tnfr 4.5.2__py3-none-any.whl → 8.5.0__py3-none-any.whl

tnfr/operators/health_analyzer.py

@@ -0,0 +1,809 @@
+"""Structural health metrics analyzer for TNFR operator sequences.
+
+Provides quantitative assessment of sequence structural quality through
+canonical TNFR metrics: coherence, balance, sustainability, and efficiency.
+"""
+
+from __future__ import annotations
+
+from functools import lru_cache
+from typing import Any, List, Tuple, TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from ..types import TNFRGraph
+
+from ..compat.dataclass import dataclass
+from ..config.operator_names import (
+    COHERENCE,
+    DISSONANCE,
+    RECURSIVITY,
+    RESONANCE,
+    SELF_ORGANIZATION,
+    SILENCE,
+    TRANSITION,
+    DESTABILIZERS,
+    TRANSFORMERS,
+)
+
+__all__ = [
+    "SequenceHealthMetrics",
+    "SequenceHealthAnalyzer",
+]
+
+
+# Operator categories for health analysis
+_STABILIZERS = frozenset({COHERENCE, SELF_ORGANIZATION, SILENCE, RESONANCE})
+_REGENERATORS = frozenset({TRANSITION, RECURSIVITY})  # NAV, REMESH
+
+
+@dataclass
+class SequenceHealthMetrics:
+    """Structural health metrics for a TNFR operator sequence.
+
+    All metrics range from 0.0 (poor) to 1.0 (excellent), measuring different
+    aspects of sequence structural quality according to TNFR principles.
+
+    Attributes
+    ----------
+    coherence_index : float
+        Global sequential flow quality (0.0-1.0). Measures how well operators
+        transition and whether the sequence forms a recognizable pattern.
+    balance_score : float
+        Equilibrium between stabilizers and destabilizers (0.0-1.0). Ideal
+        sequences have balanced structural forces.
+    sustainability_index : float
+        Capacity for long-term maintenance (0.0-1.0). Considers final stabilization,
+        resolved dissonance, and regenerative elements.
+    complexity_efficiency : float
+        Value-to-complexity ratio (0.0-1.0). Penalizes unnecessarily long sequences
+        that don't provide proportional structural value.
+    frequency_harmony : float
+        Structural frequency transition smoothness (0.0-1.0). High when transitions
+        respect νf harmonics.
+    pattern_completeness : float
+        How complete the detected pattern is (0.0-1.0). Full cycles score higher.
+    transition_smoothness : float
+        Quality of operator transitions (0.0-1.0). Measures valid transitions vs
+        total transitions.
+    overall_health : float
+        Composite health index (0.0-1.0). Weighted average of primary metrics.
+    sequence_length : int
+        Number of operators in the sequence.
+    dominant_pattern : str
+        Detected structural pattern type (e.g., "activation", "therapeutic", "unknown").
+    recommendations : List[str]
+        Specific suggestions for improving sequence health.
+    """
+
+    coherence_index: float
+    balance_score: float
+    sustainability_index: float
+    complexity_efficiency: float
+    frequency_harmony: float
+    pattern_completeness: float
+    transition_smoothness: float
+    overall_health: float
+    sequence_length: int
+    dominant_pattern: str
+    recommendations: List[str]
+
+
+class SequenceHealthAnalyzer:
+    """Analyzer for structural health of TNFR operator sequences.
+
+    Evaluates sequences along multiple dimensions to provide quantitative
+    assessment of structural quality, coherence, and sustainability.
+
+    Uses caching to optimize repeated analysis of identical sequences,
+    which is common in pattern exploration and batch validation workflows.
+
+    Examples
+    --------
+    >>> from tnfr.operators.health_analyzer import SequenceHealthAnalyzer
+    >>> analyzer = SequenceHealthAnalyzer()
+    >>> sequence = ["emission", "reception", "coherence", "silence"]
+    >>> health = analyzer.analyze_health(sequence)
+    >>> health.overall_health
+    0.82
+    >>> health.recommendations
+    []
+    """
+
+    def __init__(self) -> None:
+        """Initialize the health analyzer with caching support."""
