-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathloadModel.py
215 lines (174 loc) · 8.56 KB
/
loadModel.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
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
212
213
214
215
""" Load the data from the files and make plots
"""
__author__ = 'Atoosa Parsa'
__copyright__ = 'Copyright 2024, Atoosa Parsa'
__credits__ = 'Atoosa Parsa'
__license__ = 'MIT License'
__version__ = '2.0.0'
__maintainer__ = 'Atoosa Parsa'
__email__ = 'atoosa.parsa@gmail.com'
__status__ = "Dev"
import warnings
warnings.simplefilter(action='ignore', category=FutureWarning)
import time
import copy
import argparse
import sys
import os
import io
import pickle
import torch
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import matplotlib
import CGMtorch as CGMtorch
import myUtils as myUtils
import myIO as myIO
import loadData
font = {'family' : 'sans-serif'}
plt.rc('font', **font)
matplotlib.rcParams['pdf.fonttype'] = 42
matplotlib.rcParams['ps.fonttype'] = 42
sns.set_style("ticks", {"xtick.major.size": 8, "ytick.major.size": 8, 'font.family': 'serif', 'font.serif': 'Times New Roman'})
colors = ["cyan", "dark pink", "ocean green", "tan", "light red", "grey", "apricot"]
sns.set_palette(sns.xkcd_palette(colors), desat=.9)
parser = argparse.ArgumentParser()
parser.add_argument('--name', type=str, default=time.strftime('%Y%m%d%H%M%S'),
help='Name to use when saving or loading the model file.')
parser.add_argument('--savedir', type=str, default='./exp/',\
help='Directory in which the model file is saved.')
parser.add_argument('--seed', type=int, default=5,\
help='Number of seeds')
parser.add_argument('--output', type=list, default='plots',\
help='Output directory')
parser.add_argument('--epoch', type=int, default=None,\
help='Epoch.')
parser.add_argument('--plotName', type=str, default=None,\
help='Plot name.')
class CPU_Unpickler(pickle.Unpickler):
def find_class(self, module, name):
if module == 'torch.storage' and name == '_load_from_bytes':
return lambda b: torch.load(io.BytesIO(b), map_location='cpu')
else:
return super().find_class(module, name)
if __name__ == '__main__':
args = parser.parse_args()
if not os.path.exists(args.savedir):
print("input directory does not exist.")
if args.epoch != None:
modelName = args.name+'_'+str(args.epoch)+'.pt'
else:
modelName = args.name+'_final'+'.pt'
savepath = os.path.join(args.savedir, 'plots')
if not os.path.exists(savepath):
os.makedirs(savepath)
seeds = np.arange(1, int(args.seed)+1)
losses = []
case = ['case 01', 'case 10', 'case 11']
stiffnesses = []
for s in seeds:
plotPath = os.path.join(savepath, f'seed{s}/')
if not os.path.exists(plotPath):
os.makedirs(plotPath)
path = os.path.join(args.savedir, f'seed{s}/')
model, dynamics, history, history_state, params = myIO.load_model(path+modelName)
dev = torch.device('cpu')
print('Running on ', dev)
torch.set_default_dtype(torch.float64)
f = open(path+"dataset.dat", 'rb')
data = CPU_Unpickler(f).load()
f.close()
datasetParams = data['params']
datasetParams['timeSteps'] = datasetParams['window_length']
train_dl, test_dl = loadData.loadData(dev, datasetParams, shuffle=False)
N_epochs = np.array(history.loc[:, 'epoch'])[-1]
state_size = (datasetParams['batchSize'],) + dynamics.initial_X.size()
N_epochs = np.array(history.loc[:, 'epoch'])[-1]
figsize = 8
fig = plt.figure(figsize=(1.5*figsize, 1*figsize))
gs = fig.add_gridspec(3, 2, hspace=0, wspace=0, top=0.9)
axs = gs.subplots(sharey=True, sharex=True)
#fig.subplots_adjust(wspace=0, hspace=0)
for num, (xb, yb) in enumerate(train_dl):
X = torch.zeros(state_size, device=dev)
Y = torch.zeros(state_size, device=dev)
V_x = torch.zeros(state_size, device=dev)
