# Whether to have unique mapping b/w triggers
uniqueArgument = 'nonunique'
# Maximum number of coincidences
maxNumCoinc = len(trigXData)*len(trigHData)
# Get current working director
folder = os.popen('pwd').readlines()[0].split('\n')[0]
print 'folder: ', folder
channelXName = trigxfile.split('_', 3)[3].split('.')[0]
channelHName = trighfile.split('_', 3)[3].split('.')[0]
segIdx = np.intersect1d(np.where(trigXData[:, 2] > startTime)[0], np.where(trigXData[:,2] < endTime)[0])
trigXData = trigXData[segIdx]
segIdh = np.intersect1d(np.where(trigHData[:, 2] > startTime)[0], np.where(trigHData[:,2] < endTime)[0])
trigHData = trigHData[segIdh]
[coincTrigH, coincTrigX] = bcv.mcoinc(maxNumCoinc, trigHData[:, 2], trigXData[:, 2], COINC_TIME_WINDOW, segLength, uniqueArgument)
plt.figure()
plt.suptitle('Triggers between %s and %s' %(gtime.from_gps(startTime), gtime.from_gps(endTime)))
plt.subplot(2,1,1)
ax = plt.gca()
ax.errorbar(trigXData[:, 2], trigXData[:, 3], xerr=[trigXData[:,2] - trigXData[:, 0], trigXData[:, 1] - trigXData[:, 2]], fmt='o', label='Triggers')
plt.plot(trigXData[coincTrigX,2], trigXData[coincTrigX,3], '*', label='coincident triggers')
plt.xlabel('Central Times')
plt.ylabel('Central Freqs')
ax.set_title('Triggers of %s'%(channelXName))
plt.subplot(2,1,2)
ax = plt.gca()
ax.errorbar(trigHData[:, 2], trigHData[:, 3], xerr=[trigHData[:,2] - trigHData[:, 0], trigHData[:, 1] - trigHData[:, 2]], fmt='o', label='Triggers')
plt.plot(trigHData[coincTrigH,2], trigHData[coincTrigH,3], '*', label='coincident triggers')
def vetoanalysis(frameCache, chanHName, chanXName, frameTypeChanH, frameTypeChanX, samplFreqH, samplFreqX,
highPassCutoff, TriggerHList, TriggerXList, couplingModel,
transFnXtoH, analysisStartTime, analysisEndTime,
timeShift, outDir, logFid, debugLevel):
#VETOANALYSIS - Peform veto analysis using instrumental couplings on a set of triggers
#in the GW channel making use of a set of trigges in the an instrumental channel X and
#and the transfer function from channel X to the GW channel
#
# Usage : vetoanalysis(TriggerHList, TriggerXList, transFnXtoH,numberOfChannels, timeShift)
# TriggerHList - A structure containing the start times, and times and central times
# of the triggers in channel H
# TriggerXList - A structure containing the strat times, end times and central times of
# the triggers in channel X
# TransFnXtoH - A structure containing the transfer functions (Frequency and Amplitude) from channel X to Channel H
# analysisStartTime - GPS start time lower bound of candidate events
# analysisEndTime - GPS end times upper bound of candidate events
# timeShift - The number of seconds by whuch the coincident triggers
# will be time shifted after identification
# Sudarshan Ghonge <*****@*****.**>
# Original code written by
# Aaron B. Pearlman <*****@*****.**> and
# P. Ajith <ajith.icts.res.in>
#maximum number of allowed coincidences
maxNumCoinc = len(TriggerHList.centralTime)*len(TriggerXList.centralTime)
#Time window for identifying coincidences between channel H and X
COINC_TIME_WINDOW = 0.5
#Maximum length of one data segment used for the analysis
MAX_LENGTH_DATA_SEG = 64
################Ask Ajith what this means#
segLength = 3600
uniqueArgument = 'nonunique'
logFid.write( 'LOG: Finding coincidences... Num trigs H = %d, Num trigs X = %d\n'\
%(len(TriggerHList.centralTime), len(TriggerXList.centralTime)))
[coincTrigH, coincTrigX] = bcv.mcoinc(maxNumCoinc, TriggerHList.centralTime, TriggerXList.centralTime + timeShift, COINC_TIME_WINDOW, segLength, uniqueArgument)
trigHCentralTimeVec = TriggerHList.centralTime[coincTrigH]
trigXCentralTimeVec = TriggerXList.centralTime[coincTrigX]
trigHStartTimeVec = TriggerHList.startTime[coincTrigH]
trigXStartTimeVec = TriggerXList.startTime[coincTrigX]
trigHEndTimeVec = TriggerHList.endTime[coincTrigH]
trigXEndTimeVec = TriggerXList.endTime[coincTrigX]
trigHCentFreqVec = TriggerHList.centralFrequency[coincTrigH]
trigXCentFreqVec = TriggerXList.centralFrequency[coincTrigX]
trigHSignificVec = TriggerHList.triggerSignificance[coincTrigH]
trigXSignificVec = TriggerXList.triggerSignificance[coincTrigX]
trigHDurationVec = trigHEndTimeVec - trigHStartTimeVec
trigXDurationVec = trigXEndTimeVec - trigXStartTimeVec
del TriggerHList, TriggerXList
##Read Data Segments Using a Time-Window Around Coi-incident Triggers
##Check if the sample frequencies of Channel H X are the same. If not, throw error
#if(samplFreqH!=samplFreqX):
#sys.exit('Error: samplFreqH DOES NOT equal samplFreqX')
#else:
#samplFreq = samplFreqH
samplFreq = samplFreqH
# Apply time shift to the X/Y data. In case of bilinear coupling, there are multiple
# channels (X and Y); thus a vector of length 2 or more
if(couplingModel=='linear'):
timeShiftX = np.array([timeShift])
elif(couplingModel=='bilinear'):
timeShiftX = np.zeros(len(chanXName))
for iX in range(len(chanXName)):
timeShiftX[iX] = timeShift
#Check number of triggers in channel H that are coincident in channel X (coincTrigH)
#is the same the number of triggers in Channel X that are coincident in Channel H
#(coincTrigX)
if(len(coincTrigH)==len(coincTrigX) & len(coincTrigH)>0):
#Read data for timeShift from frameCache
logFid.write('LOG: Performing veto analysis for time shift %d..\n' %(timeShift))
logFid.write('LOG: Number of coincidences : %d...\n' %(len(coincTrigH)))
#Initializing empty lists to store the post analysis data
timeShiftVec = []
rHPMat = []
rMaxHPMat = []
