morphml 1.0.0__py3-none-any.whl

morphml/optimizers/multi_objective/visualization.py

@@ -0,0 +1,427 @@
+"""Visualization tools for multi-objective optimization results.
+
+Provides plotting functions for Pareto fronts, objective trade-offs,
+and optimization convergence.
+
+Author: Eshan Roy <eshanized@proton.me>
+Organization: TONMOY INFRASTRUCTURE & VISION
+"""
+
+from typing import List, Optional
+
+import numpy as np
+
+from morphml.logging_config import get_logger
+from morphml.optimizers.multi_objective.nsga2 import MultiObjectiveIndividual
+
+logger = get_logger(__name__)
+
+
+class ParetoVisualizer:
+    """
+    Visualization tools for Pareto fronts.
+
+    Provides various plotting methods for multi-objective optimization results.
+
+    Example:
+        >>> visualizer = ParetoVisualizer()
+        >>> visualizer.plot_2d(pareto_front, 'accuracy', 'latency')
+        >>> visualizer.plot_3d(pareto_front, 'accuracy', 'latency', 'params')
+    """
+
+    @staticmethod
+    def plot_2d(
+        pareto_front: List[MultiObjectiveIndividual],
+        obj1_name: str,
+        obj2_name: str,
+        title: Optional[str] = None,
+        save_path: Optional[str] = None,
+        show_labels: bool = False,
+    ) -> None:
+        """
+        Plot 2D Pareto front.
+
+        Args:
+            pareto_front: List of Pareto-optimal individuals
+            obj1_name: Name of first objective (x-axis)
+            obj2_name: Name of second objective (y-axis)
+            title: Optional plot title
+            save_path: Optional path to save figure
+            show_labels: Whether to show point labels
+
+        Example:
+            >>> ParetoVisualizer.plot_2d(
+            ...     pareto_front,
+            ...     obj1_name='accuracy',
+            ...     obj2_name='latency',
+            ...     save_path='pareto_front.png'
+            ... )
+        """
+        try:
+            import matplotlib.pyplot as plt
+        except ImportError:
+            logger.warning("matplotlib not available, cannot plot")
+            return
+
+        # Extract objective values
+        x = [ind.objectives[obj1_name] for ind in pareto_front]
+        y = [ind.objectives[obj2_name] for ind in pareto_front]
+
+        # Create plot
+        plt.figure(figsize=(10, 6))
+        plt.scatter(x, y, s=100, alpha=0.6, c="blue", edgecolors="black", linewidth=1.5)
+
+        # Connect points to show Pareto front
+        sorted_indices = np.argsort(x)
+        x_sorted = [x[i] for i in sorted_indices]
+        y_sorted = [y[i] for i in sorted_indices]
+        plt.plot(x_sorted, y_sorted, "r--", alpha=0.3, linewidth=1)
+
+        # Labels
+        plt.xlabel(obj1_name.replace("_", " ").title(), fontsize=14, fontweight="bold")
+        plt.ylabel(obj2_name.replace("_", " ").title(), fontsize=14, fontweight="bold")
+
+        if title is None:
+            title = f"Pareto Front: {obj1_name} vs {obj2_name}"
+        plt.title(title, fontsize=16, fontweight="bold")
+
+        # Add labels if requested
+        if show_labels:
+            for i, (xi, yi) in enumerate(zip(x, y)):
+                plt.annotate(f"{i}", (xi, yi), fontsize=8, alpha=0.7)
+
+        plt.grid(True, alpha=0.3, linestyle="--")
+        plt.tight_layout()
+
+        # Save or show
+        if save_path:
+            plt.savefig(save_path, dpi=300, bbox_inches="tight")
+            logger.info(f"Saved plot to {save_path}")
+        else:
+            plt.show()
+
+        plt.close()
+
+    @staticmethod
+    def plot_3d(
+        pareto_front: List[MultiObjectiveIndividual],
+        obj1_name: str,
+        obj2_name: str,
+        obj3_name: str,
+        title: Optional[str] = None,
+        save_path: Optional[str] = None,
+    ) -> None:
+        """
+        Plot 3D Pareto front.