+        self._recommendations: List[str] = []
+        # Cache for single-pass analysis results keyed by sequence tuple
+        # Using maxsize=128 to avoid unbounded growth while caching common sequences
+        self._analysis_cache = lru_cache(maxsize=128)(self._compute_single_pass)
+
+    def _compute_single_pass(
+        self, sequence_tuple: Tuple[str, ...]
+    ) -> Tuple[int, int, int, int, int, List[Tuple[str, str]]]:
+        """Compute sequence statistics in a single pass for efficiency.
+
+        This method scans the sequence once and extracts all the information
+        needed for the various health metrics, avoiding redundant iterations.
+
+        Parameters
+        ----------
+        sequence_tuple : Tuple[str, ...]
+            Immutable sequence of operators (tuple for hashability in cache).
+
+        Returns
+        -------
+        Tuple containing:
+            - stabilizer_count: int
+            - destabilizer_count: int
+            - transformer_count: int
+            - regenerator_count: int
+            - unique_ops: int
+            - problematic_transitions: List[(op1, op2)] pairs
+
+        Notes
+        -----
+        This function is cached using lru_cache to optimize repeated analysis
+        of identical sequences, which is common in batch validation and
+        pattern exploration workflows.
+        """
+        sequence = list(sequence_tuple)
+
+        # Initialize counters
+        stabilizer_count = 0
+        destabilizer_count = 0
+        transformer_count = 0
+        regenerator_count = 0
+        unique_ops_set = set()
+        problematic_transitions = []
+
+        # Single pass through sequence
+        for i, op in enumerate(sequence):
+            unique_ops_set.add(op)
+
+            # Count operator categories
+            if op in _STABILIZERS:
+                stabilizer_count += 1
+            if op in DESTABILIZERS:
+                destabilizer_count += 1
+            if op in TRANSFORMERS:
+                transformer_count += 1
+            if op in _REGENERATORS:
+                regenerator_count += 1
+
+            # Check transitions
+            if i < len(sequence) - 1:
+                next_op = sequence[i + 1]
+                # Destabilizer → destabilizer is problematic
+                if op in DESTABILIZERS and next_op in DESTABILIZERS:
+                    problematic_transitions.append((op, next_op))
+
+        return (
+            stabilizer_count,
+            destabilizer_count,
+            transformer_count,
+            regenerator_count,
+            len(unique_ops_set),
+            problematic_transitions,
+        )
+
+    def analyze_health(self, sequence: List[str]) -> SequenceHealthMetrics:
+        """Perform complete structural health analysis of a sequence.
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence to analyze (canonical names like "emission", "coherence").
+
+        Returns
+        -------
+        SequenceHealthMetrics
+            Comprehensive health metrics for the sequence.
+
+        Examples
+        --------
+        >>> analyzer = SequenceHealthAnalyzer()
+        >>> health = analyzer.analyze_health(["emission", "reception", "coherence", "silence"])
+        >>> health.coherence_index > 0.7
+        True
+        """
+        self._recommendations = []
+
+        # Use single-pass analysis for efficiency (cached)
+        sequence_tuple = tuple(sequence)
+        analysis = self._analysis_cache(sequence_tuple)
+
+        # Extract results from single-pass analysis
+        (
+            stabilizer_count,
+            destabilizer_count,
+            transformer_count,
+            regenerator_count,
+            unique_count,
+            problematic_transitions,
+        ) = analysis
+
+        coherence = self._calculate_coherence(sequence, problematic_transitions)
+        balance = self._calculate_balance(
+            sequence, stabilizer_count, destabilizer_count
+        )
+        sustainability = self._calculate_sustainability(
+            sequence, stabilizer_count, destabilizer_count, regenerator_count
+        )
+        efficiency = self._calculate_efficiency(sequence, unique_count)
+        frequency = self._calculate_frequency_harmony(sequence)
+        completeness = self._calculate_completeness(
+            sequence, stabilizer_count, destabilizer_count, transformer_count
+        )
+        smoothness = self._calculate_smoothness(sequence, problematic_transitions)
+
+        # Calculate overall health as weighted average
+        # Primary metrics weighted more heavily
+        overall = (
+            coherence * 0.20
+            + balance * 0.20
+            + sustainability * 0.20
+            + efficiency * 0.15
+            + frequency * 0.10
+            + completeness * 0.10
+            + smoothness * 0.05
+        )
+
+        pattern = self._detect_pattern(sequence)
+
+        return SequenceHealthMetrics(
+            coherence_index=coherence,
+            balance_score=balance,
+            sustainability_index=sustainability,
+            complexity_efficiency=efficiency,
+            frequency_harmony=frequency,
+            pattern_completeness=completeness,
+            transition_smoothness=smoothness,
+            overall_health=overall,
+            sequence_length=len(sequence),
+            dominant_pattern=pattern,
+            recommendations=self._recommendations.copy(),
+        )
+
+    def _calculate_coherence(
+        self, sequence: List[str], problematic_transitions: List[Tuple[str, str]]
+    ) -> float:
+        """Calculate coherence index: how well the sequence flows.
