/
psrplot.py
301 lines (201 loc) · 9.24 KB
/
psrplot.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
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
import scipy.stats as stats
import math
import sidereal as sd
from scipy.optimize import fsolve
from sys import exit
# AUXILIARY FUNCTIONS
def lin_fit(lins):
# assuming input is a series
# get injections
lins = lins.drop([0])
# put vectors in proper shape
x = np.reshape(lins.index, (len(lins), 1))
y = np.reshape(lins, (len(lins), 1))
# fit
p, _, _, _ = np.linalg.lstsq(x, y)
return np.poly1d([p[0][0], 0])
def noise_line(d):
# input can be a dataframe
# find elements with index 0 (no injection) and take their max over columns and rows
try:
max_n = d[0].max()
except KeyError:
max_n =d.T[0].max().max()
# return corresponding constant polynomial
return np.poly1d([max_n])
def rmse(d):
# computes RMSE between d and its best-fit line
# get injections
d = d.drop([0])
x = np.reshape(d.index, (len(d), 1)) # take injections
y = np.reshape(d, (len(d), 1)) # take recovered values
# fit and obtain sum of residuals (squared Euclidean 2-norm for each column in b - a*x)
_, residuals, _, _ = np.linalg.lstsq(x, y)
# sum of residuals = Sum[(x-y)^2], so RMSE = sqrt(s.o.r/N)
return np.sqrt(residuals[0]/len(d))
def min_det_h(d):
# assumes input is a series
# get max noise
noise = noise_line(d)(1)
# find max injection below that
d2 = d[d<noise]
det_injs = d2[d2.index>0]
try:
print 'Get min'
return det_injs.index[np.argmax(det_injs)]
except ValueError:
print 'ValueError'
return 0
def fit_intersect_noise(d):
# assumes input is a series
noise = noise_line(d) # noise line
fit = lin_fit(d) # fit line
hdetmin = min_det_h(d) # min detected h (probably around intersection)
h_intersect = fsolve(lambda x: fit(x) - noise(x), hdetmin)
return h_intersect[0]
# PLOTTING FUNCTIONS
def p(hinj=[], hrec=[], s=[], psrname='', detname='', style=sd.default_style, methods=[]):
for method in methods:
# First Calculate the interquartile range
#(http://comments.gmane.org/gmane.comp.python.scientific.user/19755)
data = np.sort(hrec)
upperQuartile = stats.scoreatpercentile(data,.75)
lowerQuartile = stats.scoreatpercentile(data,.25)
IQR = upperQuartile - lowerQuartile
# Get ideal bin size
#(http://en.wikipedia.org/wiki/Freedman%E2%80%93Diaconis_rule)
# fdsize = 3.49*np.std(data)*len(data)**(-1./3.)
fdsize = 2 * IQR * len(data)**(-1./3.)
#Get number of bins
#(http://stats.stackexchange.com/questions/798/calculating-optimal-number-of-bins-in-a-histogram-for-n-where-n-ranges-from-30)
num_bins = int((np.amax(data) - np.amin(data))/fdsize)
cumfreqs, lowlim, binsize, _ = stats.cumfreq(data, num_bins)
pv = [1. - cdf/max(cumfreqs) for cdf in cumfreqs]
bins = np.linspace(lowlim, num_bins*binsize, num_bins)
plt.plot(bins, pv, style, color=sd.sd.pltcolor[method], label=method)
plt.yscale('log')
plt.title(detname + ' PSR ' + psrname)
plt.xlabel('$h_{rec}$')
plt.ylabel('1 - CDF (log scale)')
plt.legend(numpoints=1)
plt.savefig('plots/p_' + detname + '_' + psrname, bbox_inches='tight')
print 'Plotted and saved in: ',
print 'plots/p_' + detname + '_' + psrname
plt.close()
# compound plots
def hinjs(hinj=[], hrec=[], s=[], style=sd.default_style, methods=[]):
method_plot = plt.plot(hinj, s[methods], style)
for i in range(0, len(method_plot)):
plt.setp(method_plot[i], color=sd.pltcolor[methods[i]], label=methods[i])
plt.xlabel('$h_{inj}$')
plt.ylabel('Significance')
plt.legend(numpoints=1, loc=2)
def hinjlins(hinj=[], hrec=[], s=[], style=sd.default_style, methods=[]):
lins = s.applymap(math.sqrt)
method_plot = plt.plot(hinj, lins[methods], style)
slope = []
for i in range(0, len(method_plot)):
plt.setp(method_plot[i], color=sd.pltcolor[methods[i]], label=methods[i])
# fit
line = lin_fit(lins[methods[i]])
slope += [line(1)]
