script update

scripts/gen_values.py

@@ -11,7 +11,7 @@ def generate_test_values(noise_bits, number, p_bits):
 
     max_noise = (1 << noise_bits) - 1  # 2^noise_bits - 1
 
-    a = [str(p * random.randint(1, 2) + random.randint(0, max_noise)) for _ in range(number)]
+    a = [str(p * random.randint(1, max_noise) + random.randint(0, max_noise)) for _ in range(number)]
 
     return noise_bits, a, p
 

scripts/script.py

@@ -6,74 +6,108 @@ import tempfile
 import os
 import numpy as np
 import matplotlib.pyplot as plt
+import time
+import math
 from gen_values import generate_test_file
 
 def run_agcd(input_file):
     cmd = ["./target/release/approximate-gcd", "agcd", input_file]
-
     try:
+        start = time.perf_counter()
         result = subprocess.run(cmd, capture_output=True, text=True, check=True)
+        duration = time.perf_counter() - start
         output = result.stdout
 
         match = re.search(r"Recovered p: (\d+)", output)
-        return int(match.group(1)) if match else None
+        return int(match.group(1)) if match else None, duration
 
     except subprocess.CalledProcessError as e:
         print(f"Error running command for input_file={input_file}: {e}")
-        return None
+        return None, 0.0
     except (AttributeError, ValueError) as e:
         print(f"Error parsing output for input_file={input_file}: {e}")
-        return None
+        return None, 0.0
 
-def plot_curves(noise_bits, p_bits, test_numbers, success_rates):
-    plt.figure(figsize=(10, 6))
-    plt.plot(test_numbers, success_rates, marker='o')
-    plt.xlabel('Number of Test Values')
-    plt.ylabel('Success Rate')
-    plt.title(f'Success Rate vs. Number of Test Values\n(noise_bits={noise_bits}, p_bits={p_bits})')
-    plt.grid(True)
-    plt.ylim(-0.1, 1.1)
-    plt.xticks(test_numbers)
-    plt.savefig('success_rate_plot.png')
+def plot_all(noise_bits, p_bits, test_numbers, success_rates, mean_distances, mean_times):
+    num_points = len(test_numbers)
+    max_ticks = 20
+    step = max(1, num_points // max_ticks)
+    xticks = list(test_numbers)[::step]
+    if test_numbers[-1] not in xticks:
+        xticks.append(test_numbers[-1])
+
+    fig, axs = plt.subplots(3, 1, figsize=(10, 15), sharex=True)
+
+    # Success rate plot
+    axs[0].plot(test_numbers, success_rates, linestyle='-')
+    axs[0].set_ylabel('Success Rate')
+    axs[0].set_ylim(-0.1, 1.1)
+    axs[0].grid(True)
+    axs[0].set_title(f'Success Rate (noise_bits={noise_bits}, p_bits={p_bits})')
+
+    # Mean distance plot
+    axs[1].plot(test_numbers, mean_distances, linestyle='-')
+    axs[1].set_ylabel('Mean Distance to True p')
+    axs[1].grid(True)
+    axs[1].set_title(f'Mean Distance (noise_bits={noise_bits}, p_bits={p_bits})')
+
+    # Mean runtime plot
+    axs[2].plot(test_numbers, mean_times, linestyle='-')
+    axs[2].set_ylabel('Mean Runtime (s)')
+    axs[2].set_xlabel('Number of Test Values')
+    axs[2].grid(True)
+    axs[2].set_title(f'Mean Runtime (noise_bits={noise_bits}, p_bits={p_bits})')
+
+    axs[2].set_xticks(xticks)
+
+    fig.tight_layout()
+    plt.savefig('agcd_plots.png')
     plt.show()
 
 def main():
     parser = argparse.ArgumentParser(description='Test AGCD with varying number of test values.')
-    parser.add_argument('--noise_bits', type=int, default=0, help='Number of noise bits')
-    parser.add_argument('--p_bits', type=int, default=10000, help='Number of bits for p')
-    parser.add_argument('--min_values', type=int, default=2, help='Minimum number of test values')
-    parser.add_argument('--max_values', type=int, default=100, help='Maximum number of test values')
+    parser.add_argument('--noise-bits', type=int, default=1, help='Number of noise bits')
+    parser.add_argument('--p-bits', type=int, default=10000, help='Number of bits for p')
+    parser.add_argument('--min-values', type=int, default=2, help='Minimum number of test values')
+    parser.add_argument('--max-values', type=int, default=50, help='Maximum number of test values')
+    parser.add_argument('--trials', type=int, default=50, help='Number of trials per setting')
     args = parser.parse_args()
 
     noise_bits = args.noise_bits
     p_bits = args.p_bits
     test_numbers = range(args.min_values, args.max_values + 1)
+
     success_rates = []
-    num_trials = 100
+    mean_distances = []
+    mean_times = []
 
     for num_values in test_numbers:
         successes = 0
-        for _ in range(num_trials):
-            # Create temporary test file
+        distances = []
+        runtimes = []
+
+        for _ in range(args.trials):
             with tempfile.NamedTemporaryFile(mode='w', delete=False, suffix='.txt') as tmp_file:
                 true_p = generate_test_file(noise_bits, num_values, p_bits, tmp_file.name)
+            recovered_p, duration = run_agcd(tmp_file.name)
+            os.unlink(tmp_file.name)
 
-                # Run AGCD
-                recovered_p = run_agcd(tmp_file.name)
-
-                # Check if recovery was successful
-                if recovered_p is not None and abs(recovered_p-true_p) <= 2000:
+            if recovered_p is not None:
+                diff = abs(recovered_p - true_p)
+                threshold = math.isqrt(true_p.bit_length())
+                if diff <= threshold:
                     successes += 1
+                    distances.append(diff)
+                    runtimes.append(duration)
 
-                # Clean up
-                os.unlink(tmp_file.name)
+        success_rates.append(successes / args.trials)
+        mean_distances.append(np.mean(distances) if distances else float('nan'))
+        mean_times.append(np.mean(runtimes))
 
-        success_rate = successes / num_trials
-        success_rates.append(success_rate)
-        print(f"Number of values: {num_values}, Success rate: {success_rate:.3f} ({successes}/{num_trials})")
+        print(f"Values: {num_values}, Success rate: {success_rates[-1]:.3f} ({successes}/{args.trials}), "
+              f"Mean distance: {mean_distances[-1]:.2f}, Mean time: {mean_times[-1]:.4f}s")
 
-    # Plot the results
-    plot_curves(noise_bits, p_bits, test_numbers, success_rates)
+    plot_all(noise_bits, p_bits, list(test_numbers), success_rates, mean_distances, mean_times)
 
 if __name__ == "__main__":
     main()