+
+        Factors:
+        - Valid transitions between operators
+        - Recognizable pattern structure
+        - Structural closure (proper ending)
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+        problematic_transitions : List[Tuple[str, str]]
+            Pre-computed list of problematic transition pairs
+
+        Returns
+        -------
+        float
+            Coherence score (0.0-1.0)
+        """
+        if not sequence:
+            return 0.0
+
+        # Transition quality: use pre-computed problematic transitions
+        if len(sequence) < 2:
+            transition_quality = 1.0
+        else:
+            total_transitions = len(sequence) - 1
+            # Each problematic transition gets 0.5 penalty
+            penalty = len(problematic_transitions) * 0.5
+            transition_quality = max(0.0, 1.0 - (penalty / total_transitions))
+
+        # Pattern clarity: does it form a recognizable structure?
+        pattern_clarity = self._assess_pattern_clarity(sequence)
+
+        # Structural closure: does it end properly?
+        structural_closure = self._assess_closure(sequence)
+
+        return (transition_quality + pattern_clarity + structural_closure) / 3.0
+
+    def _calculate_balance(
+        self, sequence: List[str], stabilizer_count: int, destabilizer_count: int
+    ) -> float:
+        """Calculate balance score: equilibrium between stabilizers and destabilizers.
+
+        Ideal sequences have roughly equal stabilization and transformation forces.
+        Severe imbalance reduces structural health.
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+        stabilizer_count : int
+            Pre-computed count of stabilizing operators
+        destabilizer_count : int
+            Pre-computed count of destabilizing operators
+
+        Returns
+        -------
+        float
+            Balance score (0.0-1.0)
+        """
+        if not sequence:
+            return 0.5  # Neutral for empty
+
+        # If neither present, neutral balance
+        if stabilizer_count == 0 and destabilizer_count == 0:
+            return 0.5
+
+        # Calculate ratio: closer to 1.0 means better balance
+        max_count = max(stabilizer_count, destabilizer_count)
+        min_count = min(stabilizer_count, destabilizer_count)
+
+        if max_count == 0:
+            return 0.5
+
+        ratio = min_count / max_count
+
+        # Penalize severe imbalance (difference > half the sequence length)
+        imbalance = abs(stabilizer_count - destabilizer_count)
+        if imbalance > len(sequence) // 2:
+            ratio *= 0.7  # Apply penalty
+            self._recommendations.append(
+                "Severe imbalance detected: add stabilizers or reduce destabilizers"
+            )
+
+        return ratio
+
+    def _calculate_sustainability(
+        self,
+        sequence: List[str],
+        stabilizer_count: int,
+        destabilizer_count: int,
+        regenerator_count: int,
+    ) -> float:
+        """Calculate sustainability index: capacity to maintain without collapse.