V_y = torch.zeros(state_size, device=dev)
A_x = torch.zeros(state_size, device=dev)
A_y = torch.zeros(state_size, device=dev)
with torch.no_grad():
if num >= 3:
break
outs, X, Y, V_x, V_y, A_x, A_y = model(xb, X, Y, V_x, V_y, A_x, A_y, params['fire_reset'], propagate=True)
for i in range(xb.size(0)):
x = torch.transpose(xb[i, :, :], 1, 0).detach().cpu().numpy()
axs[num][0].plot(x[:, 0], color='blue', linestyle='-', linewidth=2, label='input 1')
axs[num][0].plot(x[:, 1], color='green', linestyle='--', linewidth=2, label='input 2')
axs[num][0].grid(which='major', color='gray', linestyle='-', linewidth=0.3)
axs[num][0].set_axisbelow(True)
axs[num][0].tick_params(axis='both', which='major', labelsize=7, color='k')
if num == 0:
axs[num][0].legend(loc='upper right', fontsize=16)
axs[num][0].text(0, 0.0008, case[num], fontsize=16, weight='demibold', backgroundcolor='w')
axs[num][0].set_ylim((-1.1e-3, 1.1e-3))
y = outs[i, :, :].detach().cpu().numpy()
axs[num][1].plot(y[:, 0], color='tomato', linestyle='-', linewidth=2, label='output')
axs[num][1].grid(which='major', color='gray', linestyle='-', linewidth=0.3)
axs[num][1].set_axisbelow(True)
axs[num][1].tick_params(axis='both', which='major', labelsize=7, color='k')
if num == 0:
axs[num][1].legend(loc='upper right', fontsize=16)
axs[num][1].set_ylim((-1.1e-3, 1.1e-3))
fig.text(0.5, 0.07, "Time Steps", ha='center', va='center', fontsize=14)
fig.text(0.07, 0.5, "Displacement", ha='center', va='center', rotation='vertical', fontsize=14)
fig.suptitle(f"Particle Displacements - {args.plotName}", fontsize=16, y=0.93, fontweight='bold')
for ax in fig.get_axes():
ax.label_outer()
plt.tight_layout()
savename = 'waves_seed'+str(s)+'.png'
saving = os.path.join(plotPath, savename)
plt.show()
fig.savefig(saving, bbox_inches='tight', dpi=300) #, transparent=False)
plt.close(fig)
print("Seed: "+str(s))
stiffnesses.append(dynamics.material.K_padded.detach().cpu().numpy().tolist())
#print(dynamics.material.K_padded)
CGMtorch.plot.plot_config(N_epochs, dynamics.material.N_, dynamics.material.X_ini, dynamics.material.Y_ini,\
dynamics.material.D,\
dynamics.material.Lx, dynamics.material.Ly, dynamics.material.K_padded, \
params['k_min'], params['k_max'], dynamics.sources, dynamics.probes,\
plotdir=plotPath)
losses.append(np.array(history.loc[:, 'loss_train']))
print('Seed: '+str(s))
times = np.array(history.loc[:, 'time'])
print('Total Time: %.2f min\n' % ((times[-1] - times[0]) / 60))
print(stiffnesses)
losses = np.array(losses)
epochs = np.array(history.loc[:, 'epoch'])
mean = np.mean(losses, axis=0)
std = np.std(losses, axis=0)
fig, ax = plt.subplots(1, 1)
fig.set_figwidth(8)
fig.set_figheight(4)
for i in range(losses.shape[0]):
ax.plot(epochs, losses[i], color='blue', linewidth=1, alpha=0.15)
ax.plot(epochs, mean, color='blue', linewidth=2.5, label='$Mean$')
ax.plot(epochs, mean-std, color='blue', linewidth=1.5, linestyle='dashed', label='$\pm 1 \; STD$')
ax.plot(epochs, mean+std, color='blue', linewidth=1.5, linestyle='dashed')
ax.set_axisbelow(True)
#ax.grid(which='minor', color='gray', linestyle=':', linewidth=0.2)
ax.grid(which='major', color='gray', linestyle='-', linewidth=0.2)
ax.minorticks_on()
ax.set_xlabel('Epochs', fontsize=14, fontweight='bold')
ax.set_ylabel('Loss', fontsize=14, fontweight='bold')
ax.legend(loc='upper right', fontsize=14)
ax.tick_params(axis='both', which='major', labelsize=12, width=2)
ax.tick_params(axis='both', which='minor', labelsize=12, width=1)
plt.setp(ax.get_xticklabels(), visible=True)
plt.setp(ax.get_yticklabels(), visible=True)
plt.tight_layout()
fig.savefig(savepath+'/loss.png', bbox_inches='tight', dpi=300, transparent=False)
plt.show()
plt.close(fig)