def load_data(data_path,pars_path,n_samples):
"""
Loads txt files into a numpy array called data.
"""
# load xml file containing all waveform parameters
for _,xml_filename in enumerate(glob.iglob('%s/*.xml' % pars_path)):
print("Loading xml file: {}".format(xml_filename))
data_pars = np.array([])
mdcset = mdctools.MDCSet(['H1'])
mdcset.load_xml(xml_filename)
# load each waveform txt time series file
for counter,filename in enumerate(glob.iglob('%s/*.txt' % data_path)):
print("Loading waveform #{}".format(counter))
offset = int(np.random.uniform(-100,100)) # vary position of waveform
if counter == 0:
data = np.loadtxt(filename)
data = resample(data, 512)
data /= np.max(data)
data = np.roll(data,offset)
# store parameters in array
data_pars = np.array([(512/2)+offset,mdcset.waveforms[counter].frequency])
# plot first waveform in training set
plt.close()
#plt.plot(data)
#plt.savefig('/home/hunter.gabbard/public_html/Burst/sine-gaussian_runs/resampled_waveform.png')
#plt.close()
elif counter == n_samples:
break
else:
data_new = np.loadtxt(filename)
data_new = resample(data_new, 512)
data_new /= np.max(data_new)
data_new = np.roll(data_new,offset)
# plot next waveform in training set
#plt.plot(data_new)
#plt.savefig('/home/hunter.gabbard/public_html/Burst/sine-gaussian_runs/resampled_waveform.png')
data = np.vstack((data,data_new))
# store waveforms parameters in array
data_pars = np.vstack((data_pars, [(512/2)+offset,mdcset.waveforms[counter].frequency]))
return data, data_pars
from minke import mdctools, distribution, sources
mdcset = mdctools.MDCSet(['L1', 'H1'])
times = distribution.uniform_time(1126620016, 1136995216, number=1000)
hrss_values = distribution.log_uniform(5e-23, 1e-20, len(times))
wnb = sources.WhiteNoiseBurst(duration=0.1,
bandwidth=10,
frequency=1000,
hrss=1e-23,
time=1126630000,
seed=3)
mdcset + wnb
for hrss, time in zip(hrss_values, times):
wnb = sources.WhiteNoiseBurst(duration=0.1,
bandwidth=10,
frequency=1000,
hrss=hrss,
time=time,
seed=3)
mdcset + wnb
mdcset.save_xml('wnb1000b10tau0d1.xml.gz')
from minke import mdctools, distribution, sources
from random import randint
mdcset = mdctools.MDCSet(['H1'])
times = distribution.uniform_time(1126620016, 1136995216, number=1000)
hrss_values = distribution.log_uniform(1e-23, 1e-23, len(times))
sineGauss = sources.SineGaussian(q=15,
frequency=100,
polarisation='linear',
hrss=1e-23,
time=1126630000,
seed=3)
mdcset + sineGauss
for hrss, time in zip(hrss_values, times):
sineGauss = sources.SineGaussian(q=15,
frequency=randint(100, 200),
polarisation='linear',
hrss=hrss,
time=time,
seed=3)
mdcset + sineGauss
mdcset.save_xml('sineGauss100b10tau0d1.xml.gz')