+
+        Factors:
+        - Final operator is a stabilizer
+        - Dissonance is resolved (not left unbalanced)
+        - Contains regenerative elements
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+        stabilizer_count : int
+            Pre-computed count of stabilizing operators
+        destabilizer_count : int
+            Pre-computed count of destabilizing operators
+        regenerator_count : int
+            Pre-computed count of regenerative operators
+
+        Returns
+        -------
+        float
+            Sustainability score (0.0-1.0)
+        """
+        if not sequence:
+            return 0.0
+
+        sustainability = 0.0
+
+        # Factor 1: Ends with stabilizer (0.4 points)
+        has_final_stabilizer = sequence[-1] in _STABILIZERS
+        if has_final_stabilizer:
+            sustainability += 0.4
+        else:
+            sustainability += 0.1  # Some credit for other endings
+            self._recommendations.append(
+                "Consider ending with a stabilizer (coherence, silence, resonance, or self_organization)"
+            )
+
+        # Factor 2: Resolved dissonance (0.3 points)
+        unresolved_dissonance = self._count_unresolved_dissonance(sequence)
+        if unresolved_dissonance == 0:
+            sustainability += 0.3
+        else:
+            penalty = min(0.3, unresolved_dissonance * 0.1)
+            sustainability += max(0, 0.3 - penalty)
+            if unresolved_dissonance > 1:
+                self._recommendations.append(
+                    "Multiple unresolved dissonances detected: add stabilizers after destabilizing operators"
+                )
+
+        # Factor 3: Regenerative elements (0.3 points)
+        # Use pre-computed regenerator count
+        if regenerator_count > 0:
+            sustainability += 0.3
+        else:
+            sustainability += 0.1  # Some credit even without
+
+        return min(1.0, sustainability)
+
+    def _calculate_efficiency(self, sequence: List[str], unique_count: int) -> float:
+        """Calculate complexity efficiency: value achieved relative to length.
+
+        Penalizes unnecessarily long sequences that don't provide proportional value.
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+        unique_count : int
+            Pre-computed count of unique operators in sequence
+
+        Returns
+        -------
+        float
+            Efficiency score (0.0-1.0)
+        """
+        if not sequence:
+            return 0.0
+
+        # Assess structural value using pre-computed unique count
+        diversity_score = min(
+            1.0, unique_count / 6.0
+        )  # 6+ operators is excellent diversity
+
+        # Note: We still need to call _assess_pattern_value for category coverage
+        # This is minimal overhead as it's a single pass checking set memberships
+        pattern_value = self._assess_pattern_value_optimized(sequence, unique_count)
+
+        # Length penalty: sequences longer than 10 operators get penalized
+        # Optimal range is 3-8 operators
+        length = len(sequence)
+        if length < 3:
+            length_factor = 0.7  # Too short, limited value
+        elif length <= 8:
+            length_factor = 1.0  # Optimal range
+        else:
+            # Gradual penalty for length > 8
+            excess = length - 8
+            length_factor = max(0.5, 1.0 - (excess * 0.05))
+
+        if length > 12:
+            self._recommendations.append(
+                f"Sequence is long ({length} operators): consider breaking into sub-sequences"
+            )
+
+        return pattern_value * length_factor
+
+    def _calculate_frequency_harmony(self, sequence: List[str]) -> float:
+        """Calculate frequency harmony: smoothness of νf transitions.
+
+        TODO: Full implementation requires integration with STRUCTURAL_FREQUENCIES
+        and FREQUENCY_TRANSITIONS from the grammar module. Currently returns
+        a conservative estimate based on transition patterns.
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+
+        Returns
+        -------
+        float
+            Harmony score (0.0-1.0)
+        """
+        # Conservative estimate: assume good harmony unless obvious issues detected
+        # Future enhancement: integrate with grammar.STRUCTURAL_FREQUENCIES
+        return 0.85
+
+    def _calculate_completeness(
+        self,
+        sequence: List[str],
+        stabilizer_count: int,
+        destabilizer_count: int,
+        transformer_count: int,
+    ) -> float:
+        """Calculate pattern completeness: how complete the pattern is.
+
+        Complete patterns (with activation, transformation, stabilization) score higher.
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+        stabilizer_count : int
+            Pre-computed count of stabilizing operators
+        destabilizer_count : int
+            Pre-computed count of destabilizing operators
+        transformer_count : int
+            Pre-computed count of transforming operators
+
+        Returns
+        -------
+        float
+            Completeness score (0.0-1.0)
+        """
+        if not sequence:
+            return 0.0
+
+        # Check for key phases using pre-computed counts and minimal checks
+        has_activation = any(op in {"emission", "reception"} for op in sequence)
+        has_transformation = destabilizer_count > 0 or transformer_count > 0
+        has_stabilization = stabilizer_count > 0
+        has_completion = any(op in {"silence", "transition"} for op in sequence)
+
+        phase_count = sum(
+            [has_activation, has_transformation, has_stabilization, has_completion]
+        )
+
+        # All 4 phases = 1.0, 3 phases = 0.75, 2 phases = 0.5, 1 phase = 0.25
+        return phase_count / 4.0
+
+    def _calculate_smoothness(
+        self, sequence: List[str], problematic_transitions: List[Tuple[str, str]]
+    ) -> float:
+        """Calculate transition smoothness: quality of operator transitions.