meanYMat = []
varYMat = []
varYMat = []
minYMat = []
maxYMat = []
mindYMat = []
maxdYMat = []
meandYMat = []
trigHAnalysdCentTimeVec = []
trigXAnalysdCentTimeVec = []
trigHAnalysdCentFreqVec = []
trigXAnalysdCentFreqVec = []
trigHAnalysdSignificVec = []
trigXAnalysdSignificVec = []
trigHAnalysdDurationVec = []
trigXAnalysdDurationVec = []
#testId = open('test_' + '%d'%(analysisStartTime) + '.dat', 'w+')
#np.savetxt(testId, trigHCentralTimeVec)
#testId.close()
#outFileString = '/corrstat_timeshift%d_seg%d-%d.dat'%(timeShift, analysisStartTime, analysisEndTime)
#outFileName = outDir[iP] + '/' + outFileString
#outFid = open(outFileName, 'w+')
analysedTrigIdx = 0
for coincIndex in range(len(coincTrigH)):
trigHCentTime = trigHCentralTimeVec[coincIndex]
trigXCentTime = trigXCentralTimeVec[coincIndex]
trigHStartTime = trigHStartTimeVec[coincIndex]
trigXStartTime = trigXStartTimeVec[coincIndex]
trigHEndTime = trigHEndTimeVec[coincIndex]
trigXEndTime = trigXEndTimeVec[coincIndex]
trigHCentFreq = trigHCentFreqVec[coincIndex]
trigXCentFreq = trigXCentFreqVec[coincIndex]
trigHSignific = trigHSignificVec[coincIndex]
trigXSignific = trigXSignificVec[coincIndex]
trigHDuration = trigHEndTime - trigHStartTime
trigXDuration = trigXEndTime - trigXStartTime
# Find the segent of data used for analysis
segStartTime = min(trigHStartTime, trigXStartTime+timeShift)
segEndTime = max(trigHEndTime, trigXEndTime + timeShift)
meanTrigCentTime = (segStartTime + segEndTime)/2.0
totalDuration = segEndTime - segStartTime
# Calculate the total number of samples, rounded to the closest integer
# power of 2
totalNumberSamples = bcv.roundtopowertwo(totalDuration*samplFreq, 1024.0)
# Calculate the length of the data segment used for veto analysis
segmentDuration = totalNumberSamples / samplFreq
# Find the start and end times of a samll segment of dara that we want to
# analyse
segStartTime = meanTrigCentTime - segmentDuration / 2.0
segEndTime = meanTrigCentTime + segmentDuration/2.0
# Round the start and end times of a small segment of dara that we want to
# analyse to an integer of gps time.
# IMPORTANT NOTE: One second of data in the beginning is used to train the high
# pass filter and hecne should not be used for the analysis (this explains
# the "floor(segStartTime) - 1" below).
bcvreadStartTime = np.floor(segStartTime) - 1
bcvreadEndTime = np.ceil(segEndTime) + 1
if (debugLevel >=1):
#print values of different variables
bcv.printvar('--- Analysis Window ---',''
,'analysisStartTime',analysisStartTime,
'analysisEndTime', analysisEndTime,
'wreadStartTime', bcvreadStartTime,
'wreadEndTime', bcvreadEndTime,
'segStartTime', segStartTime,
'segEndTime', segEndTime,
'--- Trigger Parameters ---','',
'trigHStartTime', trigHStartTime,
'trigXStartTime', trigXStartTime,
'trigHEndTime', trigHEndTime,
'trigXEndTime', trigXEndTime,
'trigHCentTime', trigHCentTime,
'trigXCentTime', trigXCentTime,
'trigHCentFreq', trigHCentFreq,
'trigXCentFreq', trigXCentFreq,
'trigHSignific', trigHSignific,
'trigXSignific', trigXSignific,
'--- Veto Analysis Parameters ---', '',
'meanTrigCentTime',meanTrigCentTime,
'totalDuration', totalDuration)
#Check if data is sensible
if((segStartTime<analysisStartTime) | (segEndTime>analysisEndTime)):
logFid.write('ERROR: segment startTime (%d)/stopTime (%d) is outside the analysis interval [%d, %d]\n'%( bcvreadEndTime, bcvreadEndTime, analysisStartTime, analysisEndTime))
elif(bcvreadEndTime-bcvreadStartTime>MAX_LENGTH_DATA_SEG):
logFid.write('ERROR: Segment length %f is larger than the allowed max length of %f..\n'
%(bcvreadEndTime-bcvreadStartTime, MAX_LENGTH_DATA_SEG))
else:
#Read the segment for channel H and channel X
[dataH, samplFreqH] = bcv.readData(frameCache, chanHName, frameTypeChanH, bcvreadStartTime,
bcvreadEndTime, [0], debugLevel)
[dataX, samplFreqX] = bcv.readData(frameCache, chanXName, frameTypeChanX, bcvreadStartTime,
bcvreadEndTime, timeShiftX, debugLevel)
#Check for a read error in the channel H data.
if(not all(samplFreqH)):
logFid.write('ERROR: Cannot load frame data for channel H...\n')
logFid.write('ERROR: Channel H - bcvreadStartTime: %f bcvreadEndTime: %f\n'%( bcvreadStartTime, bcvreadEndTime))
elif(not all(samplFreqX)):
if(len(samplFreqX)==1):
logFid.write('ERROR: Cannot load frame data for channel X\n')
logFid.write('ERROR: Channel X - bcvreadStartTime: %f bcvreadEndTime: %f..\n'%(bcvreadEndTime, bcvreadEndTime ))
elif(len(samplFreqX)==2):
if(samplFreqX[1]==0):
logFid.write('ERROR: Cannot load frame data for channel Y\n')
logFid.write('ERROR: Channel Y - bcvreadStartTime: %f bcvreadEndTime: %f..\n'%(bcvreadEndTime, bcvreadEndTime ))
else:
tempArray = []
# If sampling frequency is different from the one specified,
# resample the data
if(samplFreqH != samplFreq):
dataH = np.asarray([bcv.resample2(dataH[0], samplFreqH[0], samplFreq)])
for iChan in xrange(len(dataX)):
if(samplFreqX[iChan]==samplFreq):
tempArray.append(dataX[iChan])
else:
tempArray.append(bcv.resample2(dataX[iChan], samplFreqX[iChan], samplFreq))
dataX = np.asarray(tempArray)
#if(not all(samplFreqX==samplFreq)):
#index = np.where(samplFreqX!=samplFreq)[0]
#for iDs in index:
#tempArray = bcv.resample2(dataX[iDs], samplFreqX[iDs], samplFreq)
timeH = np.arange(bcvreadStartTime, bcvreadEndTime, 1.0/samplFreq)
timeX = np.arange(bcvreadStartTime-timeShift, bcvreadEndTime-timeShift, 1.0/samplFreq)
if(highPassCutoff>0):
dataH = bcv.highpass(dataH, samplFreq, highPassCutoff)
