def bispectrumplot(self, plotcand={}, show=1, save=0):
""" Produces plot of bispectrum for given candidate.
"""
if len(plotcand) != 1:
print 'plotcands should have exactly one beam.'
return 0
# collect candidate params
(ra, dec) = plotcand.keys()[0]
cand = plotcand[(ra,dec)][0]
startint = self.startints[0]
print 'Producing bispectrum plot for beam', ra, dec, ', candidate, ', cand
self.obs = tpipe.pipe_msdisp2(file=self.datafile, nints=self.nints, nskip=startint, profile='default', selectpol=self.selectpol, scan=self.scan, spw=self.spw, chans=self.chans, dmarr=self.dmarr, datacol='corrected_data')
# select data for candidate
self.obs.time_filter(cand[0], self.filtershape, bgwindow=self.bgwindow)
noiseperbl = self.obs.data.mean(axis=3).mean(axis=2).real.std() # measure single noise for input to detect_bispectra
l1 = n.radians(ra); m1 = n.radians(dec)
if ( (l1 != self.obs.l0[0]) | (m1 != self.obs.m0[0]) ):
dl = l1 - self.obs.l0[0]; dm = m1 - self.obs.m0[0] # find phase shift for new location
self.obs.phaseshift(dl, dm)
# find candidates
self.obs.make_bispectra(stokes=self.stokes)
(candsnr, candstd, cands) = self.obs.detect_bispectra(sigma=self.sigma, Q=noiseperbl, show=show, save=save)
def candplot2(self, plotcands={}, save=0):
""" Builds summary plots, including phased beam and imaging for each plot candidate.
This is a "tier 2" plot, since it only uses all interferometric info for a given candidate data location.
Candidate should be dictionary with keys of beam location and value of a list of 3-element lists (dt, int, dmbin).
Note: this starts from scratch by reading data to produce a set of plots tailored to each candidate.
No requirement for do_search and find_candidates to be run first.
"""
print 'Building Tier 2 candidate plot.'
if n.array([len(plotcands[beam]) for beam in plotcands.keys()]).sum() == 0:
print 'No candidates available...'
return
# twindow = 128 # number of integrations to read for plots
figi = 1
for beam in plotcands.keys():
if len(plotcands[beam]) == 0:
continue
if (n.ndim(plotcands[beam]) != 2) | (n.shape(plotcands[beam])[1] != 3):
print 'Candidate does not seem to be in proper format. Skipping.'
continue
(ra, dec) = beam
print 'Generating plots for candidate with delaycenter: (%s, %s)' % (str(ra), str(dec))
for cand in plotcands[beam]:
(dtind, ii, dmind) = cand
print 'Candidate at (dtind, ii, dmind) = (%d, %d, %d)' % (dtind, ii, dmind)
if self.datatype == 'mir':
# older version centered pulse plot in time
# obs = tpipe.pipe_mirdisp2(file=self.datafile, nints=twindow, nskip=self.startints[0]+ii-twindow/2, profile='default', dmarr=self.dmarr, chans=self.chans)
self.obs = tpipe.pipe_mirdisp2(file=self.datafile, nints=self.nints, nskip=self.startints[0], profile='default', dmarr=self.dmarr, chans=self.chans)
elif self.datatype == 'ms':
# obs = tpipe.pipe_msdisp2(file=self.datafile, nints=twindow, nskip=self.startints[0]+ii-twindow/2, profile='default', selectpol=self.selectpol, scan=self.scan, spw=self.spw, chans=self.chans, dmarr=self.dmarr, datacol='corrected_data')
self.obs = tpipe.pipe_msdisp2(file=self.datafile, nints=self.nints, nskip=self.startints[0], profile='default', selectpol=self.selectpol, scan=self.scan, spw=self.spw, chans=self.chans, dmarr=self.dmarr, datacol='corrected_data')
elif self.datatype == 'sim':
print 'For datatype=\'sim\', we have no original data.'
