kailash 0.1.5__py3-none-any.whl → 0.2.1__py3-none-any.whl

kailash/workflow/cycle_profiler.py

@@ -0,0 +1,741 @@
+"""
+Advanced Performance Profiling and Analysis for Cyclic Workflows.
+
+This module provides comprehensive performance profiling capabilities for cyclic
+workflows, including detailed statistical analysis, resource usage monitoring,
+bottleneck identification, and automated optimization recommendations based on
+execution patterns and performance characteristics.
+
+Examples:
+    Basic profiling setup:
+
+    >>> profiler = CycleProfiler(enable_advanced_metrics=True)
+    >>> # Add execution traces
+    >>> profiler.add_trace(execution_trace)
+    >>> # Analyze performance
+    >>> metrics = profiler.analyze_performance()
+    >>> print(f"Overall efficiency: {metrics.efficiency_score}")
+
+    Comparative analysis:
+
+    >>> # Compare multiple cycles
+    >>> comparison = profiler.compare_cycles(["cycle_1", "cycle_2", "cycle_3"])
+    >>> print(f"Best cycle: {comparison['best_cycle']}")
+    >>> print(f"Performance gaps: {comparison['significant_differences']}")
+
+    Optimization recommendations:
+
+    >>> # Get actionable recommendations
+    >>> recommendations = profiler.get_optimization_recommendations()
+    >>> for rec in recommendations:
+    ...     print(f"{rec['priority']}: {rec['description']}")
+    ...     print(f"   Suggestion: {rec['suggestion']}")
+    ...     print(f"   Current: {rec['current_value']}")
+    ...     print(f"   Target: {rec['target_improvement']}")
+
+    Comprehensive reporting:
+
+    >>> # Generate detailed report
+    >>> report = profiler.generate_performance_report()
+    >>> # Export for external analysis
+    >>> profiler.export_profile_data("performance_analysis.json")
+"""
+
+import logging
+import statistics
+from dataclasses import dataclass, field
+from datetime import datetime
+from typing import Any, Dict, List, Optional, Tuple
+
+from kailash.workflow.cycle_debugger import CycleExecutionTrace
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class PerformanceMetrics:
+    """
+    Comprehensive performance metrics for cycle analysis.
+
+    This class aggregates and analyzes performance data from cycle executions,
+    providing detailed insights into timing, resource usage, and efficiency
+    characteristics of cyclic workflows.
+
+    Attributes:
+        total_cycles: Number of cycles analyzed.
+        total_iterations: Total iterations across all cycles.
+        avg_cycle_time: Average cycle execution time.
+        avg_iteration_time: Average iteration execution time.
+        min_iteration_time: Fastest iteration time.
+        max_iteration_time: Slowest iteration time.
+        iteration_time_stddev: Standard deviation of iteration times.
+        memory_stats: Memory usage statistics.
+        cpu_stats: CPU usage statistics.
+        convergence_stats: Convergence analysis.
+        bottlenecks: Identified performance bottlenecks.
+        optimization_opportunities: Suggested optimizations.
+    """
+
+    total_cycles: int = 0
+    total_iterations: int = 0
+    avg_cycle_time: float = 0.0
+    avg_iteration_time: float = 0.0
+    min_iteration_time: float = float("inf")
+    max_iteration_time: float = 0.0
+    iteration_time_stddev: float = 0.0
+    memory_stats: Dict[str, float] = field(default_factory=dict)
+    cpu_stats: Dict[str, float] = field(default_factory=dict)
+    convergence_stats: Dict[str, Any] = field(default_factory=dict)
+    bottlenecks: List[str] = field(default_factory=list)
+    optimization_opportunities: List[str] = field(default_factory=list)
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert metrics to dictionary for serialization.
+
+        Returns:
+            Dictionary representation of performance metrics.
+        """
+        return {
+            "total_cycles": self.total_cycles,
+            "total_iterations": self.total_iterations,
+            "timing": {
+                "avg_cycle_time": self.avg_cycle_time,
+                "avg_iteration_time": self.avg_iteration_time,
+                "min_iteration_time": (
+                    self.min_iteration_time
+                    if self.min_iteration_time != float("inf")
+                    else 0.0
+                ),
+                "max_iteration_time": self.max_iteration_time,
+                "iteration_time_stddev": self.iteration_time_stddev,
+            },
+            "memory_stats": self.memory_stats,
+            "cpu_stats": self.cpu_stats,
+            "convergence_stats": self.convergence_stats,
+            "bottlenecks": self.bottlenecks,
+            "optimization_opportunities": self.optimization_opportunities,
+        }
+
+
+class CycleProfiler:
+    """
+    Advanced profiling and performance analysis for cyclic workflows.