# In case of bilinear coupling multiply the X and Y channels
# to form a pseudo channel (which a combination of X and Y)
# Also compute some parameters descrining the slow channels(s)
# and store them in vectors.
if(couplingModel=='bilinear'):
segIdx = np.intersect1d(np.where(timeX + timeShift>=segStartTime)[0], np.where(timeX+timeShift<segEndTime)[0])
meanY = np.mean(dataX[1][segIdx])
varY = np.var(dataX[1][segIdx])
maxY = np.max(dataX[1][segIdx])
minY = np.min(dataX[1][segIdx])
maxYMat.append(maxY)
meanYMat.append(meanY)
varYMat.append(varY)
minYMat.append(minY)
#mindY = np.min(np.diff(dataX[iChan][segIdx]))
#maxdY = np.max(np.diff(dataX[iChan][segIdx]))
#meandY= np.mean(np.diff(dataX[iChan][segIdx]))
dataP = np.asarray([dataX[0]*dataX[1]])
#del dataX
#dataX = dataP
#del dataP
else:
meanY = 0
varY = 0
maxY = 0
minY = 0
maxYMat.append(maxY)
meanYMat.append(meanY)
varYMat.append(varY)
minYMat.append(minY)
dataP = np.asarray([dataX[0]])
#mindY = 0
#maxdY = 0
#meandY = 0
if(highPassCutoff>0):
dataP = bcv.highpass(dataP, samplFreq, highPassCutoff)
if(couplingModel=='linear'):
[rHP, rMaxHP] = bcv.linearCouplingCoeff(dataH[0], dataP, timeH, timeX,
transFnXtoH, segStartTime, segEndTime,
timeShift, samplFreq, logFid, debugLevel)
print '%d %f' %(timeShift, rHP)
else:
[rHP, rMaxHP] = bcv.bilinearCouplingCoeff(dataH[0],
dataP, timeH, timeX, segStartTime,
segEndTime,timeShift, samplFreq, logFid,
debugLevel)
analysedTrigIdx+=1
SIGNIFICANCE_THRESH_H = 200.0
SIGNIFICANCE_THRESH_X = 10.0
if (debugLevel>=2):
print 'Printing debug plots\n'
if((trigHSignific>=SIGNIFICANCE_THRESH_H) & (trigXSignific>=SIGNIFICANCE_THRESH_X)):
if(highPassCutoff>0):
tdataX = bcv.highpass(np.asarray([dataX[0]]), samplFreq, highPassCutoff)
import os
print 'making debug plts\n'
debugPlotsFolder = 'debug_plots/' + 'timeshift%d'%(timeShift)
debugPlotsDir = outDir[0] + '/' + debugPlotsFolder
if(not os.path.exists(debugPlotsDir)):
os.system('mkdir -p %s'%(debugPlotsDir))
plot_folder = debugPlotsDir + '/CentXTime_%f_CentHTime_%f'%(trigXCentTime, trigHCentTime)
os.system('mkdir -p %s'%(plot_folder))
props = dict(boxstyle='round', facecolor='wheat', alpha=0.5)
# Plot time series data for Channel X
plt.figure(figsize=(12, 6))
if(len(dataX)>1):
plt.subplot(3, 1, 1)
else:
plt.subplot(2, 1, 1)
plt.plot(timeX - min(timeX), tdataX[0], label='x(t)')
ax = plt.gca()
xmin = segStartTime
xmax = segEndTime
ax.axvline(trigXCentTime-min(timeX), color='r', linestyle='-')
idx = np.intersect1d(np.where(timeX>=segStartTime - timeShift)[0], np.where(timeX<=segEndTime - timeShift))
ymin = np.min(tdataX[0][idx])
ymax = np.max(tdataX[0][idx])
plt.xlim((xmin - timeShift-min(timeX), xmax - timeShift-min(timeX)))
plt.ylim((ymin, ymax))
ax.axvline(trigXStartTime - min(timeX), color='m', linestyle='--')
ax.axvline(trigXEndTime - min(timeX), color='m', linestyle = '--')
ax.text(trigXCentTime-min(timeX),ymax/10.0, '%f'%(trigXCentTime-min(timeX)) )
ax.text(0.3, 0.9, 'Duration=%f\nSignificance=%f\n'%(trigXDuration,trigXSignific ), ha='center', va = 'center', transform=ax.transAxes, fontsize=14,
verticalalignment='top', bbox=props)
plt.xlabel('t[sec] since')
plt.ylabel('Time series data: ' + chanXName[0])
plt.legend()
#Plot time series data for channel H
if(len(dataX)>1):
plt.subplot(3, 1, 2)
else:
plt.subplot(2, 1, 2)
plt.plot(timeH - min(timeH), dataH[0], label='h(t)')
ax = plt.gca()
xmin = segStartTime
xmax = segEndTime
idx = np.intersect1d(np.where(timeH>=segStartTime)[0], np.where(timeH<=segEndTime)[0])
ymin = np.min(dataH[0][idx])
ymax = np.max(dataH[0][idx])
plt.xlim((xmin - min(timeH), xmax - min(timeH)))
plt.ylim((ymin, ymax))
ax.axvline(trigHCentTime-min(timeH), color='r', linestyle='-')
ax.axvline(trigHStartTime -min(timeH), color='m', linestyle='--')
ax.axvline(trigHEndTime-min(timeH), color='m', linestyle = '--')
ax.text(trigHCentTime-min(timeH),ymax/10.0, '%f'%(trigHCentTime-min(timeH)) )
ax.text(0.3, 0.9, 'Duration=%f\nSignificance=%f, r=%f\n'%(trigHDuration,trigHSignific, rHP[0] ), ha='center', va = 'center', transform=ax.transAxes, fontsize=14,
verticalalignment='top', bbox=props)
plt.xlabel('t[sec] since')
plt.ylabel('Time series data: ' + chanHName[0])
plt.legend()
# Plot time series data for channel Y
if(len(dataX)>1):
plt.subplot(3, 1, 3)
plt.plot(timeX - min(timeX), dataX[1], label='y(t)')
xmin = segStartTime
xmax = segEndTime
plt.xlim((xmin - timeShift-min(timeX), xmax - timeShift-min(timeX)))
plt.xlabel('t[sec] since')
plt.ylabel('Time series data: ' + chanXName[1])
plt.legend()
plt.savefig(plot_folder + '/TimeSeries.png', dpi=200)
plt.close()
# Plot spectrogram of X data and Y data
plt.figure(figsize=(12, 8))
import matplotlib.mlab as mlab
plt.subplot(2,1,1)
idx = np.intersect1d(np.where(timeX>=segStartTime - timeShift)[0], np.where(timeX<=segEndTime - timeShift))
Pxx, freq, t = mlab.specgram(tdataX[0][idx], noverlap=0, Fs=samplFreq)
#freqidx = np.intersect1d( np.where(freq>10**(np.floor(np.log10(trigXCentFreq))))[0], np.where(freq<10**(np.ceil(np.log10(trigXCentFreq))))[0])
#if(len(freqidx)==0):