+
+        Measures ratio of valid/smooth transitions vs total transitions.
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+        problematic_transitions : List[Tuple[str, str]]
+            Pre-computed list of problematic transition pairs
+
+        Returns
+        -------
+        float
+            Smoothness score (0.0-1.0)
+        """
+        if len(sequence) < 2:
+            return 1.0  # No transitions to assess
+
+        total_transitions = len(sequence) - 1
+        # Each problematic transition gets 0.5 penalty (same as in _calculate_coherence)
+        penalty = len(problematic_transitions) * 0.5
+        return max(0.0, 1.0 - (penalty / total_transitions))
+
+    def _assess_pattern_clarity(self, sequence: List[str]) -> float:
+        """Assess how clearly the sequence forms a recognizable pattern.
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+
+        Returns
+        -------
+        float
+            Pattern clarity score (0.0-1.0)
+        """
+        if len(sequence) < 3:
+            return 0.5  # Too short for clear pattern
+
+        # Check for canonical patterns
+        pattern = self._detect_pattern(sequence)
+
+        if pattern in {"activation", "therapeutic", "regenerative", "transformative"}:
+            return 0.9  # Clear, recognized pattern
+        elif pattern in {"stabilization", "exploratory"}:
+            return 0.7  # Recognizable but simpler
+        else:
+            return 0.5  # No clear pattern
+
+    def _assess_closure(self, sequence: List[str]) -> float:
+        """Assess structural closure quality.
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+
+        Returns
+        -------
+        float
+            Closure quality score (0.0-1.0)
+        """
+        if not sequence:
+            return 0.0
+
+        # Valid endings per grammar
+        valid_endings = {SILENCE, TRANSITION, RECURSIVITY, DISSONANCE}
+
+        if sequence[-1] in valid_endings:
+            # Stabilizer endings are best
+            if sequence[-1] in _STABILIZERS:
+                return 1.0
+            # Other valid endings are good
+            return 0.8
+
+        # Invalid ending
+        return 0.3
+
+    def _count_unresolved_dissonance(self, sequence: List[str]) -> int:
+        """Count destabilizers not followed by stabilizers within reasonable window.
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+
+        Returns
+        -------
+        int
+            Count of unresolved dissonant operators
+        """
+        unresolved = 0
+        window = 3  # Look ahead up to 3 operators
+
+        for i, op in enumerate(sequence):
+            if op in DESTABILIZERS:
+                # Check if a stabilizer appears in the next 'window' operators
+                lookahead = sequence[i + 1 : i + 1 + window]
+                if not any(stabilizer in _STABILIZERS for stabilizer in lookahead):
+                    unresolved += 1
+
+        return unresolved
+
+    def _assess_pattern_value_optimized(
+        self, sequence: List[str], unique_count: int
+    ) -> float:
+        """Assess the structural value of the pattern using pre-computed unique count.
+
+        Value is higher when:
+        - Multiple operator types present (diversity)
+        - Key structural phases are included
+        - Balance between forces
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+        unique_count : int
+            Pre-computed count of unique operators in sequence
+
+        Returns
+        -------
+        float
+            Pattern value score (0.0-1.0)
+        """
+        if not sequence:
+            return 0.0
+
+        # Diversity: use pre-computed unique count
+        diversity_score = min(
+            1.0, unique_count / 6.0
+        )  # 6+ operators is excellent diversity
+
+        # Coverage: how many operator categories are represented
+        # This is still a minimal single-pass check
+        categories_present = 0
+        if any(op in {"emission", "reception"} for op in sequence):
+            categories_present += 1  # Activation
+        if any(op in _STABILIZERS for op in sequence):
+            categories_present += 1  # Stabilization
+        if any(op in DESTABILIZERS for op in sequence):
+            categories_present += 1  # Destabilization
+        if any(op in TRANSFORMERS for op in sequence):
+            categories_present += 1  # Transformation
+
+        coverage_score = categories_present / 4.0
+
+        # Combine factors
+        return (diversity_score * 0.5) + (coverage_score * 0.5)
+
+    def _detect_pattern(self, sequence: List[str]) -> str:
+        """Detect the dominant structural pattern type.