+
+    This class provides comprehensive performance analysis capabilities for
+    cycles, including statistical analysis, bottleneck identification,
+    comparative analysis across multiple cycles, and detailed optimization
+    recommendations based on execution patterns.
+
+    Examples:
+        >>> profiler = CycleProfiler()
+        >>> profiler.add_trace(execution_trace)
+        >>> metrics = profiler.analyze_performance()
+        >>> recommendations = profiler.get_optimization_recommendations()
+    """
+
+    def __init__(self, enable_advanced_metrics: bool = True):
+        """
+        Initialize cycle profiler.
+
+        Args:
+            enable_advanced_metrics: Whether to enable advanced statistical analysis.
+        """
+        self.enable_advanced_metrics = enable_advanced_metrics
+        self.traces: List[CycleExecutionTrace] = []
+        self.performance_history: List[PerformanceMetrics] = []
+
+    def add_trace(self, trace: CycleExecutionTrace):
+        """
+        Add a cycle execution trace for analysis.
+
+        Args:
+            trace: Completed execution trace to analyze.
+
+        Examples:
+            >>> profiler.add_trace(execution_trace)
+        """
+        self.traces.append(trace)
+        logger.debug(
+            f"Added trace for cycle '{trace.cycle_id}' with {len(trace.iterations)} iterations"
+        )
+
+    def analyze_performance(self) -> PerformanceMetrics:
+        """
+        Perform comprehensive performance analysis on all traces.
+
+        Analyzes all collected traces to generate comprehensive performance
+        metrics, identify bottlenecks, and provide optimization recommendations
+        based on statistical analysis of execution patterns.
+
+        Returns:
+            Comprehensive performance analysis results.
+
+        Examples:
+            >>> metrics = profiler.analyze_performance()
+            >>> print(f"Average cycle time: {metrics.avg_cycle_time:.3f}s")
+        """
+        if not self.traces:
+            logger.warning("No traces available for performance analysis")
+            return PerformanceMetrics()
+
+        # Collect all timing data
+        cycle_times = []
+        iteration_times = []
+        memory_values = []
+        cpu_values = []
+        convergence_data = []
+
+        for trace in self.traces:
+            if trace.total_execution_time:
+                cycle_times.append(trace.total_execution_time)
+
+            for iteration in trace.iterations:
+                if iteration.execution_time:
+                    iteration_times.append(iteration.execution_time)
+                if iteration.memory_usage_mb:
+                    memory_values.append(iteration.memory_usage_mb)
+                if iteration.cpu_usage_percent:
+                    cpu_values.append(iteration.cpu_usage_percent)
+                if iteration.convergence_value:
+                    convergence_data.append(iteration.convergence_value)
+
+        # Calculate basic metrics
+        metrics = PerformanceMetrics(
+            total_cycles=len(self.traces),
+            total_iterations=sum(len(trace.iterations) for trace in self.traces),
+        )
+
+        # Timing analysis
+        if cycle_times:
+            metrics.avg_cycle_time = statistics.mean(cycle_times)
+
+        if iteration_times:
+            metrics.avg_iteration_time = statistics.mean(iteration_times)
+            metrics.min_iteration_time = min(iteration_times)
+            metrics.max_iteration_time = max(iteration_times)
+            if len(iteration_times) > 1:
+                metrics.iteration_time_stddev = statistics.stdev(iteration_times)
+
+        # Memory analysis
+        if memory_values:
+            metrics.memory_stats = {
+                "avg": statistics.mean(memory_values),
+                "min": min(memory_values),
+                "max": max(memory_values),
+                "stddev": (
+                    statistics.stdev(memory_values) if len(memory_values) > 1 else 0.0
+                ),
+                "median": statistics.median(memory_values),
+            }
+
+        # CPU analysis
+        if cpu_values:
+            metrics.cpu_stats = {
+                "avg": statistics.mean(cpu_values),
+                "min": min(cpu_values),
+                "max": max(cpu_values),
+                "stddev": statistics.stdev(cpu_values) if len(cpu_values) > 1 else 0.0,
+                "median": statistics.median(cpu_values),
+            }
+
+        # Convergence analysis
+        if convergence_data:
+            metrics.convergence_stats = self._analyze_convergence_performance(
+                convergence_data
+            )
+
+        # Advanced analysis if enabled
+        if self.enable_advanced_metrics:
+            metrics.bottlenecks = self._identify_bottlenecks(metrics)
+            metrics.optimization_opportunities = self._identify_optimizations(metrics)
+
+        # Store in history
+        self.performance_history.append(metrics)
+
+        logger.info(
+            f"Analyzed performance for {metrics.total_cycles} cycles, "
+            f"{metrics.total_iterations} iterations, "
+            f"avg cycle time: {metrics.avg_cycle_time:.3f}s"
+        )
+
+        return metrics
+
+    def compare_cycles(self, cycle_ids: List[str]) -> Dict[str, Any]:
+        """
+        Compare performance across multiple specific cycles.
