Plot expectation value benchmarks

benchmarks/scripts/plot_expval_benchmarks_over_time.py

@@ -0,0 +1,73 @@
+import os
+import glob
+import pandas as pd
+import matplotlib.pyplot as plt
+
+# Step 1: Get the directory of the current script
+directory_of_this_file = os.path.dirname(os.path.abspath(__file__))
+
+# Step 2: Construct the correct path to the results folder
+results_folder = os.path.join(directory_of_this_file, "../results")
+
+# Step 3: Use glob to find all CSV files in the results folder
+csv_files = glob.glob(os.path.join(results_folder, "expval_*.csv"))
+
+# Step 4: Iterate through all CSV files and load them into dataframes with date column
+dfs = []  # List to store dataframes
+for file in csv_files:
+    # Extract the date from the filename (assuming the date is after 'expval_' and before '.csv')
+    date_label = str(file).split('_')[1].split('.')[0]
+
+    # Load the CSV file into a DataFrame
+    df = pd.read_csv(file, comment="#")
+
+    # Add the extracted date as a new column in the dataframe
+    df['date'] = date_label
+
+    # Append the dataframe to the list
+    dfs.append(df)
+
+# Step 5: Concatenate all dataframes into one large dataframe
+df_all = pd.concat(dfs, ignore_index=True)
+
+# Convert the 'date' column to datetime
+df_all['date'] = pd.to_datetime(df_all['date'])
+
+# Step 6: Group by date and compiler, and calculate the average relative error
+summary = df_all.groupby(['date', 'compiler']).agg(
+    avg_relative_error=('relative_error', 'mean')
+).reset_index()
+
+# Step 7: Set up the figure
+fig, ax = plt.subplots(figsize=(12, 6))
+
+# Set color map for different compilers
+unique_compilers = sorted(summary['compiler'].unique())
+colormap = plt.get_cmap("tab10", len(unique_compilers))
+color_map = {compiler: colormap(i) for i, compiler in enumerate(unique_compilers)}
+
+# Step 8: Plot average relative error over time
+for compiler in unique_compilers:
+    compiler_data = summary[summary['compiler'] == compiler]
+
+    # Plot average relative error
+    ax.plot(
+        compiler_data['date'], 
+        compiler_data['avg_relative_error'], 
+        label=compiler, 
+        color=color_map[compiler], 
+        marker='o'
+    )
+
+# Step 9: Customize plot
+ax.set_title("Average Relative Error Over Time")
+ax.set_xlabel("Date")
+ax.set_ylabel("Average Relative Error")
+ax.grid(True)
+ax.legend(title="Compiler")
+
+# Adjust layout and save the figure
+plt.tight_layout()
+filename = os.path.join(directory_of_this_file, "../average_relative_error_over_time.png")
+print(f"\n Saving plot to {filename}")
+fig.savefig(filename)

benchmarks/scripts/plot_latest_expval_benchmarks_per_circuit.py

@@ -0,0 +1,105 @@
+import os
+import glob
+import pandas as pd
+import matplotlib.pyplot as plt
+
+# Step 1: Get the directory of the current script
+directory_of_this_file = os.path.dirname(os.path.abspath(__file__))
+
+# Step 2: Construct the correct path to the results folder
+results_folder = os.path.join(directory_of_this_file, "../results")
+
+# Step 3: Use glob to find all CSV files in the results folder
+csv_files = glob.glob(os.path.join(results_folder, "expval_*.csv"))
+
+# Step 4: Iterate through all CSV files and load them into dataframes with date column
+dfs = []  # List to store dataframes
+for file in csv_files:
+    # Extract the date from the filename (assuming the date is after 'expval_' and before '.csv')
+    date_label = str(file).split('_')[1].split('.')[0]
+
+    # Load the CSV file into a DataFrame
+    df = pd.read_csv(file, comment="#")
+
+    # Add the extracted date as a new column in the dataframe
+    df['date'] = date_label
+
+    # Append the dataframe to the list
+    dfs.append(df)
+
+# Step 5: Concatenate all dataframes into one large dataframe
+df_all = pd.concat(dfs, ignore_index=True)
+
+# Find the most recent date in the 'date' column
+latest_date = df_all['date'].max()
+print("latest date is", latest_date)
+
+# Filter the dataframe to only include rows from the most recent date
+df_latest = df_all[df_all['date'] == latest_date]
+
+# Step 6: Define the bar width and create x-axis positions for the circuits
+bar_width = 0.2
+circuit_names = df_latest['circuit_name'].unique()
+x_positions = range(len(circuit_names))  # X positions for each circuit
+
+# Create a dictionary to map circuit names to indices
+circuit_name_to_index = {name: i for i, name in enumerate(circuit_names)}
+
+# Step 7: Set up the figure and axes
+fig, ax = plt.subplots(1, 2, figsize=(14, 7))
+
+# Set color map for different compilers
+# Get unique compilers and sort them alphabetically
+unique_compilers = sorted(df_latest['compiler'].unique())
+
+# Set color map for different compilers
+colormap = plt.get_cmap("tab10", len(unique_compilers))
+color_map = {compiler: colormap(i) for i, compiler in enumerate(unique_compilers)}
+
+# Step 8: Plot relative error and absolute error for each compiler
+for i, (key, grp) in enumerate(df_latest.groupby("compiler")):
+    # Get indices for each circuit in the current compiler group
+    grp_indices = grp['circuit_name'].map(circuit_name_to_index)
+
+    # Plot relative error
+    ax[0].bar(
+        [grp_indices + i * bar_width for grp_indices in grp_indices],  # Shift bars for each compiler
+        grp['relative_error'],  # Relative error data
+        width=bar_width,
+        label=key,
+        color=color_map[key]
+    )
+
+    # Plot absolute error
+    ax[1].bar(
+        [grp_indices + i * bar_width for grp_indices in grp_indices],  # Shift bars for each compiler
+        grp['absoluate_error'],  # Absolute error data
+        width=bar_width,
+        label=key,
+        color=color_map[key]
+    )
+
+# Step 9: Customize plots
+ax[0].set_title(f"Relative Error by Compiler on Circuits (Date: {latest_date})")
+ax[0].set_xlabel("Circuit Name")
+ax[0].set_ylabel("Relative Error")
+ax[0].set_xticks(x_positions)
+ax[0].set_xticklabels(circuit_names, rotation=75)
+ax[0].set_yscale("log")
+
+ax[1].set_title(f"Absolute Error by Compiler on Circuits (Date: {latest_date})")
+ax[1].set_xlabel("Circuit Name")
+ax[1].set_ylabel("Absolute Error")
+ax[1].set_xticks(x_positions)
+ax[1].set_xticklabels(circuit_names, rotation=75)
+ax[1].set_yscale("log")
+
+# Step 10: Add legend
+ax[0].legend(title="Compiler")
+ax[1].legend(title="Compiler")
+
+# Adjust layout and save the figure
+plt.tight_layout()
+filename = os.path.join(directory_of_this_file, "../latest_relative_absolute_errors_by_circuit.png")
+print(f"\n Saving plot to {filename}")
+fig.savefig(filename)