#freqidx = np.intersect1d( np.where(freq>10**(np.floor(np.log10(trigXCentFreq))-1))[0], np.where(freq<10**(np.ceil(np.log10(trigXCentFreq))+1))[0])
t = t + xmin - timeShift-min(timeX)
plt.xlabel('t[sec] since')
plt.ylabel('Fourier Frequencies')
plt.title('channel X specgram')
ax = plt.gca()
#ax.set_yscale('log')
imshow = ax.pcolor(t, freq, np.log10(Pxx))
ax.axhline(trigXCentFreq, color='w',linestyle='--', linewidth=1 )
centTime = trigXCentTime - min(timeX)
ax.axvline(centTime, color='w', linestyle = '--', linewidth = 1)
ax.text(centTime, trigXCentFreq, '(%f, %f)'%(centTime, trigXCentFreq))
#ax.set_ylim((1.0, ymax))
#ax.set_xlim((0.0, bcvreadEndTime - bcvSeadStartTime))
plt.colorbar(imshow)
plt.subplot(2,1,2)
idx = np.intersect1d(np.where(timeH>=segStartTime)[0], np.where(timeH<=segEndTime)[0])
Pxx, freq, t = mlab.specgram(dataH[0][idx], noverlap=0, Fs=samplFreq)
#freqidx = np.intersect1d( np.where(freq>10**(np.floor(np.log10(trigHCentFreq))))[0], np.where(freq<10**(np.ceil(np.log10(trigHCentFreq))))[0])
#if(len(freqidx)==0):
#freqidx = np.intersect1d( np.where(freq>10**(np.floor(np.log10(trigXCentFreq))-1))[0], np.where(freq<10**(np.ceil(np.log10(trigXCentFreq))+1))[0])
t = t+ xmin - min(timeH)
plt.xlabel('t[sec] since')
plt.ylabel('Fourier Frequencies')
plt.title('channel H specgram')
ax = plt.gca()
#ax.set_yscale('log')
imshow = ax.pcolor(t, freq, np.log10(Pxx))
ax.axhline(trigHCentFreq, color='w',linestyle='--', linewidth=1 )
centTime = trigHCentTime - min(timeH)
ax.axvline(centTime, color='w', linestyle='--', linewidth=1)
ax.text(centTime, trigHCentFreq, '(%f,%f)'%(centTime, trigHCentFreq))
plt.colorbar(imshow)
#ax.set_ylim((, ymax))
#ax.set_xlim((0.0, bcvreadEndTime - bcvSeadStartTime))
plt.savefig(plot_folder +'/Specgram.png')
timeShiftVec.append(timeShift)
rHPMat.append(rHP)
rMaxHPMat.append(rMaxHP)
#mindYMat.append(mindY)
#maxdYMat.append(maxdY)
#meandYMat.append(meandY)
trigHAnalysdCentTimeVec.append(trigHCentTime)
trigXAnalysdCentTimeVec.append(trigXCentTime)
trigHAnalysdCentFreqVec.append(trigHCentFreq)
trigXAnalysdCentFreqVec.append(trigXCentFreq)
trigHAnalysdSignificVec.append(trigHSignific)
trigXAnalysdSignificVec.append(trigXSignific)
trigHAnalysdDurationVec.append(trigHDuration)
trigXAnalysdDurationVec.append(trigXDuration)
timeShiftVec = np.asarray(timeShiftVec)
rHPMat = np.asarray(rHPMat)
rMaxHPMat = np.asarray(rMaxHPMat)
meanYMat = np.asarray(meanYMat)
varYMat = np.asarray(varYMat)
minYMat = np.asarray(minYMat)
#mindYMat = np.asarray(mindYMat)
#maxYMat = np.asarray(maxYMat)
#maxdYMat = np.asarray(maxdYMat)
#meandYMat = np.asarray(meandYMat)
trigHAnalysdCentTimeVec = np.asarray(trigHAnalysdCentTimeVec)
trigXAnalysdCentTimeVec = np.asarray(trigXAnalysdCentTimeVec)
trigHAnalysdCentFreqVec = np.asarray(trigHAnalysdCentFreqVec)
trigXAnalysdCentFreqVec = np.asarray(trigXAnalysdCentFreqVec)
trigHAnalysdSignificVec = np.asarray(trigHAnalysdSignificVec)
trigXAnalysdSignificVec = np.asarray(trigXAnalysdSignificVec)
trigHAnalysdDurationVec = np.asarray(trigHAnalysdDurationVec)
trigXAnalysdDurationVec = np.asarray(trigXAnalysdDurationVec)
# combine the correlation from multiple pseudochannels
# assmuing that all pseudochannels are orthogonal
# (not strictly true)
#rHPCombVec = np.sqrt(np.sum(rHPMat**2.0, axis=1))
#rMaxHPCombVec = np.sqrt(np.sum(rMaxHPMat**2.0, axis=1))
outFileString = '/corrstat_timeshift%d_seg%d-%d.dat'%(timeShift, analysisStartTime, analysisEndTime)
#save the analysis results from each pseudochannel in a separate file
if(analysedTrigIdx>0):
for iP in range(0, len(outDir)):
outFileName = outDir[iP] + '/' + outFileString
resultsMatrix = np.asarray([timeShiftVec, rHPMat[:, iP], rMaxHPMat[:, iP],
trigHAnalysdCentTimeVec, trigXAnalysdCentTimeVec, trigHAnalysdCentFreqVec, trigXAnalysdCentFreqVec, trigHAnalysdSignificVec, trigXAnalysdSignificVec, trigHAnalysdDurationVec, trigXAnalysdDurationVec, meanYMat, varYMat,
maxYMat, minYMat], dtype=np.float64).transpose()
resultsMatrix = resultsMatrix[resultsMatrix[:,5 ].argsort()]
np.savetxt(outFileName, resultsMatrix, delimiter = ' ', fmt = '%f')
del resultsMatrix, timeShiftVec, rHPMat, rMaxHPMat
del trigHAnalysdCentTimeVec, trigXAnalysdCentTimeVec, trigHAnalysdCentFreqVec
del trigXAnalysdCentFreqVec, trigHAnalysdSignificVec, trigXAnalysdSignificVec
del trigHAnalysdDurationVec, trigXAnalysdDurationVec
del meanYMat, varYMat, maxYMat, minYMat, mindYMat, maxdYMat, meandYMat
else:
logFid.write('WARNING: No coincident triggers found for timeshift = %d..\n' %(timeShift))
def vetoanalysis(frameCache, chanHName, chanXName, frameTypeChanH,
frameTypeChanX, samplFreqH, samplFreqX, highPassCutoff,
TriggerHList, TriggerXList, couplingModel, transFnXtoH,
analysisStartTime, analysisEndTime, timeShift, outDir, logFid,
debugLevel):
#VETOANALYSIS - Peform veto analysis using instrumental couplings on a set of triggers
#in the GW channel making use of a set of trigges in the an instrumental channel X and
#and the transfer function from channel X to the GW channel