+
+        Provides detailed comparative analysis between specific cycles,
+        highlighting performance differences, convergence patterns, and
+        relative efficiency metrics.
+
+        Args:
+            cycle_ids (List[str]): List of cycle IDs to compare
+
+        Returns:
+            Dict[str, Any]: Comparative analysis results
+
+        Side Effects:
+            None - this is a pure analysis method
+
+        Example:
+            >>> comparison = profiler.compare_cycles(["cycle_1", "cycle_2"])
+            >>> print(f"Best performing cycle: {comparison['best_cycle']}")
+        """
+        relevant_traces = [
+            trace for trace in self.traces if trace.cycle_id in cycle_ids
+        ]
+
+        if len(relevant_traces) < 2:
+            return {"error": "Need at least 2 cycles for comparison"}
+
+        comparison = {
+            "cycles_compared": len(relevant_traces),
+            "cycle_details": {},
+            "performance_ranking": [],
+            "significant_differences": [],
+        }
+
+        # Analyze each cycle
+        for trace in relevant_traces:
+            stats = trace.get_statistics()
+
+            comparison["cycle_details"][trace.cycle_id] = {
+                "execution_time": trace.total_execution_time,
+                "iterations": len(trace.iterations),
+                "converged": trace.converged,
+                "efficiency_score": stats["efficiency_score"],
+                "avg_iteration_time": stats["avg_iteration_time"],
+                "convergence_rate": stats["convergence_rate"],
+            }
+
+        # Rank by efficiency score
+        ranking = sorted(
+            comparison["cycle_details"].items(),
+            key=lambda x: x[1]["efficiency_score"],
+            reverse=True,
+        )
+        comparison["performance_ranking"] = [cycle_id for cycle_id, _ in ranking]
+        comparison["best_cycle"] = ranking[0][0] if ranking else None
+        comparison["worst_cycle"] = ranking[-1][0] if ranking else None
+
+        # Identify significant differences
+        if len(ranking) >= 2:
+            best_score = ranking[0][1]["efficiency_score"]
+            worst_score = ranking[-1][1]["efficiency_score"]
+
+            if best_score - worst_score > 0.2:
+                comparison["significant_differences"].append(
+                    f"Large efficiency gap: {best_score:.2f} vs {worst_score:.2f}"
+                )
+
+            # Compare convergence rates
+            convergence_rates = [details["convergence_rate"] for _, details in ranking]
+            if max(convergence_rates) - min(convergence_rates) > 0.3:
+                comparison["significant_differences"].append(
+                    "Significant variation in convergence rates"
+                )
+
+        return comparison
+
+    def get_optimization_recommendations(
+        self, trace: Optional[CycleExecutionTrace] = None
+    ) -> List[Dict[str, Any]]:
+        """
+        Generate detailed optimization recommendations.
+
+        Provides specific, actionable optimization recommendations based on
+        performance analysis, including parameter tuning suggestions,
+        algorithmic improvements, and resource optimization strategies.