# obs = tpipe.pipe_simdisp2(array='vla_d', nints=twindow, profile='default', dmarr=self.dmarr)
# obs.add_transient(0., 0., 1., twindow/2)
self.obs = tpipe.pipe_simdisp2(array='vla_d', nints=self.nints, profile='default', dmarr=self.dmarr)
self.obs.add_transient(0., 0., 1., ii)
self.obs.time_filter(self.timescales[dtind], self.filtershape, bgwindow=self.bgwindow) # tophat filter (bgwindow not used)
noiseperbl = self.obs.data.mean(axis=3).mean(axis=2).real.std() # measure single noise for input to detect_bispectra
# shift phase center to candidate
l1 = n.radians(ra); m1 = n.radians(dec)
if ( (l1 != self.obs.l0[0]) | (m1 != self.obs.m0[0]) ):
dl = l1 - self.obs.l0[0]; dm = m1 - self.obs.m0[0] # find phase shift for new location (in degrees)
self.obs.phaseshift(dl, dm)
# reproduce bisp candidates
self.obs.make_bispectra(stokes=self.stokes)
(candsnr, candstd, cands) = self.obs.detect_bispectra(sigma=self.sigma, Q=noiseperbl, save=0)
print 'Reproduced bispectrum candidate, ', cands
# make image of field
# im = obs.imagetrack(obs.tracksub(dmind, twindow/2), i=twindow/2, pol='i', size=self.size, res=self.res, clean=1, show=0)
im = self.obs.imagetrack(self.obs.tracksub(dmind, ii), i=ii, pol='i', size=self.size, res=self.res, clean=1, show=0)
imsize = float(self.size)/self.res
pixelscale = 1./self.size
peakm, peakl = n.where(im == im.max())
dm = -(imsize/2. - peakm[0])*pixelscale # indexed starting at top...
dl = -(peakl[0] - imsize/2.)*pixelscale # indexed starting at left.
# phase shift to center for spectral analysis
self.obs.phaseshift(dl, dm)
# specmod = obs.specmod(dmind, twindow/2)
specmod = self.obs.specmod(dmind, ii)
# now plot
p.figure(figi)
# text description of candidate
p.subplot(221, axisbg='white')
p.title('Candidate @ Tier 2')
p.xticks([])
p.yticks([])
p.text(0.1, 0.8, self.datafile, fontname='sans-serif')
beamra = n.round(beam, 3)[0]
beamdec = n.round(beam, 3)[1]
p.text(0.1, 0.6, 'Beam: ' + str(beamra) + ', ' + str(beamdec) + ', dt: ' + str(self.timescales[cand[0]]), fontname='sans-serif')
p.text(0.1, 0.4, 'Integration: ' + str(cand[1]) + ', DM: ' + str(self.dmarr[cand[2]]), fontname='sans-serif')
# ww = n.where([n.all(mastercand == cand) for mastercand in self.masterloc[beam]])[0][0]
p.text(0.1, 0.2, 'SNR: ' + str(n.round(candsnr, 1)), fontname='sans-serif')
# image of dispersed visibilities for candidate
p.subplot(222)
fov = n.degrees(1./self.res)*3600. # field of view in arcseconds
p.imshow(im, aspect='auto', origin='lower', interpolation='nearest', extent=[fov/2, -fov/2, -fov/2, fov/2])
p.colorbar()
p.xlabel('RA/l Offset (arcsec)')
p.ylabel('Dec/m Offset (arcsec)')
p.subplot(223)
# spec = obs.dedisperse(dmind).mean(axis=3).mean(axis=1).real[twindow/2] # stokes I
spec = self.obs.dedisperse(dmind).mean(axis=3).mean(axis=1).real[ii] # stokes I
p.plot(self.obs.freq, spec, '.')
p.text(0.1, 0.1, 'specmod =' + str(specmod))
p.xlabel('Frequency (GHz)')
p.ylabel('Flux density (Jy)')
p.subplot(224)
dataph = n.rot90(self.obs.dataph)
sh = dataph.shape
im = p.imshow(dataph, aspect='auto', origin='upper', interpolation='nearest', extent=(0, sh[1], 0, sh[0]))
p.colorbar()
p.plot(self.obs.dmtrack0[dmind][0], self.obs.dmtrack0[dmind][1],'k.') # reference dispersed track
p.xlabel('Integration')
p.ylabel('Channel')
if self.datatype == 'mir': # casapy doesn't get along with recent matplotlib stuff
p.tight_layout()
if save:
p.savefig(self.fileroot + '_sc' + str(self.scan) + 'sp' + str(self.spw[0]) + 'i' + str(self.startints[0]) + '_tier1_' + str(figi) + '.png')
figi += 1
def do_search(self):
""" Bispectrum transient search over all delay beams.