#
# Usage : vetoanalysis(TriggerHList, TriggerXList, transFnXtoH,numberOfChannels, timeShift)
# TriggerHList - A structure containing the start times, and times and central times
# of the triggers in channel H
# TriggerXList - A structure containing the strat times, end times and central times of
# the triggers in channel X
# TransFnXtoH - A structure containing the transfer functions (Frequency and Amplitude) from channel X to Channel H
# analysisStartTime - GPS start time lower bound of candidate events
# analysisEndTime - GPS end times upper bound of candidate events
# timeShift - The number of seconds by whuch the coincident triggers
# will be time shifted after identification
# Sudarshan Ghonge <*****@*****.**>
# Original code written by
# Aaron B. Pearlman <*****@*****.**> and
# P. Ajith <ajith.icts.res.in>
#maximum number of allowed coincidences
maxNumCoinc = len(TriggerHList.centralTime) * len(TriggerXList.centralTime)
#Time window for identifying coincidences between channel H and X
COINC_TIME_WINDOW = 0.5
#Maximum length of one data segment used for the analysis
MAX_LENGTH_DATA_SEG = 64
################Ask Ajith what this means#
segLength = 3600
uniqueArgument = 'nonunique'
logFid.write( 'LOG: Finding coincidences... Num trigs H = %d, Num trigs X = %d\n'\
%(len(TriggerHList.centralTime), len(TriggerXList.centralTime)))
[coincTrigH,
coincTrigX] = bcv.mcoinc(maxNumCoinc, TriggerHList.centralTime,
TriggerXList.centralTime + timeShift,
COINC_TIME_WINDOW, segLength, uniqueArgument)
trigHCentralTimeVec = TriggerHList.centralTime[coincTrigH]
trigXCentralTimeVec = TriggerXList.centralTime[coincTrigX]
trigHStartTimeVec = TriggerHList.startTime[coincTrigH]
trigXStartTimeVec = TriggerXList.startTime[coincTrigX]
trigHEndTimeVec = TriggerHList.endTime[coincTrigH]
trigXEndTimeVec = TriggerXList.endTime[coincTrigX]
trigHCentFreqVec = TriggerHList.centralFrequency[coincTrigH]
trigXCentFreqVec = TriggerXList.centralFrequency[coincTrigX]
trigHSignificVec = TriggerHList.triggerSignificance[coincTrigH]
trigXSignificVec = TriggerXList.triggerSignificance[coincTrigX]
trigHDurationVec = trigHEndTimeVec - trigHStartTimeVec
trigXDurationVec = trigXEndTimeVec - trigXStartTimeVec
del TriggerHList, TriggerXList
##Read Data Segments Using a Time-Window Around Coi-incident Triggers
##Check if the sample frequencies of Channel H X are the same. If not, throw error
#if(samplFreqH!=samplFreqX):
#sys.exit('Error: samplFreqH DOES NOT equal samplFreqX')
#else:
#samplFreq = samplFreqH
samplFreq = samplFreqH
# Apply time shift to the X/Y data. In case of bilinear coupling, there are multiple
# channels (X and Y); thus a vector of length 2 or more
if (couplingModel == 'linear'):
timeShiftX = np.array([timeShift])
elif (couplingModel == 'bilinear'):
timeShiftX = np.zeros(len(chanXName))
for iX in range(len(chanXName)):
timeShiftX[iX] = timeShift
#Check number of triggers in channel H that are coincident in channel X (coincTrigH)
#is the same the number of triggers in Channel X that are coincident in Channel H
#(coincTrigX)
if (len(coincTrigH) == len(coincTrigX) & len(coincTrigH) > 0):
#Read data for timeShift from frameCache
logFid.write('LOG: Performing veto analysis for time shift %d..\n' %
(timeShift))
logFid.write('LOG: Number of coincidences : %d...\n' %
(len(coincTrigH)))
#Initializing empty lists to store the post analysis data
timeShiftVec = []
rHPMat = []
rMaxHPMat = []
meanYMat = []
varYMat = []
varYMat = []
minYMat = []
maxYMat = []
mindYMat = []
maxdYMat = []
meandYMat = []
trigHAnalysdCentTimeVec = []
trigXAnalysdCentTimeVec = []
trigHAnalysdCentFreqVec = []
trigXAnalysdCentFreqVec = []
trigHAnalysdSignificVec = []
trigXAnalysdSignificVec = []
trigHAnalysdDurationVec = []
trigXAnalysdDurationVec = []
#testId = open('test_' + '%d'%(analysisStartTime) + '.dat', 'w+')
#np.savetxt(testId, trigHCentralTimeVec)
#testId.close()
#outFileString = '/corrstat_timeshift%d_seg%d-%d.dat'%(timeShift, analysisStartTime, analysisEndTime)
#outFileName = outDir[iP] + '/' + outFileString
#outFid = open(outFileName, 'w+')
analysedTrigIdx = 0
for coincIndex in range(len(coincTrigH)):
trigHCentTime = trigHCentralTimeVec[coincIndex]
trigXCentTime = trigXCentralTimeVec[coincIndex]
trigHStartTime = trigHStartTimeVec[coincIndex]
trigXStartTime = trigXStartTimeVec[coincIndex]
trigHEndTime = trigHEndTimeVec[coincIndex]
trigXEndTime = trigXEndTimeVec[coincIndex]
trigHCentFreq = trigHCentFreqVec[coincIndex]
trigXCentFreq = trigXCentFreqVec[coincIndex]
trigHSignific = trigHSignificVec[coincIndex]
trigXSignific = trigXSignificVec[coincIndex]
trigHDuration = trigHEndTime - trigHStartTime
trigXDuration = trigXEndTime - trigXStartTime
# Find the segent of data used for analysis
segStartTime = min(trigHStartTime, trigXStartTime + timeShift)
segEndTime = max(trigHEndTime, trigXEndTime + timeShift)
meanTrigCentTime = (segStartTime + segEndTime) / 2.0
totalDuration = segEndTime - segStartTime
# Calculate the total number of samples, rounded to the closest integer
# power of 2
totalNumberSamples = bcv.roundtopowertwo(totalDuration * samplFreq,
1024.0)