+
+        Parameters
+        ----------
+        sequence : List[str]
+            Operator sequence
+
+        Returns
+        -------
+        str
+            Pattern name (e.g., "activation", "therapeutic", "unknown")
+        """
+        if not sequence:
+            return "empty"
+
+        # Check for common patterns
+        starts_with_emission = sequence[0] == "emission"
+        has_reception = "reception" in sequence
+        has_coherence = COHERENCE in sequence
+        has_dissonance = DISSONANCE in sequence
+        has_self_org = SELF_ORGANIZATION in sequence
+        has_regenerator = any(op in _REGENERATORS for op in sequence)
+
+        # Pattern detection logic
+        if starts_with_emission and has_reception and has_coherence:
+            if has_dissonance and has_self_org:
+                return "therapeutic"
+            elif has_regenerator:
+                return "regenerative"
+            else:
+                return "activation"
+
+        if has_dissonance and has_self_org:
+            return "transformative"
+
+        if sum(1 for op in sequence if op in _STABILIZERS) > len(sequence) // 2:
+            return "stabilization"
+
+        if sum(1 for op in sequence if op in DESTABILIZERS) > len(sequence) // 2:
+            return "exploratory"
+
+        return "unknown"
+
+    def analyze_thol_coherence(self, G: TNFRGraph) -> dict[str, Any] | None:
+        """Analyze collective coherence of THOL bifurcations across the network.
+
+        Examines all nodes that have undergone THOL bifurcation and provides
+        statistics on their collective coherence metrics.
+
+        Parameters
+        ----------
+        G : TNFRGraph
+            Graph containing nodes with potential THOL bifurcations
+
+        Returns
+        -------
+        dict or None
+            Dictionary containing coherence statistics:
+            - mean_coherence: Average coherence across all THOL nodes
+            - min_coherence: Lowest coherence value observed
+            - max_coherence: Highest coherence value observed
+            - nodes_below_threshold: Count of nodes with coherence < 0.3
+            - total_thol_nodes: Total nodes with sub-EPIs
+            Returns None if no THOL bifurcations exist in the network.
+
+        Notes
+        -----
+        TNFR Principle: Collective coherence measures the structural alignment
+        of emergent sub-EPIs. Low coherence may indicate chaotic fragmentation
+        rather than controlled emergence.
+
+        This metric is particularly useful for:
+        - Detecting pathological bifurcation patterns
+        - Monitoring network-wide self-organization quality
+        - Identifying nodes requiring stabilization
+
+        Examples
+        --------
+        >>> analyzer = SequenceHealthAnalyzer()
+        >>> # After running THOL operations on graph G
+        >>> coherence_stats = analyzer.analyze_thol_coherence(G)
+        >>> if coherence_stats:
+        ...     print(f"Mean coherence: {coherence_stats['mean_coherence']:.3f}")
+        ...     print(f"Nodes below threshold: {coherence_stats['nodes_below_threshold']}")
+        """
+        # Find all nodes with sub-EPIs (THOL bifurcation occurred)
+        thol_nodes = []
+        for node in G.nodes():
+            if G.nodes[node].get("sub_epis"):
+                thol_nodes.append(node)
+
+        if not thol_nodes:
+            return None
+
+        # Collect coherence values
+        coherences = []
+        for node in thol_nodes:
+            coh = G.nodes[node].get("_thol_collective_coherence")
+            if coh is not None:
+                coherences.append(coh)
+
+        if not coherences:
+            return None
+
+        # Compute statistics
+        mean_coherence = sum(coherences) / len(coherences)
+        min_coherence = min(coherences)
+        max_coherence = max(coherences)
+
+        # Get threshold from graph config
+        threshold = float(G.graph.get("THOL_MIN_COLLECTIVE_COHERENCE", 0.3))
+        nodes_below_threshold = sum(1 for c in coherences if c < threshold)
+
+        return {
+            "mean_coherence": mean_coherence,
+            "min_coherence": min_coherence,
+            "max_coherence": max_coherence,
+            "nodes_below_threshold": nodes_below_threshold,
+            "total_thol_nodes": len(thol_nodes),
+            "threshold": threshold,
+        }