.value
+                if isinstance(validated_glyph, Glyph)
+                else str(glyph)
+            )
+        else:
+            glyph = glyph.value
+    except ValidationError as e:
+        step_idx = glyph_history.current_step_idx(node)
+        hist = glyph_history.ensure_history(node)
+        glyph_history.append_metric(
+            hist,
+            "events",
+            (
+                "warn",
+                {
+                    "step": step_idx,
+                    "node": getattr(node, "n", None),
+                    "msg": f"invalid glyph: {e}",
+                },
+            ),
+        )
+        raise ValueError(f"invalid glyph: {e}") from e
+
+    # Try direct glyph code first
+    try:
+        g = Glyph(str(glyph))
+    except ValueError:
+        # Try structural function name mapping
+        g = function_name_to_glyph(glyph)
+        if g is None:
+            step_idx = glyph_history.current_step_idx(node)
+            hist = glyph_history.ensure_history(node)
+            glyph_history.append_metric(
+                hist,
+                "events",
+                (
+                    "warn",
+                    {
+                        "step": step_idx,
+                        "node": getattr(node, "n", None),
+                        "msg": f"unknown glyph: {glyph}",
+                    },
+                ),
+            )
+            raise ValueError(f"unknown glyph: {glyph}")
+
+    op = GLYPH_OPERATIONS.get(g)
+    if op is None:
+        raise ValueError(f"glyph has no registered operator: {g}")
+    if window is None:
+        window = int(get_param(node, "GLYPH_HYSTERESIS_WINDOW"))
+    gf = get_glyph_factors(node)
+    op(node, gf)
+    glyph_history.push_glyph(node._glyph_storage(), g.value, window)
+    node.epi_kind = g.value
+
+
+def apply_glyph(
+    G: TNFRGraph, n: NodeId, glyph: Glyph | str, *, window: int | None = None
+) -> None:
+    """Adapter to operate on ``networkx`` graphs."""
+    from ..validation.input_validation import (
+        ValidationError,
+        validate_node_id,
+        validate_tnfr_graph,
+    )
+
+    # Validate graph and node parameters
+    try:
+        validate_tnfr_graph(G)
+        validate_node_id(n)
+    except ValidationError as e:
+        raise ValueError(f"Invalid parameters for apply_glyph: {e}") from e
+
+    NodeNX = get_nodenx()
+    if NodeNX is None:
+        raise ImportError("NodeNX is unavailable")
+    node = NodeNX(G, n)
+    apply_glyph_obj(node, glyph, window=window)