# Calculate the length of the data segment used for veto analysis
segmentDuration = totalNumberSamples / samplFreq
# Find the start and end times of a samll segment of dara that we want to
# analyse
segStartTime = meanTrigCentTime - segmentDuration / 2.0
segEndTime = meanTrigCentTime + segmentDuration / 2.0
# Round the start and end times of a small segment of dara that we want to
# analyse to an integer of gps time.
# IMPORTANT NOTE: One second of data in the beginning is used to train the high
# pass filter and hecne should not be used for the analysis (this explains
# the "floor(segStartTime) - 1" below).
bcvreadStartTime = np.floor(segStartTime) - 1
bcvreadEndTime = np.ceil(segEndTime) + 1
if (debugLevel >= 1):
#print values of different variables
bcv.printvar(
'--- Analysis Window ---', '', 'analysisStartTime',
analysisStartTime, 'analysisEndTime', analysisEndTime,
'wreadStartTime', bcvreadStartTime, 'wreadEndTime',
bcvreadEndTime, 'segStartTime', segStartTime, 'segEndTime',
segEndTime, '--- Trigger Parameters ---', '',
'trigHStartTime', trigHStartTime, 'trigXStartTime',
trigXStartTime, 'trigHEndTime', trigHEndTime,
'trigXEndTime', trigXEndTime, 'trigHCentTime',
trigHCentTime, 'trigXCentTime', trigXCentTime,
'trigHCentFreq', trigHCentFreq, 'trigXCentFreq',
trigXCentFreq, 'trigHSignific', trigHSignific,
'trigXSignific', trigXSignific,
'--- Veto Analysis Parameters ---', '', 'meanTrigCentTime',
meanTrigCentTime, 'totalDuration', totalDuration)
#Check if data is sensible
if ((segStartTime < analysisStartTime) |
(segEndTime > analysisEndTime)):
logFid.write(
'ERROR: segment startTime (%d)/stopTime (%d) is outside the analysis interval [%d, %d]\n'
% (bcvreadEndTime, bcvreadEndTime, analysisStartTime,
analysisEndTime))
elif (bcvreadEndTime - bcvreadStartTime > MAX_LENGTH_DATA_SEG):
logFid.write(
'ERROR: Segment length %f is larger than the allowed max length of %f..\n'
% (bcvreadEndTime - bcvreadStartTime, MAX_LENGTH_DATA_SEG))
else:
#Read the segment for channel H and channel X
[dataH,
samplFreqH] = bcv.readData(frameCache, chanHName,
frameTypeChanH, bcvreadStartTime,
bcvreadEndTime, [0], debugLevel)
[dataX,
samplFreqX] = bcv.readData(frameCache, chanXName,
frameTypeChanX, bcvreadStartTime,
bcvreadEndTime, timeShiftX,
debugLevel)
#Check for a read error in the channel H data.
if (not all(samplFreqH)):
logFid.write(
'ERROR: Cannot load frame data for channel H...\n')
logFid.write(
'ERROR: Channel H - bcvreadStartTime: %f bcvreadEndTime: %f\n'
% (bcvreadStartTime, bcvreadEndTime))
elif (not all(samplFreqX)):
if (len(samplFreqX) == 1):
logFid.write(
'ERROR: Cannot load frame data for channel X\n')
logFid.write(
'ERROR: Channel X - bcvreadStartTime: %f bcvreadEndTime: %f..\n'
% (bcvreadEndTime, bcvreadEndTime))
elif (len(samplFreqX) == 2):
if (samplFreqX[1] == 0):
logFid.write(
'ERROR: Cannot load frame data for channel Y\n'
)
logFid.write(
'ERROR: Channel Y - bcvreadStartTime: %f bcvreadEndTime: %f..\n'
% (bcvreadEndTime, bcvreadEndTime))
else:
tempArray = []
# If sampling frequency is different from the one specified,
# resample the data
if (samplFreqH != samplFreq):
dataH = np.asarray([
bcv.resample2(dataH[0], samplFreqH[0], samplFreq)
])
for iChan in xrange(len(dataX)):
if (samplFreqX[iChan] == samplFreq):
tempArray.append(dataX[iChan])
else:
tempArray.append(
bcv.resample2(dataX[iChan], samplFreqX[iChan],
samplFreq))
dataX = np.asarray(tempArray)
#if(not all(samplFreqX==samplFreq)):
#index = np.where(samplFreqX!=samplFreq)[0]
#for iDs in index:
#tempArray = bcv.resample2(dataX[iDs], samplFreqX[iDs], samplFreq)
timeH = np.arange(bcvreadStartTime, bcvreadEndTime,
1.0 / samplFreq)
timeX = np.arange(bcvreadStartTime - timeShift,
bcvreadEndTime - timeShift,
1.0 / samplFreq)
if (highPassCutoff > 0):
dataH = bcv.highpass(dataH, samplFreq, highPassCutoff)
