svZeroDEstimator/Tune_NelderMead.py at main · ComBiNE-Fluid-Dynamics-Lab/svZeroDEstimator · GitHub

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
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
import numpy as np
#import json
from scipy.optimize import minimize

#import torch
#from sbi import utils as utils
#from sbi.utils.sbiutils import seed_all_backends
#from sbi.inference.base import infer
#from sbi.inference import SNPE, prepare_for_sbi, simulate_for_sbi
##from sbi.simulators.simutils import tqdm_joblib
##from tqdm.auto import tqdm
##from joblib import Parallel, delayed
##import pickle
#from multiprocessing import Pool

#   config_file = './svzerod_tuning.json'
#   f = open(config_file)
#   config = json.load(f)
#   params = np.genfromtxt('optParams.txt')

#   for i in range(100):
#       model = closed_loop_model.svZeroD_ClosedLoop(config)
#       frozen_param_idxs, variable_param_idxs = model.read_frozen_params('frozen_parameters.csv')
#       results = model.run_with_frozen_params(params[variable_param_idxs])


#   model = closed_loop_model.svZeroD_ClosedLoop(config)
#   for i in range(100):
#       frozen_param_idxs, variable_param_idxs = model.read_frozen_params('frozen_parameters.csv')
#       results = model.run_with_frozen_params(params[variable_param_idxs])

class Tune_NelderMead:

#   base_config = None
#   model_metadata = None
#   frozen_param_idxs = None
#   variable_param_idxs = None
#   frozen_param_values = None
    model = None
    num_restarts = None
    adaptive = False
    max_iter = None
    num_params = None
    num_results = None
    bounds = None
    initial_guess = None

#   def __init__(self, config_file):
#       f = open(config_file)
#       self.base_config = json.load(f)
#       model = closed_loop_model.svZeroD_ClosedLoop(self.base_config)
#       self.model_metadata = model.get_data_members()
#       self.frozen_param_idxs, self.variable_param_idxs, self.frozen_param_values = model.read_frozen_params('frozen_parameters.csv')
#       self.num_params = model.get_num_parameters()
#       self.num_results = model.get_num_results()

    def __init__(self, model, optimization_params):
        self.num_restarts = optimization_params["num_restarts"]
        self.max_iter = optimization_params["max_iterations"]
        self.convergence_tol = optimization_params.get("convergence_tol", 0.1)
        self.model = model
        self.num_params = model.num_parameters()
        self.num_results = model.num_results()
        self.bounds = model.parameter_limits_tuples()
        self.options = {}
        #self.options["bounds"] = self.bounds
        self.options["maxiter"] = self.max_iter
        self.options["adaptive"] = optimization_params.get("adaptive", False)
        self.options["disp"] = optimization_params.get("verbose", True)
        self.print_info()

    def print_info(self):
        print("--------------------------------")
        print("Running Nelder Mead optimization")
        print(f"Number of restarts: {self.num_restarts}")
        print(f"Max iterations: {self.max_iter}")
        print(f"Number of parameters: {self.num_params}")
        print("--------------------------------")

    def set_initial_guess(self, initial_params):
        if len(initial_params) != self.num_params:
            raise RuntimeError("len(initial_guess) != self.num_params")
        self.initial_guess = initial_params

    def run(self):
        init_guess = self.initial_guess
        for i_restart in range(self.num_restarts):
            print(f"\nRunning optimization round {i_restart+1}/{self.num_restarts} \n")
            print("Initial guess:")
            print(init_guess)
            res = minimize(self.model.evaluate_error, init_guess, method='Nelder-Mead',
                           tol=self.convergence_tol, bounds=self.bounds, options=self.options)
            init_guess = res.x
        return res.x

#   def evaluate(self, variable_params):
#       full_param_vector = np.zeros(self.num_params)
#       full_param_vector[self.frozen_param_idxs] = self.frozen_param_values
#       full_param_vector[self.variable_param_idxs] = variable_params
#       new_config = self.update_config_params(self.base_config, self.model_metadata, full_param_vector)
#       model = closed_loop_model.svZeroD_ClosedLoop(new_config)
#       try:
#           results = model.run_model()
#           return results
#       except:
#           print('Invalid result for parameters: ', variable_params)
#           return np.empty(self.num_results)*np.nan

#   def evaluate_simple(self, params):
#       model = closed_loop_model.svZeroD_ClosedLoop(self.base_config)
#       frozen_param_idxs, variable_param_idxs, _ = model.read_frozen_params('frozen_parameters.csv')
#       try:
#           results = model.run_with_frozen_params(params)
#           return results
#       except:
#           print('Invalid result for parameters: ', params)
#           return np.empty(self.num_results)*np.nan