"""
self.masterloc = {} # "location" of candidate: dt, integration (over all data in pipe), dmbin
self.masterprop = {} # properties of candidate: snr and std of bispectra for dedispersed trial
self.masterdata = {} # data around the candidate: bisp lightcurve and spectrogram
self.masternoise = {} # noise per bl as measured by detect_bispectra
for (ra,dec) in self.delaycenters:
self.masterloc[(ra,dec)] = []
self.masterprop[(ra,dec)] = []
self.masterdata[(ra,dec)] = []
self.masternoise[(ra,dec)] = []
for startint in self.startints:
if self.datatype == 'mir':
self.obs = tpipe.pipe_mirdisp2(file=self.datafile, nints=self.nints, nskip=startint, profile='default', dmarr=self.dmarr, chans=self.chans)
elif self.datatype == 'ms':
self.obs = tpipe.pipe_msdisp2(file=self.datafile, nints=self.nints, nskip=startint, profile='default', selectpol=self.selectpol, scan=self.scan, spw=self.spw, chans=self.chans, dmarr=self.dmarr, datacol='corrected_data')
elif self.datatype == 'sim':
self.obs = tpipe.pipe_simdisp2(array='vla_a', nints=self.nints, profile='default', dmarr=self.dmarr, chans=range(64))
# self.obs.add_transient(0., 0., 1., n.random.randint(self.nints))
else:
print 'Unknown datatype!'
return
# accumulate running reltime
if startint == self.startints[0]:
self.tarr = self.obs.reltime
else:
self.tarr = n.append(self.tarr, self.obs.reltime + self.tarr[-1] + self.obs.inttime)
for dtind in range(len(self.timescales)):
# after first pass, need to reset the working data so effect of time_filter is as expected
# if self.timescales.index(dt) > 0:
if dtind > 0:
print 'Resetting data to unfiltered version...'
self.obs.prep()
self.obs.time_filter(self.timescales[dtind], self.filtershape, bgwindow=self.bgwindow)
for (ra,dec) in self.delaycenters:
print
print 'Searching int %d, dt %d, and delay beam (ra, dec) = %.3f, %.3f' % (startint, self.timescales[dtind], ra, dec)
l1 = n.radians(ra); m1 = n.radians(dec)
if ( (l1 != self.obs.l0[0]) | (m1 != self.obs.m0[0]) ):
dl = l1 - self.obs.l0[0]; dm = m1 - self.obs.m0[0] # find phase shift for new location
self.obs.phaseshift(dl, dm)
# measure single noise for input to detect_bispectra
noiseperbl = self.obs.data.mean(axis=3).mean(axis=2).real.std()
# find candidates
self.obs.make_bispectra(stokes=self.stokes)
# add a way to cut candidates near edges (due to time_filter)?
(candsnr, candstd, cands) = self.obs.detect_bispectra(sigma=self.sigma, Q=noiseperbl, save=0)
# build master list of candidate location from bispectrum search
loclist = self.masterloc[(ra,dec)]
proplist = self.masterprop[(ra,dec)]
datalist = self.masterdata[(ra,dec)]
for i in range(len(cands)):
loclist.append( [dtind, (startint-self.startints[0])+cands[i][0], cands[i][1]] ) # candidate location
proplist.append( [candsnr[i], candstd[i]] ) # candidate bisp properties
intmin = max(0, cands[i][0] - 64) # candidate lightcuve and spectrogram
intmax = min(cands[i][0] + 64, self.nints)
bamean = self.obs.bispectra.real.mean(axis=2)[cands[i][1], intmin:intmax] # trim in time to save space
bastd = self.obs.bispectra.real.std(axis=2)[cands[i][1], intmin:intmax]
spec = self.obs.dataph[intmin:intmax]
datalist.append( (bamean, bastd, spec) )
self.masterloc[(ra,dec)] = loclist
self.masterprop[(ra,dec)] = proplist
self.masterdata[(ra,dec)] = datalist
self.masternoise[(ra,dec)] = (dtind, self.obs.Q)
print '%d candidates found in beam of %s at int %d' % (len(loclist), self.datafile, startint)