# In case of bilinear coupling multiply the X and Y channels
# to form a pseudo channel (which a combination of X and Y)
# Also compute some parameters descrining the slow channels(s)
# and store them in vectors.
if (couplingModel == 'bilinear'):
segIdx = np.intersect1d(
np.where(timeX + timeShift >= segStartTime)[0],
np.where(timeX + timeShift < segEndTime)[0])
meanY = np.mean(dataX[1][segIdx])
varY = np.var(dataX[1][segIdx])
maxY = np.max(dataX[1][segIdx])
minY = np.min(dataX[1][segIdx])
maxYMat.append(maxY)
meanYMat.append(meanY)
varYMat.append(varY)
minYMat.append(minY)
#mindY = np.min(np.diff(dataX[iChan][segIdx]))
#maxdY = np.max(np.diff(dataX[iChan][segIdx]))
#meandY= np.mean(np.diff(dataX[iChan][segIdx]))
dataP = np.asarray([dataX[0] * dataX[1]])
#del dataX
#dataX = dataP
#del dataP
else:
meanY = 0
varY = 0
maxY = 0
minY = 0
maxYMat.append(maxY)
meanYMat.append(meanY)
varYMat.append(varY)
minYMat.append(minY)
dataP = np.asarray([dataX[0]])
#mindY = 0
#maxdY = 0
#meandY = 0
if (highPassCutoff > 0):
dataP = bcv.highpass(dataP, samplFreq, highPassCutoff)
if (couplingModel == 'linear'):
[rHP, rMaxHP] = bcv.linearCouplingCoeff(
dataH[0], dataP, timeH, timeX, transFnXtoH,
segStartTime, segEndTime, timeShift, samplFreq,
logFid, debugLevel)
print '%d %f' % (timeShift, rHP)
else:
[rHP, rMaxHP] = bcv.bilinearCouplingCoeff(
dataH[0], dataP, timeH, timeX, segStartTime,
segEndTime, timeShift, samplFreq, logFid,
debugLevel)
analysedTrigIdx += 1
SIGNIFICANCE_THRESH_H = 200.0
SIGNIFICANCE_THRESH_X = 10.0
if (debugLevel >= 2):
print 'Printing debug plots\n'
if ((trigHSignific >= SIGNIFICANCE_THRESH_H) &
(trigXSignific >= SIGNIFICANCE_THRESH_X)):
if (highPassCutoff > 0):
tdataX = bcv.highpass(np.asarray([dataX[0]]),
samplFreq,
highPassCutoff)
import os
print 'making debug plts\n'
debugPlotsFolder = 'debug_plots/' + 'timeshift%d' % (
timeShift)
debugPlotsDir = outDir[0] + '/' + debugPlotsFolder
if (not os.path.exists(debugPlotsDir)):
os.system('mkdir -p %s' % (debugPlotsDir))
plot_folder = debugPlotsDir + '/CentXTime_%f_CentHTime_%f' % (
trigXCentTime, trigHCentTime)
os.system('mkdir -p %s' % (plot_folder))
props = dict(boxstyle='round',
facecolor='wheat',
alpha=0.5)
# Plot time series data for Channel X
plt.figure(figsize=(12, 6))
if (len(dataX) > 1):
plt.subplot(3, 1, 1)
else:
plt.subplot(2, 1, 1)
plt.plot(timeX - min(timeX),
tdataX[0],
label='x(t)')
ax = plt.gca()
xmin = segStartTime
xmax = segEndTime
ax.axvline(trigXCentTime - min(timeX),
color='r',
linestyle='-')
idx = np.intersect1d(
np.where(timeX >= segStartTime - timeShift)[0],
np.where(timeX <= segEndTime - timeShift))
ymin = np.min(tdataX[0][idx])
ymax = np.max(tdataX[0][idx])
plt.xlim((xmin - timeShift - min(timeX),
xmax - timeShift - min(timeX)))
plt.ylim((ymin, ymax))
ax.axvline(trigXStartTime - min(timeX),
color='m',
linestyle='--')
ax.axvline(trigXEndTime - min(timeX),
color='m',
linestyle='--')
ax.text(trigXCentTime - min(timeX), ymax / 10.0,
'%f' % (trigXCentTime - min(timeX)))
ax.text(0.3,
0.9,
'Duration=%f\nSignificance=%f\n' %
(trigXDuration, trigXSignific),
ha='center',
va='center',
transform=ax.transAxes,
fontsize=14,
verticalalignment='top',
bbox=props)
plt.xlabel('t[sec] since')
plt.ylabel('Time series data: ' + chanXName[0])
plt.legend()
#Plot time series data for channel H
if (len(dataX) > 1):
plt.subplot(3, 1, 2)
else:
plt.subplot(2, 1, 2)
plt.plot(timeH - min(timeH),
dataH[0],
label='h(t)')
ax = plt.gca()
xmin = segStartTime
xmax = segEndTime
idx = np.intersect1d(
np.where(timeH >= segStartTime)[0],
np.where(timeH <= segEndTime)[0])
ymin = np.min(dataH[0][idx])
ymax = np.max(dataH[0][idx])
plt.xlim((xmin - min(timeH), xmax - min(timeH)))
plt.ylim((ymin, ymax))
ax.axvline(trigHCentTime - min(timeH),
color='r',
linestyle='-')
ax.axvline(trigHStartTime - min(timeH),
color='m',
linestyle='--')
ax.axvline(trigHEndTime - min(timeH),
color='m',
linestyle='--')
ax.text(trigHCentTime - min(timeH), ymax / 10.0,
'%f' % (trigHCentTime - min(timeH)))
ax.text(0.3,
0.9,
'Duration=%f\nSignificance=%f, r=%f\n' %
(trigHDuration, trigHSignific, rHP[0]),
ha='center',
va='center',
transform=ax.transAxes,
fontsize=14,
verticalalignment='top',
bbox=props)
plt.xlabel('t[sec] since')
plt.ylabel('Time series data: ' + chanHName[0])
plt.legend()
# Plot time series data for channel Y
if (len(dataX) > 1):
plt.subplot(3, 1, 3)
plt.plot(timeX - min(timeX),
dataX[1],
label='y(t)')
xmin = segStartTime
xmax = segEndTime
plt.xlim((xmin - timeShift - min(timeX),
xmax - timeShift - min(timeX)))
plt.xlabel('t[sec] since')
plt.ylabel('Time series data: ' + chanXName[1])
plt.legend()
plt.savefig(plot_folder + '/TimeSeries.png',
dpi=200)
plt.close()
# Plot spectrogram of X data and Y data
plt.figure(figsize=(12, 8))
import matplotlib.mlab as mlab
plt.subplot(2, 1, 1)
idx = np.intersect1d(
np.where(timeX >= segStartTime - timeShift)[0],
np.where(timeX <= segEndTime - timeShift))
Pxx, freq, t = mlab.specgram(tdataX[0][idx],
noverlap=0,
Fs=samplFreq)
#freqidx = np.intersect1d( np.where(freq>10**(np.floor(np.log10(trigXCentFreq))))[0], np.where(freq<10**(np.ceil(np.log10(trigXCentFreq))))[0])
#if(len(freqidx)==0):
#freqidx = np.intersect1d( np.where(freq>10**(np.floor(np.log10(trigXCentFreq))-1))[0], np.where(freq<10**(np.ceil(np.log10(trigXCentFreq))+1))[0])
