Clean up

Saurav Agarwal · Saurav Agarwal · commit 8ee545b3d613 · 2024-09-07T21:44:56.000-04:00
diff --git a/python/utils/plot_constrained_learning_results.py b/python/utils/plot_constrained_learning_results.py
@@ -7,46 +7,35 @@
 plt.rc('text', usetex=True)
 plt.rc('font', family='serif')
 
-# Define the parameters
 # eta_duals = [0.1, 1.0, 10.0, 100.0]
 # T_0s = [25, 50, 75, 100]
 eta_duals = [1]
 T_0s = [25]
 envs = list(range(100))  # Assuming this corresponds to the environment IDs
 eval_dir = sys.argv[1]  # Path to the directory containing the evaluation results
 
-# Set fixed axis ranges
-x_axis_range = (0, 3000)  # Assuming 3000 columns as per your description
-y_axis_range = (0, .3)  # Adjust this based on the expected range of max values
+# x_axis_range = (0, 3000)
+y_axis_range = (0, .30)
 
-# Initialize a figure for the grid of plots
 fig, axes = plt.subplots(len(eta_duals), len(T_0s), figsize=(24, 24))
 # fig.suptitle('Max Objective Values for Different $\\eta$ and $T_0$', fontsize=24)
 
-# Iterate over eta_duals and T_0s to read data and plot
 for i, eta_dual in enumerate(eta_duals):
     for j, T_0 in enumerate(T_0s):
-        # Initialize a list to store max values for each environment
         all_max_values = []
         all_obj_values = []
         
         for env_id in envs:
-            # Construct the file path
             file_path = f"{eval_dir}/{eta_dual}_{T_0}/obj_values_{env_id}.csv"
             
-            # Check if the file exists
             if os.path.exists(file_path):
-                # Load the obj_values from the CSV file
                 obj_values = np.loadtxt(file_path, delimiter=",")
                 all_obj_values.append(obj_values)
                 
-                # Compute the max value for each column
                 max_values = np.max(obj_values, axis=0)
                 
-                # Append to the list of max values
                 all_max_values.append(max_values)
                 
-                # Plot individual environment max values with light, thin lines
                 if len(eta_duals) > 1 and len(T_0s) > 1:
                     ax = axes[i, j]
                 elif len(eta_duals) > 1:
@@ -57,24 +46,20 @@
                     ax = axes
                 ax.plot(max_values, color='lightgray', linewidth=0.5)
         
-        # Plot the average max values across all environments
-        if all_max_values:
-            avg_max_values = np.mean(all_max_values, axis=0)
-            median_max_values = np.median(all_max_values, axis=0)
-            interquartile_range = np.percentile(all_max_values, 75, axis=0) - np.percentile(all_max_values, 25, axis=0)
-            ax.plot(avg_max_values, linewidth=2, label='Average')
-            ax.plot(median_max_values, color='red', linewidth=2, label='Median')
-            ax.fill_between(range(len(avg_max_values)), median_max_values - interquartile_range, median_max_values + interquartile_range, color='red', alpha=0.2, label='Interquartile Range')
-            ax.legend(fontsize=20)
+        avg_max_values = np.mean(all_max_values, axis=0)
+        print("Average max values at end: ", np.mean(avg_max_values))
+        median_max_values = np.median(all_max_values, axis=0)
+        print("Median max values at end: ", median_max_values[-1])
+        interquartile_range = np.percentile(all_max_values, 75, axis=0) - np.percentile(all_max_values, 25, axis=0)
+        ax.plot(avg_max_values, linewidth=2, label='Average')
+        ax.plot(median_max_values, color='red', linewidth=2, label='Median')
+        ax.fill_between(range(len(avg_max_values)), median_max_values - interquartile_range, median_max_values + interquartile_range, color='red', alpha=0.2, label='Interquartile Range')
+        ax.legend(fontsize=20)
 
-            # Set plot title and axis ranges
-            # ax.set_xlim(x_axis_range)
-            ax.set_ylim(y_axis_range)
-            ax.set_title(rf'$\eta$: {eta_dual}, $T_0$: {T_0}', fontsize=32)
+        ax.set_ylim(y_axis_range)
+        ax.set_title(rf'$\eta$: {eta_dual}, $T_0$: {T_0}', fontsize=32)
         ax.tick_params(axis='both', which='major', labelsize=30)
 
-# Adjust layout to prevent overlap
 plt.tight_layout()
 plt.subplots_adjust(top=0.95)
-# plt.show()
-plt.savefig(sys.argv[2])  # Save the figure to a file
+plt.savefig(sys.argv[2])
