-
Notifications
You must be signed in to change notification settings - Fork 1
/
plot_data.py
executable file
·247 lines (186 loc) · 7.88 KB
/
plot_data.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
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
#!/usr/bin/env python
import argparse,logging,json,glob,sys
import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
from matplotlib import colors
logger = logging.getLogger(__name__)
def excepthook(*args):
logging.getLogger().error('Uncaught exception:', exc_info=args)
sys.excepthook = excepthook
class NPMean:
def __init__(self,shape):
self.sum = np.zeros(shape)
self.sum2 = np.zeros(shape)
self.n = 0.
def __iadd__(self,other):
assert self.sum.shape == other.shape,'shape missmatch: %s != %s' % (self.sum.shape,other.shape)
self.sum = self.sum + other
self.sum2 = self.sum2 + other*other
self.n += 1.
return self
def mean(self):
assert self.n > 2.,'mean must have more than 2 entries'
return self.sum / self.n
def sigma(self):
assert self.n > 2.,'mean must have more than 2 entries'
mean = self.mean()
return np.sqrt((1./self.n) * self.sum2 - mean * mean)
def main():
logging.basicConfig(level=logging.INFO)
''' simple starter program that can be copied for use when starting a new script. '''
parser = argparse.ArgumentParser(description='Atlas Training')
parser.add_argument('--config_file', '-c',
help='configuration in standard json format.')
args = parser.parse_args()
# load configuration
config_file = json.load(open(args.config_file))
# get file list
filelist = sorted(glob.glob(config_file['data_handling']['input_file_glob']))
np.random.shuffle(filelist)
logger.info('found %s input files',len(filelist))
# profiling data
layer_histo_bins = 110
num_layers = 15
num_eta = 1153
num_phi = 50
labels = ['u','s','c','b','g']
profile_data = {'b':{'layerhist':NPMean([num_layers,110]),
'etaphi':NPMean([num_phi,num_eta]),
'fullimg':NPMean([num_layers,num_phi,num_eta])},
'c':{'layerhist':NPMean([num_layers,110]),
'etaphi':NPMean([num_phi,num_eta]),
'fullimg':NPMean([num_layers,num_phi,num_eta])},
's':{'layerhist':NPMean([num_layers,110]),
'etaphi':NPMean([num_phi,num_eta]),
'fullimg':NPMean([num_layers,num_phi,num_eta])},
}
image_counter = 0
for filename in filelist:
logger.info('[%s] plotting file: %s',image_counter,filename)
data = np.load(filename)
images = data['raw']
batch_size,nchannels,bins_phi,bins_eta = images.shape
truth = data['truth']
for i in range(images.shape[0]):
plt.figure(1,figsize=(16,12))
logger.debug(' index: %s',i)
image = images[i]
y = truth[i][0]
label = labels[np.argmax(y[5:])]
logger.debug('%s %s',image.shape,y.shape)
logger.debug('%s %s',y,label)
profile_data[label]['fullimg'] += image
plt.subplot(221)
histo_data = image.flatten()
n, bins, patches = plt.hist(histo_data,100)
plt.yscale('log')
plt.subplot(222)
eta_phi = np.sum(image,axis=0)
p = plt.imshow(eta_phi,aspect='auto')
plt.colorbar(p,ax=plt.gca(),orientation='vertical')
plt.gca().invert_yaxis()
profile_data[label]['etaphi'] += eta_phi
plt.subplot(223)
depth_eta = np.sum(image,axis=1)
p = plt.imshow(depth_eta,aspect='auto')
plt.colorbar(p,ax=plt.gca(),orientation='vertical')
plt.gca().invert_yaxis()
plt.subplot(224)
depth_phi = np.sum(image,axis=2)
p = plt.imshow(depth_phi,aspect='auto')
plt.colorbar(p,ax=plt.gca(),orientation='vertical')
plt.gca().invert_yaxis()
plt.gcf().savefig('%010d_overview.png' % image_counter)
