transfer test code

cfg/airsim_dagger.cfg

@@ -15,15 +15,15 @@ test_interval = 40
 n_test_episodes = 20
 
 # architecture parameters
-k = 3
+k = 2
 hidden_size = 32
 gamma = 0.99
 tau = 0.5
 
 # env parameters
-env = FlockingAirsim-v0
-v_max = 2.0
-comm_radius = 1.25
+env = FlockingAirsimAccel-v0
+v_max = 0.5
+comm_radius = 1.5
 n_agents = 50
 n_actions = 2
 n_states = 6

cfg/dagger_stoch.cfg

@@ -15,7 +15,7 @@ test_interval = 40
 n_test_episodes = 20
 
 # architecture parameters
-k = 2
+k = 4
 hidden_size = 32
 gamma = 0.99
 tau = 0.1

cfg/default.cfg

@@ -28,4 +28,5 @@ n_agents = 100
 n_actions = 2
 n_states = 6
 debug = False
+dt = 0.01
 

cfg/default_baseline.cfg

@@ -19,6 +19,7 @@ k = 2
 hidden_size = 32
 gamma = 0.99
 tau = 0.5
+dt = 0.01
 
 # env parameters
 env = FlockingRelative-v0
@@ -28,4 +29,5 @@ n_agents = 100
 n_actions = 2
 n_states = 6
 debug = False
+dt = 0.01
 

cfg/dt.cfg

@@ -0,0 +1,131 @@
+[DEFAULT]
+
+alg = dagger
+
+# learning parameters
+batch_size = 20
+buffer_size = 10000
+updates_per_step = 200
+seed = 11
+actor_lr = 5e-5
+
+n_train_episodes = 800
+beta_coeff = 0.993
+test_interval = 40
+n_test_episodes = 20
+
+# architecture parameters
+k = 2
+hidden_size = 32
+gamma = 0.99
+tau = 0.5
+
+# env parameters
+env = FlockingRelative-v0
+v_max = 3.0
+comm_radius = 1.0
+n_agents = 100
+n_actions = 2
+n_states = 6
+debug = False
+dt = 0.01
+
+
+header = k, dt, reward
+
+[1, 0.1]
+k = 1
+dt = 0.1
+
+[1, 0.075]
+k = 1
+dt = 0.075
+
+[1, 0.05]
+k = 1
+dt = 0.05
+
+[1, 0.025]
+k = 1
+dt = 0.025
+
+[1, 0.01]
+k = 1
+dt = 0.01
+
+[1, 0.0075]
+k = 1
+dt = 0.0075
+
+[2, 0.1]
+k = 2
+dt = 0.1
+
+[2, 0.075]
+k = 2
+dt = 0.075
+
+[2, 0.05]
+k = 2
+dt = 0.05
+
+[2, 0.025]
+k = 2
+dt = 0.025
+
+[2, 0.01]
+k = 2
+dt = 0.01
+
+[2, 0.0075]
+k = 2
+dt = 0.0075
+
+[3, 0.1]
+k = 3
+dt = 0.1
+
+[3, 0.075]
+k = 3
+dt = 0.075
+
+[3, 0.05]
+k = 3
+dt = 0.05
+
+[3, 0.025]
+k = 3
+dt = 0.025
+
+[3, 0.01]
+k = 3
+dt = 0.01
+
+[3, 0.0075]
+k = 3
+dt = 0.0075
+
+[4, 0.1]
+k = 4
+dt = 0.1
+
+[4, 0.075]
+k = 4
+dt = 0.075
+
+[4, 0.05]
+k = 4
+dt = 0.05
+
+[4, 0.025]
+k = 4
+dt = 0.025
+
+[4, 0.01]
+k = 4
+dt = 0.01
+
+[4, 0.0075]
+k = 4
+dt = 0.0075
+

cfg/dt_baseline.cfg

@@ -0,0 +1,83 @@
+[DEFAULT]
+
+alg = baseline
+
+# learning parameters
+batch_size = 20
+buffer_size = 10000
+updates_per_step = 200
+seed = 11
+actor_lr = 5e-5
+
+n_train_episodes = 800
+beta_coeff = 0.993
+test_interval = 40
+n_test_episodes = 20
+
+# architecture parameters
+k = 2
+hidden_size = 32
+gamma = 0.99
+tau = 0.5
+
+# env parameters
+env = FlockingRelative-v0
+v_max = 3.0
+comm_radius = 1.0
+n_agents = 100
+n_actions = 2
+n_states = 6
+debug = False
+dt = 0.01
+
+
+header = dt, centralized, reward
+
+[0.1, True]
+dt = 0.1
+centralized = True
+
+[0.1, False]
+dt = 0.1
+centralized = False
+
+[0.075, True]
+dt = 0.075
+centralized = True
+
+[0.075, False]
+dt = 0.075
+centralized = False
+
+[0.05, True]
+dt = 0.05
+centralized = True
+
+[0.05, False]
+dt = 0.05
+centralized = False
+
+[0.025, True]
+dt = 0.025
+centralized = True
+
+[0.025, False]
+dt = 0.025
+centralized = False
+
+[0.01, True]
+dt = 0.01
+centralized = True
+
+[0.01, False]
+dt = 0.01
+centralized = False
+
+[0.0075, True]
+dt = 0.0075
+centralized = True
+
+[0.0075, False]
+dt = 0.0075
+centralized = False
+

cfg/make_config.py

@@ -1,9 +1,36 @@
 import itertools
 
-default_fname = "default_baseline.cfg"
+baseline = False
+baseline = True
+param = 'vel'
+param = 'dt'
+# param = 'n'
+# param = 'rad'
+
+params = {}
+
+# params['seed'] = range(10)
+
+if baseline:
+    default_fname = "default_baseline.cfg"
+    out_fname = param + '_baseline.cfg'
+
+    params['centralized'] = ['True', 'False']
+else:
+    default_fname = 'default.cfg'
+    out_fname = param + '.cfg'
+
+    params['k'] = [1, 2, 3, 4]
+
+if param == 'vel':
+    params['v_max'] = [0.5, 1.5, 2.5, 3.5, 4.5]
+elif param == 'rad':
+    params['comm_radius'] = [3.0, 2.5, 2.0, 1.5, 1.0]
+elif param == 'n':
+    params['n_agents'] = [25, 50, 75, 100, 125, 150, 175, 200]
+elif param == 'dt':
+    params['dt'] = [0.1, 0.075, 0.05, 0.025, 0.01, 0.0075]
 
-out_fname = 'vel_baseline.cfg'
-# out_fname = 'n_baseline.cfg'
 
 out_file = open(out_fname, "w")
 
@@ -13,13 +40,6 @@
 
 out_file.write('\n')
 
-params = {}
-params['centralized'] = ['True', 'False']
-params['seed'] = range(10)
-# params['n_agents'] = [20, 40, 80, 100]
-# params['comm_radius'] = [3.0, 2.0, 1.5, 1.0]
-params['v_max'] = [0.5, 1.5, 2.5, 3.5]
-
 param_names = params.keys()
 param_values = params.values()