final before initial submission

Author: dantzert

baseline_controllers/gamma/optimize_parameters.py

@@ -722,6 +722,8 @@ def create_bo_model(data):
     lower_bounds = [2.5] * 9  # Nine parameters for gamma
     upper_bounds = [6.0] * 9
     search_space = Box(lower_bounds, upper_bounds)
+    init_space = Box([4.0] * 9, [5.0] * 9)  # Initial space for sampling
+    # a harder and higher dimensional problem, try to make sure the initial points are mostly feasible.
 
     # Import required libraries for additional optimization methods
     from scipy.optimize import dual_annealing, differential_evolution
@@ -742,7 +744,7 @@ def combined_objective(params):
         res = evaluate_cf_point(params_array)
         # If the constraint is violated, return a large penalty
         if res['constraint'] > Sim_cf.threshold:
-            return 1e6*(res['constraint'] - Sim_cf.threshold) # if all initial samples have the same value some methods will error out.
+            return 1e10#*(res['constraint'] - Sim_cf.threshold) # if all initial samples have the same value some methods will error out.
         return float(res['objective'])
 
         '''
@@ -761,13 +763,14 @@ def combined_objective(params):
 
     # Generate common initial points for all methods
     num_initial_points = 5#25  # Number of initial points
-    num_steps = 5#25  
+    num_steps = 25#25  
     initial_seed = 42  # Use fixed seed for reproducibility
     tf.random.set_seed(initial_seed)
     np.random.seed(initial_seed)
 
     # Generate initial points that all methods will use
-    initial_points = search_space.sample(num_initial_points)
+    #initial_points = search_space.sample(num_initial_points)
+    initial_points = init_space.sample(num_initial_points)
     initial_points_np = initial_points.numpy()
 
     # 1. Bayesian Optimization with Unknown Constraints (BOUC) using Trieste
@@ -840,6 +843,7 @@ def combined_objective(params):
 
     print("BOUC function calls:", bouc_fcalls_arr)
     print("BOUC combined costs:", bouc_combined_costs)
+    print("BOUC best cost:", results["BOUC"]["best_cost"])
 
     # 2. Vanilla Bayesian Optimization using Trieste
     print("\nRunning vanilla BO optimization...")
@@ -928,8 +932,8 @@ def da_callback(x, f, context):
         return False
 
     # Run optimization
-    da_max_calls = num_steps * 2
-    da_max_iter = num_steps // 5
+    da_max_calls = int(num_steps * 1.2)
+    da_max_iter = num_steps // 7
 
     da_start_time = time.time()
     res_da = dual_annealing(combined_objective, bounds=bounds_da, 
@@ -982,11 +986,11 @@ def de_callback(x, convergence):
         return False
 
     # Create initial population that includes our initial points
-    popsize = num_initial_points
+    popsize = num_initial_points 
     population = np.array(initial_points_np)
 
     # Set maxiter to ensure comparable number of function evaluations
-    de_max_iter = 2*max(1, num_steps // (popsize * len(lower_bounds)))
+    de_max_iter = 3*max(1, num_steps // (popsize * len(lower_bounds)))
     print(f"DE max iterations: {de_max_iter}, popsize: {popsize}, function evaluations: {(de_max_iter + 1) * popsize * (len(lower_bounds) - 1)}")
 
     de_start_time = time.time()
@@ -1138,15 +1142,15 @@ def rolling_min(costs, feasible=None):
             min_cost_all = min(results[method]["best_cost_so_far"])
 
     # Set y-limits from slightly below best cost to twice the best cost
-    plt.ylim(0.95 * min_cost_all, 2.0 * min_cost_all)
+    #plt.ylim(0.95 * min_cost_all, 2.0 * min_cost_all)
 
     for method, data in results.items():
         if data["fcalls"] and data["best_cost_so_far"]:
             plt.plot(data["fcalls"], data["best_cost_so_far"], 'o-', label=labels[method], markersize=8)
 
     plt.xlabel('Function Evaluations', fontsize=14)
     plt.ylabel('Best Cost Found', fontsize=14)
-    plt.title('Optimization Methods Comparison (Zoomed) - Gamma Scenario', fontsize=16)
+    plt.title('Optimization Methods Comparison - Gamma Scenario', fontsize=16)
     plt.grid(True, alpha=0.3)
     plt.legend(fontsize='x-large')
     plt.tight_layout()

baseline_controllers/gamma/v2/gamma-optim-compare.png

@@ -1,5 +1,5 @@
 method,best_cost,optimization_time_sec,num_function_calls,param_1,param_2,param_3,param_4,param_5,param_6,param_7,param_8,param_9
-BOUC,99290.67299849597,34477.51378130913,50,3.6077411389789056,5.3807122663141955,3.442493276477518,5.7232907117410665,5.337139884533713,4.190866514161556,2.729240969602435,4.5890491048784785,2.7649846856003766
-BO,98377.50329707502,535.5013947486877,50,3.931620357112107,5.10573759247095,3.6278875241137896,4.84199169640375,4.3314561496114035,2.5151946820151045,4.246216602957377,2.6639378348946607,5.570273166335934
-DA,68970.11886400828,897.9846568107605,75,3.262722931802273,4.638648509979248,3.1899866349995136,5.378272503614426,4.7226347625255585,3.721466556424275,5.884222760796547,2.5597217231988907,5.8646092265844345
-DE,87445.70084261333,954.190426826477,100,3.2419585373433937,3.9608576461699503,3.395929034032598,5.83247917716061,3.1556220277169844,2.5558112183094126,2.8870635771375257,4.553924883217957,5.1252473575821496
+BOUC,104791.93504003703,320.93161702156067,30,3.3439470709496613,4.907406024606427,3.8277356191137515,4.837456152651078,4.112413454940194,3.7550806440028617,4.338703109663414,4.458628065967337,4.24917181869433
+BO,119382.23813566855,263.12797451019287,30,4.2625012853156115,5.4209690750371715,3.6110748786739846,5.101152909223343,4.432456661682204,3.6704665958157356,4.463743370250949,4.187103050937975,3.2958336045742427
+DA,129356.94544444678,216.78493905067444,35,4.225781738758087,5.694479823112488,4.269283793279291,4.629364438354969,3.772303983569145,4.897373799933121,3.7375309988856316,2.6554581969976425,4.075709910171536
+DE,136640.54275972047,108.53766369819641,25,4.113463270214174,5.648448322023452,4.190402200035405,4.512789276829983,5.086862392526488,4.533943370949488,2.728116294254682,4.366053824488603,4.915519192913303