+
+        Args:
+            pareto_front: List of Pareto-optimal individuals
+            obj1_name: Name of first objective (x-axis)
+            obj2_name: Name of second objective (y-axis)
+            obj3_name: Name of third objective (z-axis)
+            title: Optional plot title
+            save_path: Optional path to save figure
+
+        Example:
+            >>> ParetoVisualizer.plot_3d(
+            ...     pareto_front,
+            ...     'accuracy', 'latency', 'params'
+            ... )
+        """
+        try:
+            import matplotlib.pyplot as plt
+        except ImportError:
+            logger.warning("matplotlib not available, cannot plot")
+            return
+
+        # Extract objective values
+        x = [ind.objectives[obj1_name] for ind in pareto_front]
+        y = [ind.objectives[obj2_name] for ind in pareto_front]
+        z = [ind.objectives[obj3_name] for ind in pareto_front]
+
+        # Create 3D plot
+        fig = plt.figure(figsize=(12, 9))
+        ax = fig.add_subplot(111, projection="3d")
+
+        # Scatter plot colored by third objective
+        scatter = ax.scatter(
+            x, y, z, s=100, c=z, cmap="viridis", alpha=0.7, edgecolors="black", linewidth=1
+        )
+
+        # Labels
+        ax.set_xlabel(obj1_name.replace("_", " ").title(), fontsize=12, fontweight="bold")
+        ax.set_ylabel(obj2_name.replace("_", " ").title(), fontsize=12, fontweight="bold")
+        ax.set_zlabel(obj3_name.replace("_", " ").title(), fontsize=12, fontweight="bold")
+
+        if title is None:
+            title = f"3D Pareto Front: {obj1_name}, {obj2_name}, {obj3_name}"
+        ax.set_title(title, fontsize=14, fontweight="bold", pad=20)
+
+        # Colorbar
+        cbar = plt.colorbar(scatter, ax=ax, pad=0.1)
+        cbar.set_label(obj3_name.replace("_", " ").title(), fontsize=11)
+
+        # Grid
+        ax.grid(True, alpha=0.3)
+
+        # Save or show
+        if save_path:
+            plt.savefig(save_path, dpi=300, bbox_inches="tight")
+            logger.info(f"Saved plot to {save_path}")
+        else:
+            plt.show()
+
+        plt.close()
+
+    @staticmethod
+    def plot_parallel_coordinates(
+        pareto_front: List[MultiObjectiveIndividual],
+        objective_names: Optional[List[str]] = None,
+        title: Optional[str] = None,
+        save_path: Optional[str] = None,
+    ) -> None:
+        """
+        Parallel coordinates plot for multiple objectives.
+
+        Each line represents a solution, each axis represents an objective.
+        Useful for visualizing high-dimensional Pareto fronts.
+
+        Args:
+            pareto_front: List of Pareto-optimal individuals
+            objective_names: List of objectives to plot (None = all)
+            title: Optional plot title
+            save_path: Optional path to save figure
+
+        Example:
+            >>> ParetoVisualizer.plot_parallel_coordinates(
+            ...     pareto_front,
+            ...     objective_names=['accuracy', 'latency', 'params', 'flops']
+            ... )
+        """
+        try:
+            import matplotlib.pyplot as plt
+            import pandas as pd
+            from pandas.plotting import parallel_coordinates
+        except ImportError:
+            logger.warning("matplotlib/pandas not available, cannot plot")
+            return
+
+        # Extract objective values
+        if objective_names is None:
+            objective_names = list(pareto_front[0].objectives.keys())
+
+        data = []
+        for i, ind in enumerate(pareto_front):
+            row = {name: ind.objectives[name] for name in objective_names}
+            row["solution"] = i
+            data.append(row)
+
+        df = pd.DataFrame(data)
+
+        # Normalize objectives to [0, 1] for better visualization
+        for col in objective_names:
+            min_val = df[col].min()
+            max_val = df[col].max()
+            if max_val > min_val:
+                df[col] = (df[col] - min_val) / (max_val - min_val)
+
+        # Create plot
+        plt.figure(figsize=(14, 7))
+        parallel_coordinates(df, "solution", colormap="viridis", alpha=0.6, linewidth=1.5)
+
+        # Labels
+        if title is None:
+            title = "Pareto Front - Parallel Coordinates"
+        plt.title(title, fontsize=16, fontweight="bold")
+
+        plt.ylabel("Normalized Objective Value", fontsize=12)
+        plt.legend().remove()  # Remove legend (too many solutions)
+        plt.grid(True, alpha=0.3, axis="y")
+        plt.tight_layout()
+
+        # Save or show
+        if save_path:
+            plt.savefig(save_path, dpi=300, bbox_inches="tight")
+            logger.info(f"Saved plot to {save_path}")
+        else:
+            plt.show()
+
+        plt.close()
+
+    @staticmethod
+    def plot_convergence(history: List[dict], save_path: Optional[str] = None) -> None:
+        """
+        Plot optimization convergence over generations.
+
+        Shows how Pareto front size and quality evolve.
+
+        Args:
+            history: Optimization history from NSGA2Optimizer
+            save_path: Optional path to save figure
+
+        Example:
+            >>> history = optimizer.get_history()
+            >>> ParetoVisualizer.plot_convergence(history)
+        """
+        try:
+            import matplotlib.pyplot as plt
+        except ImportError:
+            logger.warning("matplotlib not available, cannot plot")
+            return
+
+        if not history:
+            logger.warning("No history to plot")
+            return
+
+        generations = [h["generation"] for h in history]
+        pareto_sizes = [h["pareto_size"] for h in history]
+
+        plt.figure(figsize=(10, 6))
+        plt.plot(generations, pareto_sizes, "b-o", linewidth=2, markersize=6)
+
+        plt.xlabel("Generation", fontsize=14, fontweight="bold")
+        plt.ylabel("Pareto Front Size", fontsize=14, fontweight="bold")
+        plt.title("NSGA-II Convergence", fontsize=16, fontweight="bold")
+        plt.grid(True, alpha=0.3)
+        plt.tight_layout()
+
+        if save_path:
+            plt.savefig(save_path, dpi=300, bbox_inches="tight")
+            logger.info(f"Saved convergence plot to {save_path}")
+        else:
+            plt.show()
+
+        plt.close()
+
+    @staticmethod
+    def plot_objective_distribution(
+        pareto_front: List[MultiObjectiveIndividual],
+        objective_name: str,
+        title: Optional[str] = None,
+        save_path: Optional[str] = None,
+    ) -> None:
+        """
+        Plot distribution of a single objective across Pareto front.