# plot
plt.plot(s.index, line(s.index), ls='-',color=sd.pltcolor[methods[i]], linewidth=.05)
# find noise for max slope line
n_line = noise_line(lins[methods[np.argmax(slope)]])
plt.plot(s.index, n_line(s.index), '--', color='.95')
linsmax = max([np.amax(s[m].map(lambda x: np.sqrt(x))) for m in methods])
plt.xlabel('$h_{inj}$')
plt.ylabel('Linearized significance')
plt.ylim(ymax=linsmax)
plt.legend(numpoints=1, loc=2)
def hinjrec(hinj=[], hrec=[], s=[], style=sd.default_style, methods=[]):
method_plot = plt.plot(hinj, hrec[methods], style)
for i in range(0, len(method_plot)):
plt.setp(method_plot[i], color=sd.pltcolor[methods[i]], label=methods[i])
plt.legend(numpoints=1, loc=2)
hmax = max([np.amax(hrec[m]) for m in methods])
plt.xlabel('$h_{inj}$')
plt.ylabel('$h_{rec}$')
plt.ylim(ymax=hmax)
# simple plots
def inj(hinj=[], hrec=[], s=[], style=sd.default_style, methods=[]):
plt.plot(hinj, style)
plt.ylabel('$h_{inj}$')
plt.xlabel('Instantiation')
def rec(hinj=[], hrec=[], s=[], style=sd.default_style, methods=[]):
method_plot = plt.plot(h[methods], style)
for i in range(0, len(method_plot)):
plt.setp(method_plot[i], color=sd.pltcolor[methods[i]], label=methods[i])
plt.ylabel('$h_{rec}$')
plt.ylabel('Instantiations')
plt.legend(numpoints=1)
def sig(hinj=[], hrec=[], s=[], style=sd.default_style, methods=[]):
method_plot = plt.plot(s[methods], style)
for i in range(0, len(method_plot)):
plt.setp(method_plot[i], color=sd.pltcolor[methods[i]], label=methods[i])
plt.ylabel('$Significance$')
plt.ylabel('Instantiations')
plt.legend(numpoints=1)
def original(detector, psr, location='files/data/'):
try:
d = pd.HDFStore(location + 'dataPitkin_' + detector + '.hdf5', 'r')
a = d[psr].tolist()
b = [x.real for x in a]
plt.plot(b[:1000])
finally:
d.close()
def sigma(detector, psrname, location='files/analysis/psrSegmentSigma', compare=False):
if compare:
print 'comparing'
sgS5 = pd.read_table('files/data/source/S5/sigmaS5_' + detector, names=[None, 'matlab'], sep='\s+', header=None, index_col=0)
sgS5.plot(style='r+')
else:
pass
# sg1 = pd.HDFStore(location + '_mat', 'r')
try:
sg = pd.HDFStore(location, 'r') #location, 'r')
sg[psrname].plot(style='g+')
plt.legend(numpoints=1)
plt.title(psrname + detector + 'daily std')
plt.xlabel('GPS time')
plt.ylabel('Standard deviation')
plt.savefig('plots/std_' + detector + '_' + psrname, bbox_inches='tight')
print 'Plotted and saved in: ',
print 'plots/' + detector + '_' + psrname
plt.close()
finally:
sg.close()
def p_original(detector, psr, location='files/remote/source/'):
d = pd.HDFStore(location + 'dataPitkin_' + detector + '.hdf5', 'r')
a = d[psr].tolist()
b = [abs(x) for x in a]
# First Calculate the interquartile range
#(http://comments.gmane.org/gmane.comp.python.scientific.user/19755)
data = np.sort(d[psr].tolist())
upperQuartile = stats.scoreatpercentile(data,.75)
lowerQuartile = stats.scoreatpercentile(data,.25)
IQR = upperQuartile - lowerQuartile
# Get ideal bin size
#(http://en.wikipedia.org/wiki/Freedman%E2%80%93Diaconis_rule)
# fdsize = 3.49*np.std(data)*len(data)**(-1./3.)
fdsize = 2 * IQR * len(data)**(-1./3.)
#Get number of bins
#(http://stats.stackexchange.com/questions/798/calculating-optimal-number-of-bins-in-a-histogram-for-n-where-n-ranges-from-30)
num_bins = int((np.amax(data) - np.amin(data))/fdsize)
cumfreqs, lowlim, binsize, _ = stats.cumfreq(data, num_bins)
pv = [1. - cdf/max(cumfreqs) for cdf in cumfreqs]
bins = np.linspace(lowlim, num_bins*binsize, num_bins)
plt.plot(bins, pv, style, color=sd.pltcolor[method], label=method)
plt.yscale('log')
plt.title(detname + ' PSR ' + psrname)
plt.xlabel('$h_{rec}$')
plt.ylabel('1 - CDF (log scale)')
plt.legend(numpoints=1)
plt.savefig('plots/p_' + detname + '_' + psrname, bbox_inches='tight')
print 'Plotted and saved in: ',
print 'plots/p_' + detname + '_' + psrname
plt.close()