+
+        Args:
+            trace (Optional[CycleExecutionTrace]): Specific trace to analyze, or None for overall recommendations
+
+        Returns:
+            List[Dict[str, Any]]: List of optimization recommendations with details
+
+        Side Effects:
+            None - this is a pure analysis method
+
+        Example:
+            >>> recommendations = profiler.get_optimization_recommendations()
+            >>> for rec in recommendations:
+            ...     print(f"{rec['priority']}: {rec['description']}")
+        """
+        recommendations = []
+
+        if trace:
+            traces_to_analyze = [trace]
+        else:
+            traces_to_analyze = self.traces
+
+        if not traces_to_analyze:
+            return recommendations
+
+        # Analyze all traces for patterns
+        self.analyze_performance() if not trace else None
+
+        for target_trace in traces_to_analyze:
+            stats = target_trace.get_statistics()
+
+            # Efficiency recommendations
+            if stats["efficiency_score"] < 0.3:
+                recommendations.append(
+                    {
+                        "priority": "HIGH",
+                        "category": "efficiency",
+                        "description": "Very low efficiency detected",
+                        "suggestion": "Consider reducing max_iterations or improving convergence condition",
+                        "cycle_id": target_trace.cycle_id,
+                        "current_value": stats["efficiency_score"],
+                        "target_improvement": "Increase to > 0.5",
+                    }
+                )
+
+            # Convergence recommendations
+            if not target_trace.converged:
+                if stats["total_iterations"] >= (
+                    target_trace.max_iterations_configured or 0
+                ):
+                    recommendations.append(
+                        {
+                            "priority": "HIGH",
+                            "category": "convergence",
+                            "description": "Cycle reached max_iterations without converging",
+                            "suggestion": "Increase max_iterations or improve algorithm",
+                            "cycle_id": target_trace.cycle_id,
+                            "current_value": target_trace.max_iterations_configured,
+                            "target_improvement": f"Increase to {int((target_trace.max_iterations_configured or 10) * 1.5)}",
+                        }
+                    )
+
+            # Performance recommendations
+            if stats["avg_iteration_time"] > 0.5:
+                recommendations.append(
+                    {
+                        "priority": "MEDIUM",
+                        "category": "performance",
+                        "description": "High average iteration time",
+                        "suggestion": "Optimize node execution or reduce data processing",
+                        "cycle_id": target_trace.cycle_id,
+                        "current_value": stats["avg_iteration_time"],
+                        "target_improvement": "Reduce to < 0.5s per iteration",
+                    }
+                )
+
+            # Memory recommendations
+            if target_trace.memory_peak_mb and target_trace.memory_peak_mb > 1000:
+                recommendations.append(
+                    {
+                        "priority": "MEDIUM",
+                        "category": "memory",
+                        "description": "High memory usage detected",
+                        "suggestion": "Consider data streaming, chunking, or garbage collection",
+                        "cycle_id": target_trace.cycle_id,
+                        "current_value": target_trace.memory_peak_mb,
+                        "target_improvement": "Reduce to < 1000 MB",
+                    }
+                )
+
+            # Convergence pattern recommendations
+            convergence_trend = target_trace.get_convergence_trend()
+            if convergence_trend:
+                pattern_analysis = self._analyze_convergence_pattern(convergence_trend)
+                if pattern_analysis["unstable"]:
+                    recommendations.append(
+                        {
+                            "priority": "HIGH",
+                            "category": "stability",
+                            "description": "Unstable convergence pattern detected",
+                            "suggestion": "Reduce learning rate or add regularization",
+                            "cycle_id": target_trace.cycle_id,
+                            "current_value": "Unstable",
+                            "target_improvement": "Stable convergence",
+                        }
+                    )
+
+        # Sort by priority
+        priority_order = {"HIGH": 0, "MEDIUM": 1, "LOW": 2}
+        recommendations.sort(key=lambda x: priority_order.get(x["priority"], 3))
+
+        return recommendations
+
+    def generate_performance_report(self) -> Dict[str, Any]:
+        """
+        Generate comprehensive performance report.
+
+        Creates a detailed performance report including metrics analysis,
+        recommendations, trends, and comparative insights across all
+        analyzed cycles.
+
+        Returns:
+            Dict[str, Any]: Comprehensive performance report
+
+        Side Effects:
+            None - this is a pure analysis method
+
+        Example:
+            >>> report = profiler.generate_performance_report()
+            >>> print(f"Overall score: {report['overall_score']}")
+        """
+        metrics = self.analyze_performance()
+
+        # Calculate overall performance score
+        overall_score = self._calculate_overall_score(metrics)
+
+        # Generate trend analysis if we have history
+        trend_analysis = (
+            self._analyze_performance_trends()
+            if len(self.performance_history) > 1
+            else None
+        )
+
+        # Get top recommendations
+        recommendations = self.get_optimization_recommendations()
+
+        report = {
+            "summary": {
+                "overall_score": overall_score,
+                "total_cycles_analyzed": metrics.total_cycles,
+                "total_iterations": metrics.total_iterations,
+                "avg_cycle_time": metrics.avg_cycle_time,
+                "primary_bottlenecks": metrics.bottlenecks[:3],
+            },
+            "detailed_metrics": metrics.to_dict(),
+            "trend_analysis": trend_analysis,
+            "recommendations": recommendations[:10],  # Top 10 recommendations
+            "cycle_comparisons": self._get_cycle_comparisons(),
+            "generated_at": datetime.now().isoformat(),
+        }
+
+        return report
+
+    def export_profile_data(self, filepath: str, format: str = "json"):
+        """
+        Export profiling data for external analysis.