plt.close('all')
fig,ax = plt.subplots(num_layers)
fig.set_figwidth(12)
fig.set_figheight(16)
vmin = np.min(image)
vmax = np.max(image)
logger.debug('min = %s; max = %s',vmin,vmax)
norm = colors.LogNorm(vmin=1e-6, vmax=vmax)
plots = []
for layer in range(image.shape[0]):
logger.debug('plot layer: %s',layer)
axid = image.shape[0]-(layer+1)
p = ax[axid].imshow(image[layer],aspect='auto')
ax[axid].text(-0.1,0.5,'%s' % layer,fontsize=14,horizontalalignment='center',verticalalignment='center',transform=ax[axid].transAxes)
p.set_norm(norm)
p.callbacksSM.connect('changed',update)
plots.append(p)
ax[layer].label_outer()
# figB.colorbar(p,ax=axB[layer],ticks=[0,1])
fig.colorbar(plots[0], ax=ax, orientation='vertical', fraction=.1,aspect=30)
fig.text(0.01,0.95,'layer',fontsize=15)
#plt.show()
fig.savefig('%010d_layer.png' % image_counter)
plt.close('all')
plt.figure(1,figsize=(16,12))
layer_histos = []
for layer in range(image.shape[0]):
plt.subplot(num_layers,1,image.shape[0]-layer)
layer_histo = image[layer].flatten()
n, bins, patches = plt.hist(layer_histo,bins=layer_histo_bins,range=(-0.1,1.0),alpha=0.5,label='%s' % layer)
layer_histos.append(n)
plt.ylim(0.1,1e5)
plt.yscale('log')
logger.debug('layer %s max %s',layer,np.max(layer_histo))
profile_data[label]['layerhist'] += np.array(layer_histos)
plt.gcf().savefig('%010d_layerhist.png' % image_counter)
plt.close('all')
image_counter += 1
if image_counter > 2500:
break
for quarkflavor in profile_data:
image = profile_data[quarkflavor]['fullimg']
mean = image.mean()
sigma = image.sigma()
np.savez_compressed('quarkflavor_%s_fullimg.npz' % quarkflavor,mean=mean,sigma=sigma)
fig,ax = plt.subplots(num_layers)
fig.set_figwidth(12)
fig.set_figheight(16)
vmin = np.min(mean)
vmax = np.max(mean)
logger.debug('min = %s; max = %s',vmin,vmax)
norm = colors.LogNorm(vmin=1e-6, vmax=vmax)
plots = []
for layer in range(mean.shape[0]):
logger.debug('plot layer: %s',layer)
axid = mean.shape[0]-(layer+1)
p = ax[axid].imshow(mean[layer],aspect='auto')
ax[axid].text(-0.1,0.5,'%s' % layer,fontsize=14,horizontalalignment='center',verticalalignment='center',transform=ax[axid].transAxes)
p.set_norm(norm)
p.callbacksSM.connect('changed',update)
plots.append(p)
ax[layer].label_outer()
# figB.colorbar(p,ax=axB[layer],ticks=[0,1])
fig.colorbar(plots[0], ax=ax, orientation='vertical', fraction=.1,aspect=30)
fig.text(0.01,0.95,'layer',fontsize=15)
plt.gcf().savefig('quarkflavor_%s_fullimg.png' % quarkflavor)
plt.close('all')
plt.figure(1,figsize=(16,12))
image = profile_data[quarkflavor]['layerhist']
mean = image.mean()
sigma = image.sigma()
np.savez_compressed('quarkflavor_%s_layerhist.npz' % quarkflavor,mean=mean,sigma=sigma)
layer_histos = []
for layer in range(mean.shape[0]):
plt.subplot(num_layers,1,mean.shape[0]-layer)
layer_data_y = mean[layer]
layer_data_x = [-0.095 + i*0.01 for i in range(layer_histo_bins)]
layer_error_x = [0.005]*layer_histo_bins
layer_error_y = sigma[layer]
plt.errorbar(layer_data_x,layer_data_y,xerr=layer_error_x,yerr=layer_error_y)
plt.ylim(1,1e5)
plt.yscale('log')
plt.gcf().savefig('quarkflavor_%s_layerhist.png' % quarkflavor)
plt.close('all')
def update(changed_image,images):
for im in images:
if (changed_image.get_cmap() != im.get_cmap() or changed_image.get_clim() != im.get_clim()):
im.set_cmap(changed_image.get_cmap())
im.set_clim(changed_image.get_clim())
if __name__ == '__main__':
main()