-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdiscomfort_constrained_opt_pdcbo.py
148 lines (133 loc) · 5.86 KB
/
discomfort_constrained_opt_pdcbo.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
import copy
import os
import pickle
import datetime
from tune_util import get_vacbo_optimizer
# parameter configurations to enumerate
discomfort_thr_list = [5 * (k+1) for k in range(7)] #[5, 10, 15, 20]
os.environ['KMP_DUPLICATE_LIB_OK'] = 'True'
optimization_config = {
'eval_budget': 300
}
optimizer_base_config = {
'noise_level': [0.002/5, 0.75/5, 1.39/5],
'kernel_var': 1.0,
'problem_name': 'SinglePIRoomEvaluator',
'normalize_input': False
}
VARS_TO_FIX = ['high_off_time', 'low_setpoint',
'control_setpoint']
CONTEXTUAL_VARS = ['Q_irr', 'T_out', 'T_init']
tune_var_scale = 'log'
save_name_append = f'_{tune_var_scale}_with_context'
TUNE_OBJ = 'energy'
ave_discomfort_range = (0, 20)
ave_energy_range = (16, 23)
discomfort_weight = 0.0
discomfort_thr_to_ave_energy_pdcbo = dict()
discomfort_thr_to_ave_discomfort_pdcbo = dict()
discomfort_thr_to_energy_pdcbo = dict()
discomfort_thr_to_discomfort_pdcbo = dict()
discomfort_thr_to_history_dict_pdcbo = dict()
discomfort_thr_to_contexts_pdcbo = dict()
def run_opt(bo_config, optimizer_type):
discomfort_thr_to_ave_energy = dict()
discomfort_thr_to_ave_discomfort = dict()
discomfort_thr_to_energy = dict()
discomfort_thr_to_discomfort = dict()
discomfort_thr_to_history_dict = dict()
discomfort_thr_to_contexts_dict = dict()
for discomfort_thr in discomfort_thr_list:
opt, opt_total_cost_list, opt_problem = get_vacbo_optimizer(
bo_config['problem_name'], optimizer_type, bo_config,
discomfort_thr=discomfort_thr, vars_to_fix=VARS_TO_FIX,
contextual_vars=CONTEXTUAL_VARS,
discomfort_weight=discomfort_weight, tune_obj=TUNE_OBJ)
opt_obj_list = []
constraints_list = []
energy_list = []
discomfort_list = []
average_energy_list = []
average_discomfort_list = []
contexts_list = []
for _ in range(optimization_config['eval_budget']):
context_vars = opt_problem.get_context(opt_problem.simulator)
y_obj, constr_vals = opt.make_step(context_vars)
if optimizer_type == 'safe_bo':
new_cumu_cost = opt.safe_bo.cumu_vio_cost
opt_problem = opt.safe_bo.opt_problem
if optimizer_type == 'constrained_bo':
new_cumu_cost = opt.constrained_bo.cumu_vio_cost
opt_problem = opt.constrained_bo.opt_problem
if optimizer_type == 'violation_aware_bo':
new_cumu_cost = opt.violation_aware_bo.cumu_vio_cost
opt_problem = opt.violation_aware_bo.opt_problem
if optimizer_type == 'pdcbo':
new_cumu_cost = opt.pdbo.cumu_vio_cost
opt_problem = opt.pdbo.opt_problem
if optimizer_type == 'no opt':
new_cumu_cost = opt.cumu_vio_cost
opt_problem = opt.opt_problem
if optimizer_type == 'grid search':
new_cumu_cost = opt.cumu_vio_cost
opt_problem = opt.opt_problem
opt_total_cost_list.append(new_cumu_cost)
opt_obj_list.append(y_obj)
constraints_list.append(constr_vals)
energy, discomfort = \
opt_problem.simulator.get_recent_energy_discomfort_per_day()
contexts_list.append(context_vars)
energy_list.append(energy)
discomfort_list.append(discomfort)
print_log = True
if print_log:
print(f"For {opt_problem.problem_name}, in step {_}, " +
"with discomfort threshold " +
f"{discomfort_thr}, we get energy {energy}" +
f" and discomfort {discomfort}, with the point "
+ f" {opt_problem.evaluated_points_list[-1]}.")
average_energy_list.append(
opt_problem.simulator.cumulative_energy / 96.0 /
len(opt_problem.evaluated_constrs_list)
)
average_discomfort_list.append(
opt_problem.simulator.cumulative_discomfort * 0.25 /
len(opt_problem.evaluated_constrs_list)
)
discomfort_thr_to_ave_discomfort[discomfort_thr] = \
average_discomfort_list
discomfort_thr_to_ave_energy[discomfort_thr] = average_energy_list
discomfort_thr_to_discomfort[discomfort_thr] = discomfort_list
discomfort_thr_to_energy[discomfort_thr] = energy_list
discomfort_thr_to_history_dict[discomfort_thr] = \
opt_problem.simulator.history_dict
discomfort_thr_to_contexts_dict[discomfort_thr] = \
contexts_list
return discomfort_thr_to_discomfort, discomfort_thr_to_energy, \
discomfort_thr_to_ave_discomfort, discomfort_thr_to_ave_energy, \
discomfort_thr_to_history_dict, discomfort_thr_to_contexts_dict
# run PDCBO
pdcbo_config = copy.deepcopy(optimizer_base_config)
pdcbo_config.update({
'eta_0': 1.0 / 300.0,
'eta_func': lambda t: 1.0 / 300.0,
'total_eval_num': optimization_config['eval_budget'],
'init_dual': 10.0 / 300.0,
'lcb_coef': lambda t: 1.0 # 1e-6
})
optimizer_type = 'pdcbo'
discomfort_thr_to_discomfort_pdcbo, discomfort_thr_to_energy_pdcbo, \
discomfort_thr_to_ave_discomfort_pdcbo, \
discomfort_thr_to_ave_energy_pdcbo, discomfort_thr_to_history_dict_pdcbo, \
discomfort_thr_to_contexts_pdcbo = run_opt(pdcbo_config, optimizer_type)
now_time_str = datetime.datetime.now().strftime(
"%H_%M_%S-%b_%d_%Y")
with open(f'./result/discomfort_constrained_energy_min_pdcbo_{now_time_str}.pkl',
'wb') as f:
pickle.dump([
discomfort_thr_to_ave_energy_pdcbo,
discomfort_thr_to_ave_discomfort_pdcbo,
discomfort_thr_to_energy_pdcbo,
discomfort_thr_to_discomfort_pdcbo,
discomfort_thr_to_contexts_pdcbo
], f)