+
+        Args:
+            filepath (str): Output file path
+            format (str): Export format ("json", "csv")
+
+        Side Effects:
+            Creates file with profiling data
+
+        Example:
+            >>> profiler.export_profile_data("profile_analysis.json")
+        """
+        report = self.generate_performance_report()
+
+        if format == "json":
+            import json
+
+            with open(filepath, "w") as f:
+                json.dump(report, f, indent=2)
+        elif format == "csv":
+            import csv
+
+            with open(filepath, "w", newline="") as f:
+                writer = csv.writer(f)
+
+                # Write summary data
+                writer.writerow(["Metric", "Value"])
+                writer.writerow(["Overall Score", report["summary"]["overall_score"]])
+                writer.writerow(
+                    ["Total Cycles", report["summary"]["total_cycles_analyzed"]]
+                )
+                writer.writerow(["Avg Cycle Time", report["summary"]["avg_cycle_time"]])
+                writer.writerow([])
+
+                # Write detailed metrics
+                writer.writerow(["Detailed Metrics"])
+                metrics = report["detailed_metrics"]
+                for key, value in metrics.items():
+                    if isinstance(value, dict):
+                        for sub_key, sub_value in value.items():
+                            writer.writerow([f"{key}_{sub_key}", sub_value])
+                    else:
+                        writer.writerow([key, value])
+        else:
+            raise ValueError(f"Unsupported export format: {format}")
+
+        logger.info(f"Exported profiling data to {filepath} in {format} format")
+
+    def _analyze_convergence_performance(
+        self, convergence_data: List[float]
+    ) -> Dict[str, Any]:
+        """Analyze convergence performance characteristics."""
+        if not convergence_data:
+            return {}
+
+        return {
+            "avg_convergence": statistics.mean(convergence_data),
+            "min_convergence": min(convergence_data),
+            "max_convergence": max(convergence_data),
+            "convergence_stddev": (
+                statistics.stdev(convergence_data) if len(convergence_data) > 1 else 0.0
+            ),
+            "convergence_trend": (
+                "improving"
+                if convergence_data[0] > convergence_data[-1]
+                else "degrading"
+            ),
+            "data_points": len(convergence_data),
+        }
+
+    def _analyze_convergence_pattern(
+        self, convergence_trend: List[Tuple[int, Optional[float]]]
+    ) -> Dict[str, Any]:
+        """Analyze convergence pattern for stability."""
+        valid_points = [value for _, value in convergence_trend if value is not None]
+
+        if len(valid_points) < 3:
+            return {"unstable": False, "reason": "insufficient_data"}
+
+        # Calculate volatility
+        differences = [
+            abs(valid_points[i] - valid_points[i - 1])
+            for i in range(1, len(valid_points))
+        ]
+        avg_difference = statistics.mean(differences)
+        max_difference = max(differences)
+
+        # Consider unstable if large swings or high volatility
+        unstable = max_difference > (2 * avg_difference) and avg_difference > 0.1
+
+        return {
+            "unstable": unstable,
+            "avg_volatility": avg_difference,
+            "max_volatility": max_difference,
+            "reason": "high_volatility" if unstable else "stable",
+        }
+
+    def _identify_bottlenecks(self, metrics: PerformanceMetrics) -> List[str]:
+        """Identify performance bottlenecks from metrics."""