t = t + xmin - timeShift - min(timeX)
plt.xlabel('t[sec] since')
plt.ylabel('Fourier Frequencies')
plt.title('channel X specgram')
ax = plt.gca()
#ax.set_yscale('log')
imshow = ax.pcolor(t, freq, np.log10(Pxx))
ax.axhline(trigXCentFreq,
color='w',
linestyle='--',
linewidth=1)
centTime = trigXCentTime - min(timeX)
ax.axvline(centTime,
color='w',
linestyle='--',
linewidth=1)
ax.text(centTime, trigXCentFreq,
'(%f, %f)' % (centTime, trigXCentFreq))
#ax.set_ylim((1.0, ymax))
#ax.set_xlim((0.0, bcvreadEndTime - bcvSeadStartTime))
plt.colorbar(imshow)
plt.subplot(2, 1, 2)
idx = np.intersect1d(
np.where(timeH >= segStartTime)[0],
np.where(timeH <= segEndTime)[0])
Pxx, freq, t = mlab.specgram(dataH[0][idx],
noverlap=0,
Fs=samplFreq)
#freqidx = np.intersect1d( np.where(freq>10**(np.floor(np.log10(trigHCentFreq))))[0], np.where(freq<10**(np.ceil(np.log10(trigHCentFreq))))[0])
#if(len(freqidx)==0):
#freqidx = np.intersect1d( np.where(freq>10**(np.floor(np.log10(trigXCentFreq))-1))[0], np.where(freq<10**(np.ceil(np.log10(trigXCentFreq))+1))[0])
t = t + xmin - min(timeH)
plt.xlabel('t[sec] since')
plt.ylabel('Fourier Frequencies')
plt.title('channel H specgram')
ax = plt.gca()
#ax.set_yscale('log')
imshow = ax.pcolor(t, freq, np.log10(Pxx))
ax.axhline(trigHCentFreq,
color='w',
linestyle='--',
linewidth=1)
centTime = trigHCentTime - min(timeH)
ax.axvline(centTime,
color='w',
linestyle='--',
linewidth=1)
ax.text(centTime, trigHCentFreq,
'(%f,%f)' % (centTime, trigHCentFreq))
plt.colorbar(imshow)
#ax.set_ylim((, ymax))
#ax.set_xlim((0.0, bcvreadEndTime - bcvSeadStartTime))
plt.savefig(plot_folder + '/Specgram.png')
timeShiftVec.append(timeShift)
rHPMat.append(rHP)
rMaxHPMat.append(rMaxHP)
#mindYMat.append(mindY)
#maxdYMat.append(maxdY)
#meandYMat.append(meandY)
trigHAnalysdCentTimeVec.append(trigHCentTime)
trigXAnalysdCentTimeVec.append(trigXCentTime)
trigHAnalysdCentFreqVec.append(trigHCentFreq)
trigXAnalysdCentFreqVec.append(trigXCentFreq)
trigHAnalysdSignificVec.append(trigHSignific)
trigXAnalysdSignificVec.append(trigXSignific)
trigHAnalysdDurationVec.append(trigHDuration)
trigXAnalysdDurationVec.append(trigXDuration)
timeShiftVec = np.asarray(timeShiftVec)
rHPMat = np.asarray(rHPMat)
rMaxHPMat = np.asarray(rMaxHPMat)
meanYMat = np.asarray(meanYMat)
varYMat = np.asarray(varYMat)
minYMat = np.asarray(minYMat)
#mindYMat = np.asarray(mindYMat)
#maxYMat = np.asarray(maxYMat)
#maxdYMat = np.asarray(maxdYMat)
#meandYMat = np.asarray(meandYMat)
trigHAnalysdCentTimeVec = np.asarray(trigHAnalysdCentTimeVec)
trigXAnalysdCentTimeVec = np.asarray(trigXAnalysdCentTimeVec)
trigHAnalysdCentFreqVec = np.asarray(trigHAnalysdCentFreqVec)
trigXAnalysdCentFreqVec = np.asarray(trigXAnalysdCentFreqVec)
trigHAnalysdSignificVec = np.asarray(trigHAnalysdSignificVec)
trigXAnalysdSignificVec = np.asarray(trigXAnalysdSignificVec)
trigHAnalysdDurationVec = np.asarray(trigHAnalysdDurationVec)
trigXAnalysdDurationVec = np.asarray(trigXAnalysdDurationVec)
# combine the correlation from multiple pseudochannels
# assmuing that all pseudochannels are orthogonal
# (not strictly true)
#rHPCombVec = np.sqrt(np.sum(rHPMat**2.0, axis=1))
#rMaxHPCombVec = np.sqrt(np.sum(rMaxHPMat**2.0, axis=1))
outFileString = '/corrstat_timeshift%d_seg%d-%d.dat' % (
timeShift, analysisStartTime, analysisEndTime)
#save the analysis results from each pseudochannel in a separate file
if (analysedTrigIdx > 0):
for iP in range(0, len(outDir)):
outFileName = outDir[iP] + '/' + outFileString
resultsMatrix = np.asarray([
timeShiftVec, rHPMat[:, iP], rMaxHPMat[:, iP],
trigHAnalysdCentTimeVec, trigXAnalysdCentTimeVec,
trigHAnalysdCentFreqVec, trigXAnalysdCentFreqVec,
trigHAnalysdSignificVec, trigXAnalysdSignificVec,
trigHAnalysdDurationVec, trigXAnalysdDurationVec, meanYMat,
varYMat, maxYMat, minYMat
],
dtype=np.float64).transpose()
resultsMatrix = resultsMatrix[resultsMatrix[:, 5].argsort()]
np.savetxt(outFileName, resultsMatrix, delimiter=' ', fmt='%f')
del resultsMatrix, timeShiftVec, rHPMat, rMaxHPMat
del trigHAnalysdCentTimeVec, trigXAnalysdCentTimeVec, trigHAnalysdCentFreqVec
del trigXAnalysdCentFreqVec, trigHAnalysdSignificVec, trigXAnalysdSignificVec
del trigHAnalysdDurationVec, trigXAnalysdDurationVec
del meanYMat, varYMat, maxYMat, minYMat, mindYMat, maxdYMat, meandYMat
else:
logFid.write(
'WARNING: No coincident triggers found for timeshift = %d..\n' %
(timeShift))
folder = os.popen('pwd').readlines()[0].split('\n')[0]
print 'folder: ', folder
channelXName = trigxfile.split('_', 3)[3].split('.')[0]
channelHName = trighfile.split('_', 3)[3].split('.')[0]
segIdx = np.intersect1d(
np.where(trigXData[:, 2] > startTime)[0],
np.where(trigXData[:, 2] < endTime)[0])
trigXData = trigXData[segIdx]
segIdh = np.intersect1d(
np.where(trigHData[:, 2] > startTime)[0],
np.where(trigHData[:, 2] < endTime)[0])
trigHData = trigHData[segIdh]
[coincTrigH,
coincTrigX] = bcv.mcoinc(maxNumCoinc, trigHData[:, 2], trigXData[:, 2],
COINC_TIME_WINDOW, segLength, uniqueArgument)
plt.figure()
plt.suptitle('Triggers between %s and %s' %
(gtime.from_gps(startTime), gtime.from_gps(endTime)))
plt.subplot(2, 1, 1)
ax = plt.gca()
ax.errorbar(trigXData[:, 2],
trigXData[:, 3],
xerr=[
trigXData[:, 2] - trigXData[:, 0],
trigXData[:, 1] - trigXData[:, 2]
],
fmt='o',
label='Triggers')
plt.plot(trigXData[coincTrigX, 2],