 #  def create_prior(self):
 #      model = closed_loop_model.svZeroD_ClosedLoop(self.base_config)
 #      all_parameter_limits = model.parameter_limits()
 #      lower_lims = all_parameter_limits[np.arange(0,2*model.get_num_parameters(),2)]
 #      upper_lims = all_parameter_limits[np.arange(0,2*model.get_num_parameters(),2)+1]
 #      prior = utils.BoxUniform(low=torch.tensor(lower_lims[self.variable_param_idxs]), high=torch.tensor(upper_lims[self.variable_param_idxs]))
 #      return prior

 #  def run_simulations(self, num_procs, parameter_samples):
 #      p = Pool(num_procs)
 #      all_results = p.map(self.evaluate, parameter_samples)
 #      #all_results = p.map(self.evaluate_simple, parameter_samples)
 #      return torch.tensor(np.array(all_results))

#   def update_config_params(self, original_config, metadata, params):
#       new_config = original_config.copy()
#       for i, bc_idx in enumerate(metadata['idxs_corBC_l']):
#           bc_values = new_config['boundary_conditions'][bc_idx]['bc_values']
#           bc_values['Ra'] = metadata['Ra_l_base'][i]*params[26]
#           bc_values['Ram'] = metadata['Ram_l_base'][i]*params[26]
#           bc_values['Rv'] = metadata['Rv_l_base'][i]*params[27]
#           bc_values['Ca'] = metadata['Ca_l_base'][i]*params[29]
#           bc_values['Cim'] = metadata['Cim_l_base'][i]*params[28]
#
#       for i, bc_idx in enumerate(metadata['idxs_corBC_r']):
#           bc_values = new_config['boundary_conditions'][bc_idx]['bc_values']
#           bc_values['Ra'] = metadata['Ra_r_base'][i]*params[26]
#           bc_values['Ram'] = metadata['Ram_r_base'][i]*params[26]
#           bc_values['Rv'] = metadata['Rv_r_base'][i]*params[27]
#           bc_values['Ca'] = metadata['Ca_r_base'][i]*params[31]
#           bc_values['Cim'] = metadata['Cim_r_base'][i]*params[30]
#
#       for i, bc_idx in enumerate(metadata['idxs_RCRBC']):
#           bc_values = new_config['boundary_conditions'][bc_idx]['bc_values']
#           bc_values['Rp'] = metadata['Rp_rcr_base'][i]*params[32]
#           bc_values['C'] = metadata['C_rcr_base'][i]*params[33]
#           bc_values['Rd'] = metadata['Rd_rcr_base'][i]*params[32]
#
#       heart_params = new_config['closed_loop_blocks'][0]['parameters']
#       heart_params['Tsa']  = params[0]
#       heart_params['tpwave']  = params[1]
#       heart_params['Erv_s']  = params[2]
#       heart_params['Elv_s']  = params[3]
#       heart_params['iml']  = params[4]
#       heart_params['imr']  = params[34]
#       heart_params['Lra_v']  = params[7]/metadata['pConv'];
#       heart_params['Rra_v']  = params[8]/metadata['pConv'];
#       heart_params['Lrv_a']  = params[5]/metadata['pConv'];
#       heart_params['Rrv_a']  = metadata['Rrv_base']*params[6]/metadata['pConv']
#       heart_params['Lla_v'] = params[9]/metadata['pConv']
#       heart_params['Rla_v'] = params[10]/metadata['pConv']
#       heart_params['Llv_a'] = params[12]/metadata['pConv']
#       heart_params['Rlv_ao'] = metadata['Rlv_base']*params[11]/metadata['pConv']
#       heart_params['Vrv_u'] = params[13]
#       heart_params['Vlv_u'] = params[14]
#       heart_params['Rpd'] = metadata['Rpd_base']*params[15]/metadata['pConv']
#       heart_params['Cp'] = params[16]
#       heart_params['Cpa'] = params[17]
#       heart_params['Kxp_ra'] = params[18]
#       heart_params['Kxv_ra'] = params[19]
#       heart_params['Kxp_la'] = params[22]
#       heart_params['Kxv_la'] = params[23]
#       heart_params['Emax_ra'] = params[20]
#       heart_params['Emax_la'] = params[24]
#       heart_params['Vaso_ra'] = params[21]
#       heart_params['Vaso_la'] = params[25]
#
#       return new_config


if __name__ == "__main__":

    config_file = './svzerod_tuning.json'
    sbi_coronary = SBI_ClosedLoopCoronary(config_file)
#   prior = sbi_coronary.create_prior()
#   num_samples = 20
#   param_samples = prior.sample((num_samples,))
#   print(param_samples)

#   num_procs = 2
#   all_results = sbi_coronary.run_simulations(num_procs, param_samples)
#   print(all_results)
#   print(len(all_results))


#simulator, prior = prepare_for_sbi(model.run_with_frozen_params, prior)
#num_samples = 5
#params, results = create_samples_and_results(num_samples, simulator, prior)

#posterior = infer(model.run_with_frozen_params, prior, method="SNPE", num_simulations=1000)
#params, results = simulate_for_sbi(model.run_with_frozen_params, proposal = prior, num_simulations = 100, num_workers = 2)