+        bottlenecks = []
+
+        # High iteration time variance suggests inconsistent performance
+        if metrics.iteration_time_stddev > metrics.avg_iteration_time * 0.5:
+            bottlenecks.append(
+                "High iteration time variance - inconsistent node performance"
+            )
+
+        # Very slow iterations
+        if metrics.max_iteration_time > metrics.avg_iteration_time * 3:
+            bottlenecks.append(
+                "Outlier slow iterations detected - potential resource contention"
+            )
+
+        # Memory bottlenecks
+        if metrics.memory_stats and metrics.memory_stats.get("max", 0) > 2000:
+            bottlenecks.append(
+                "High memory usage - potential memory leaks or inefficient data handling"
+            )
+
+        # CPU bottlenecks
+        if metrics.cpu_stats and metrics.cpu_stats.get("avg", 0) > 80:
+            bottlenecks.append("High CPU usage - computationally intensive operations")
+
+        return bottlenecks
+
+    def _identify_optimizations(self, metrics: PerformanceMetrics) -> List[str]:
+        """Identify optimization opportunities."""
+        optimizations = []
+
+        # Low convergence rate suggests early termination opportunities
+        convergence_rate = metrics.convergence_stats.get("avg_convergence")
+        if convergence_rate and convergence_rate < 0.5:
+            optimizations.append(
+                "Add early termination conditions for faster convergence"
+            )
+
+        # High memory variance suggests optimization potential
+        if metrics.memory_stats and metrics.memory_stats.get("stddev", 0) > 100:
+            optimizations.append("Optimize memory usage patterns for consistency")
+
+        # Slow average iteration time
+        if metrics.avg_iteration_time > 0.1:
+            optimizations.append("Optimize node execution performance")
+
+        return optimizations
+
+    def _calculate_overall_score(self, metrics: PerformanceMetrics) -> float:
+        """Calculate overall performance score (0-1, higher is better)."""
+        score_components = []
+
+        # Efficiency component (convergence rate)
+        if metrics.convergence_stats:
+            avg_convergence = metrics.convergence_stats.get("avg_convergence", 0.5)
+            score_components.append(min(1.0, avg_convergence))
+
+        # Speed component (based on iteration time)
+        if metrics.avg_iteration_time > 0:
+            speed_score = max(0.0, 1.0 - min(1.0, metrics.avg_iteration_time / 2.0))
+            score_components.append(speed_score)
+
+        # Consistency component (low variance is good)
+        if metrics.iteration_time_stddev >= 0:
+            consistency_score = max(
+                0.0,
+                1.0
+                - min(
+                    1.0,
+                    (
+                        metrics.iteration_time_stddev / metrics.avg_iteration_time
+                        if metrics.avg_iteration_time > 0
+                        else 0
+                    ),
+                ),
+            )
+            score_components.append(consistency_score)
+
+        # Memory efficiency component
+        if metrics.memory_stats:
+            max_memory = metrics.memory_stats.get("max", 500)
+            memory_score = max(
+                0.0, 1.0 - min(1.0, max_memory / 2000)
+            )  # Penalty after 2GB
+            score_components.append(memory_score)
+
+        return statistics.mean(score_components) if score_components else 0.5
+
+    def _analyze_performance_trends(self) -> Dict[str, Any]:
+        """Analyze performance trends over time."""
+        if len(self.performance_history) < 2:
+            return {"trend": "insufficient_data"}
+
+        recent_scores = [
+            self._calculate_overall_score(m) for m in self.performance_history[-5:]
+        ]
+
+        if len(recent_scores) >= 2:
+            trend = "improving" if recent_scores[-1] > recent_scores[0] else "degrading"
+            trend_strength = abs(recent_scores[-1] - recent_scores[0])
+        else:
+            trend = "stable"
+            trend_strength = 0.0
+
+        return {
+            "trend": trend,
+            "trend_strength": trend_strength,
+            "recent_scores": recent_scores,
+            "performance_history_length": len(self.performance_history),
+        }
+
+    def _get_cycle_comparisons(self) -> Dict[str, Any]:
+        """Get comparative analysis across all cycles."""
+        if len(self.traces) < 2:
+            return {"comparison": "insufficient_data"}
+
+        cycle_scores = {}
+        for trace in self.traces:
+            stats = trace.get_statistics()
+            cycle_scores[trace.cycle_id] = stats["efficiency_score"]
+
+        best_cycle = max(cycle_scores.items(), key=lambda x: x[1])
+        worst_cycle = min(cycle_scores.items(), key=lambda x: x[1])
+
+        return {
+            "best_cycle": {"id": best_cycle[0], "score": best_cycle[1]},
+            "worst_cycle": {"id": worst_cycle[0], "score": worst_cycle[1]},
+            "score_range": best_cycle[1] - worst_cycle[1],
+            "total_cycles": len(cycle_scores),
+        }