def create_injection_list(N_inj, srate, start_GPS, end_GPS, theta_range = [[10,100],[10,100],[-0.8,0.8],[-0.8,0.8],[40,400],[0.,np.pi],[0.,2*np.pi]], datafile = None):
"""
Creates a list of injections.
Each injection is a dictionary with entries:
"WF": a WF array plus of variable length
"srate": a sampling rate for the WF
"GPS": GPS time for the merger
"theta": a 7 dimensional array holding params of the WF [m1,m2,s1,s2,d_L, iota, phi]
"""
if datafile is not None:
data = np.squeeze(pd.read_csv(datafile, skiprows =3).to_numpy())
g = gen.GW_generator(0)
inj_list = []
theta_range = np.array(theta_range)
for i in range(N_inj):
inj = {}
theta = np.random.uniform(*theta_range.T,size = (7,))
#theta = [35.,30.,0.,0., 440., 0., 1.]
t_min = np.random.uniform(-10,-6)
t_grid = np.linspace(t_min, 1., int((1.-t_min)*srate))
h_p, h_c = g.get_WF(theta, t_grid)
#computing antenna patterns
sky_loc = np.random.uniform(0.,np.pi,size = (3,))
F_p, F_c = antenna_patterns(*sky_loc)
h = h_p *F_p + h_c*F_c
D_eff = theta[4]/np.sqrt(F_p**2 *( (1+np.cos(theta[5])**2) /2.)**2 + ( F_c*np.cos(theta[5]) )**2) #https://arxiv.org/pdf/1603.02444.pdf
#computing SNR
#FIXME: I have serious doubt that the SNR is computed correctly
#FIXME: NUmbers look fine but I see a peak in the SNR without injection!! What the f**k?!?!?!
#TODO: smooth WF switching on!!
#for SNR you can check:
# https://gwpy.github.io/docs/stable/examples/timeseries/pycbc-snr.html
# https://git.ligo.org/lscsoft/gstlal/-/blob/master/gstlal-inspiral/bin/gstlal_inspiral_injection_snr
if datafile is not None:
h_time_series = pycbc.types.timeseries.TimeSeries(h.astype(np.float64), 1./srate)
data_time_series = pycbc.types.timeseries.TimeSeries(data.astype(np.float64), 1./srate)
#data_time_series[0:len(h_time_series)] += h_time_series
data_time_series = TimeSeries.from_pycbc(data_time_series)
data_time_series = data_time_series.highpass(15)
psd = data_time_series.psd(len(h_time_series)/srate, 5).to_pycbc()
data_time_series = data_time_series.to_pycbc()[len(data_time_series)-len(h_time_series)-2000:-2000]
#data_time_series = pycbc.types.timeseries.TimeSeries(np.random.normal(0,1,len(data_time_series)).astype(np.float64), 1./srate)
#h_time_series = pycbc.types.timeseries.TimeSeries(np.random.normal(0,1,len(h_time_series)).astype(np.float64), 1./srate)
SNR_ts = pycbc.filter.matchedfilter.matched_filter(h_time_series/len(h_time_series), data_time_series, psd, low_frequency_cutoff=20., high_frequency_cutoff = 2048) #TD template should be divided by its length for FFT purposes! np.fft is weird and doesn't normalize stuff
SNR_ts = SNR_ts[100:-100]
SNR = np.max(np.abs(np.array(SNR_ts)))
#print("Theta | Injection SNR: ", theta, SNR)
#plt.plot(data_time_series)
#plt.plot(h_time_series)
#plt.plot(np.abs(np.array(SNR_ts)))
#plt.plot(np.log(psd))
#plt.show()
else:
SNR = None
inj['theta'] = theta
inj['skyloc'] = sky_loc
inj['D_eff'] = D_eff
inj['srate'] = srate
inj['GPS'] = int(start_GPS)
inj['time'] = np.random.uniform(.3, float(end_GPS-start_GPS))
inj['WF'] = h
inj['SNR'] = SNR
inj_list.append(inj)
return inj_list
mass2=32,
f_lower=20,
f_final=2048,
delta_f=psd.df.value)
# At this point we are ready to calculate the SNR, so we import the
# :func:`pycbc.filter.matched_filter
# <pycbc.filter.matchedfilter.matched_filter>` method, and pass it
# our template, the data, and the PSD:
from pycbc.filter import matched_filter
snr = matched_filter(hp,
zoom.to_pycbc(),
psd=psd.to_pycbc(),
low_frequency_cutoff=15)
snrts = TimeSeries.from_pycbc(snr).abs()
# .. note::
#
# Here we have used the :meth:`~TimeSeries.to_pycbc` methods of the
# `~gwpy.timeseries.TimeSeries` and `~gwpy.frequencyseries.FrequencySeries`
# objects to convert from GWpy objects to something that PyCBC functions
# can understand, and then used the :meth:`~TimeSeries.from_pycbc` method
# to convert back to a GWpy object.
# We can plot the SNR `TimeSeries` around the region of interest:
plot = snrts.plot()
ax = plot.gca()
ax.set_xlim(1126259461, 1126259463)
ax.set_epoch(1126259462.427)
ax.set_ylabel('Signal-to-noise ratio (SNR)')
# template `~pycbc.types.frequencyseries.FrequencySeries`:
from pycbc.waveform import get_fd_waveform
hp, _ = get_fd_waveform(approximant="IMRPhenomD", mass1=40, mass2=32,
f_lower=20, f_final=2048, delta_f=psd.df.value)
# At this point we are ready to calculate the SNR, so we import the
# :func:`~pycbc.filter.matched_filter` method, and pass it our template,
# the data, and the PSD, using the :meth:`~TimeSeries.to_pycbc` methods of
# the `TimeSeries` and `~gwpy.frequencyseries.FrequencySeries` objects:
import numpy
from pycbc.filter import matched_filter
snr = matched_filter(hp, zoom.to_pycbc(), psd=psd.to_pycbc(),
low_frequency_cutoff=15)
snrts = TimeSeries.from_pycbc(snr).abs()
# We can plot the SNR `TimeSeries` around the region of interest:
plot = snrts.plot()
ax = plot.gca()
ax.set_xlim(1126259461, 1126259463)
ax.set_epoch(1126259462.427)
ax.set_ylabel('Signal-to-noise ratio (SNR)')
ax.set_title('LIGO-Hanford signal-correlation for GW150914')
plot.show()
# We can clearly see a large spike (above 17!) at the time of the GW150914
# signal!
# This is, in principle, how the full, blind, CBC search is performed, using
# all of the available data, and a bank of tens of thousand of signal models.