+
+        Args:
+            pareto_front: List of Pareto-optimal individuals
+            objective_name: Objective to visualize
+            title: Optional plot title
+            save_path: Optional path to save figure
+        """
+        try:
+            import matplotlib.pyplot as plt
+        except ImportError:
+            logger.warning("matplotlib not available, cannot plot")
+            return
+
+        values = [ind.objectives[objective_name] for ind in pareto_front]
+
+        fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(14, 5))
+
+        # Histogram
+        ax1.hist(values, bins=20, alpha=0.7, color="blue", edgecolor="black")
+        ax1.set_xlabel(objective_name.replace("_", " ").title(), fontsize=12)
+        ax1.set_ylabel("Count", fontsize=12)
+        ax1.set_title("Distribution", fontsize=14, fontweight="bold")
+        ax1.grid(True, alpha=0.3, axis="y")
+
+        # Box plot
+        ax2.boxplot(
+            values, vert=True, patch_artist=True, boxprops={"facecolor": "lightblue", "alpha": 0.7}
+        )
+        ax2.set_ylabel(objective_name.replace("_", " ").title(), fontsize=12)
+        ax2.set_title("Box Plot", fontsize=14, fontweight="bold")
+        ax2.grid(True, alpha=0.3, axis="y")
+
+        if title:
+            fig.suptitle(title, fontsize=16, fontweight="bold")
+
+        plt.tight_layout()
+
+        if save_path:
+            plt.savefig(save_path, dpi=300, bbox_inches="tight")
+            logger.info(f"Saved distribution plot to {save_path}")
+        else:
+            plt.show()
+
+        plt.close()
+
+    @staticmethod
+    def plot_tradeoff_matrix(
+        pareto_front: List[MultiObjectiveIndividual],
+        objective_names: Optional[List[str]] = None,
+        save_path: Optional[str] = None,
+    ) -> None:
+        """
+        Plot pairwise trade-off matrix for all objectives.
+
+        Creates a grid of scatter plots showing relationships between
+        all pairs of objectives.
+
+        Args:
+            pareto_front: List of Pareto-optimal individuals
+            objective_names: List of objectives (None = all)
+            save_path: Optional path to save figure
+
+        Example:
+            >>> ParetoVisualizer.plot_tradeoff_matrix(pareto_front)
+        """
+        try:
+            import matplotlib.pyplot as plt
+        except ImportError:
+            logger.warning("matplotlib not available, cannot plot")
+            return
+
+        if objective_names is None:
+            objective_names = list(pareto_front[0].objectives.keys())
+
+        n = len(objective_names)
+        fig, axes = plt.subplots(n, n, figsize=(4 * n, 4 * n))
+
+        for i, obj1 in enumerate(objective_names):
+            for j, obj2 in enumerate(objective_names):
+                ax = axes[i, j] if n > 1 else axes
+
+                if i == j:
+                    # Diagonal: histogram
+                    values = [ind.objectives[obj1] for ind in pareto_front]
+                    ax.hist(values, bins=15, alpha=0.7, color="blue", edgecolor="black")
+                    ax.set_ylabel("Count", fontsize=10)
+                else:
+                    # Off-diagonal: scatter plot
+                    x = [ind.objectives[obj2] for ind in pareto_front]
+                    y = [ind.objectives[obj1] for ind in pareto_front]
+                    ax.scatter(x, y, s=50, alpha=0.6, c="blue", edgecolors="black")
+
+                # Labels
+                if i == n - 1:
+                    ax.set_xlabel(obj2.replace("_", " ").title(), fontsize=10)
+                if j == 0:
+                    ax.set_ylabel(obj1.replace("_", " ").title(), fontsize=10)
+
+                ax.grid(True, alpha=0.3)
+
+        plt.suptitle("Objective Trade-off Matrix", fontsize=18, fontweight="bold")
+        plt.tight_layout()
+
+        if save_path:
+            plt.savefig(save_path, dpi=300, bbox_inches="tight")
+            logger.info(f"Saved trade-off matrix to {save_path}")
+        else:
+            plt.show()
+
+        plt.close()
+
+
+# Convenience functions
+def quick_visualize_2d(pareto_front: List[MultiObjectiveIndividual], obj1: str, obj2: str) -> None:
+    """Quick 2D visualization."""
+    ParetoVisualizer.plot_2d(pareto_front, obj1, obj2)
+
+
+def quick_visualize_3d(
+    pareto_front: List[MultiObjectiveIndividual], obj1: str, obj2: str, obj3: str
+) -> None:
+    """Quick 3D visualization."""
+    ParetoVisualizer.plot_3d(pareto_front, obj1, obj2, obj3)