def __init__(self, **kwargs):
self._run_parameters = kwargs
filename = kwargs['filename']
self._filename = filename
scrunch = kwargs.get('scrunch', 1)
self._scrunch = scrunch # For identifying process.
self._data_ring = MockRing(filename, scrunch)
#self._data_ring = power_data_io.MockRing(filename, scrunch)
#parameters = get_parameters(filename)
#print parameters
parameters = self._data_ring.get_parameters()
self._parameters = parameters
####
if kwargs.get('to_2_DM_diag', False):
freq0 = parameters['freq0']
delta_f = parameters['delta_f']
nfreq = parameters['nfreq']
freq_max = max(freq0, freq0 + delta_f * (nfreq - 1))
freq_min = min(freq0, freq0 + delta_f * (nfreq - 1))
dm_diag = misc.diag_dm(parameters['delta_t'], delta_f, freq_max)
self._max_dm = 0.9 * 2 * dm_diag
self._overlap = 1.5 * (misc.disp_delay(2 * dm_diag, freq_min) -
misc.disp_delay(2 * dm_diag, freq_max))
self._time_block = 3.0 * self._overlap
else:
self._time_block = kwargs.get('time_block', TIME_BLOCK)
self._overlap = kwargs.get('overlap', OVERLAP)
self._max_dm = kwargs.get('max_dm', MAX_DM)
#print self._overlap, self._time_block, dm_diag
self._min_search_dm = kwargs.get('min_search_dm', MIN_SEARCH_DM)
self._Transformer = dedisperse.DMTransform(
parameters['delta_t'],
parameters['nfreq'],
parameters['freq0'],
parameters['delta_f'],
self._max_dm,
jon=USE_JON_DD,
)
print(
"Filename: %s, delta_t: %f, min_dm: %f, max_dm: %f, ndm: %d,"
" time_block: %f" %
(filename, parameters['delta_t'], self._min_search_dm,
self._max_dm, self._Transformer.ndm, self._time_block))
#self._df = parameters['delta_f']
#self._nfreq = parameters['nfreq']
#self._f0 = parameters['freq0']
nrecords_block = int(
math.ceil(self._time_block /
(parameters['ntime_record'] * parameters['delta_t'])))
#initialize sim object, if there are to be simulated events
if SIMULATE:
self._sim_source = simulate.RandSource(
alpha=alpha,
f_m=f_m,
f_sd=f_sd,
bw_m=bw_m,
bw_sd=bw_sd,
t_m=t_m,
t_sd=t_sd,
s_m=s_m,
s_sd=s_sd,
dm_m=dm_m,
dm_sd=dm_sd,
event_rate=sim_rate,
file_params=self._parameters,
t_overlap=self._overlap,
nrecords_block=nrecords_block)
self._cal_spec = 1.
self._dedispersed_out_group = None
self.set_search_method()
self.set_trigger_action()
def __init__(self, **kwargs):
self._run_parameters = kwargs
filename = kwargs['filename']
self._filename = filename
scrunch = kwargs.get('scrunch', 1)
self._scrunch = scrunch # For identifying process.
self._data_ring = MockRing(filename, scrunch)
#self._data_ring = power_data_io.MockRing(filename, scrunch)
#parameters = get_parameters(filename)
#print parameters
parameters = self._data_ring.get_parameters()
self._parameters = parameters
####
if kwargs.get('to_2_DM_diag', False):
freq0 = parameters['freq0']
delta_f = parameters['delta_f']
nfreq = parameters['nfreq']
freq_max = max(freq0, freq0 + delta_f * (nfreq - 1))
freq_min = min(freq0, freq0 + delta_f * (nfreq - 1))
dm_diag = misc.diag_dm(parameters['delta_t'], delta_f, freq_max)
self._max_dm = 0.9 * 2 * dm_diag
self._overlap = 1.5 * (misc.disp_delay(2 * dm_diag, freq_min)
- misc.disp_delay(2 * dm_diag, freq_max))
self._time_block = 3.0 * self._overlap
else:
self._time_block = kwargs.get('time_block', TIME_BLOCK)
self._overlap = kwargs.get('overlap', OVERLAP)
self._max_dm = kwargs.get('max_dm', MAX_DM)
#print self._overlap, self._time_block, dm_diag
self._min_search_dm = kwargs.get('min_search_dm', MIN_SEARCH_DM)
self._Transformer = dedisperse.DMTransform(
parameters['delta_t'],
parameters['nfreq'],
parameters['freq0'],
parameters['delta_f'],
self._max_dm,
jon=USE_JON_DD,
)
print ("Filename: %s, delta_t: %f, min_dm: %f, max_dm: %f, ndm: %d,"
" time_block: %f"
% (filename, parameters['delta_t'], self._min_search_dm,
self._max_dm, self._Transformer.ndm, self._time_block))
#self._df = parameters['delta_f']
#self._nfreq = parameters['nfreq']
#self._f0 = parameters['freq0']
nrecords_block = int(math.ceil(self._time_block / (parameters['ntime_record'] * parameters['delta_t'])))
#initialize sim object, if there are to be simulated events
if SIMULATE:
self._sim_source = simulate.RandSource(alpha=alpha, f_m=f_m,f_sd=f_sd,bw_m=bw_m,bw_sd=bw_sd,t_m=t_m,
t_sd=t_sd,s_m=s_m,s_sd=s_sd,dm_m=dm_m,dm_sd=dm_sd,
event_rate=sim_rate,file_params=self._parameters,t_overlap=self._overlap,nrecords_block=nrecords_block)
self._cal_spec = 1.
self._dedispersed_out_group = None
self.set_search_method()
self.set_trigger_action()
class FileSearch(object):
def __init__(self, **kwargs):
self._run_parameters = kwargs
filename = kwargs['filename']
self._filename = filename
scrunch = kwargs.get('scrunch', 1)
self._scrunch = scrunch # For identifying process.
self._data_ring = MockRing(filename, scrunch)
#self._data_ring = power_data_io.MockRing(filename, scrunch)
#parameters = get_parameters(filename)
#print parameters
parameters = self._data_ring.get_parameters()
self._parameters = parameters
####
if kwargs.get('to_2_DM_diag', False):
freq0 = parameters['freq0']
delta_f = parameters['delta_f']
nfreq = parameters['nfreq']
freq_max = max(freq0, freq0 + delta_f * (nfreq - 1))
freq_min = min(freq0, freq0 + delta_f * (nfreq - 1))
dm_diag = misc.diag_dm(parameters['delta_t'], delta_f, freq_max)
self._max_dm = 0.9 * 2 * dm_diag
self._overlap = 1.5 * (misc.disp_delay(2 * dm_diag, freq_min) -
misc.disp_delay(2 * dm_diag, freq_max))
self._time_block = 3.0 * self._overlap
else:
self._time_block = kwargs.get('time_block', TIME_BLOCK)
self._overlap = kwargs.get('overlap', OVERLAP)
self._max_dm = kwargs.get('max_dm', MAX_DM)
#print self._overlap, self._time_block, dm_diag
self._min_search_dm = kwargs.get('min_search_dm', MIN_SEARCH_DM)
self._Transformer = dedisperse.DMTransform(
parameters['delta_t'],
parameters['nfreq'],
parameters['freq0'],
parameters['delta_f'],
self._max_dm,
jon=USE_JON_DD,
)
print(
"Filename: %s, delta_t: %f, min_dm: %f, max_dm: %f, ndm: %d,"
" time_block: %f" %
(filename, parameters['delta_t'], self._min_search_dm,
self._max_dm, self._Transformer.ndm, self._time_block))
#self._df = parameters['delta_f']
#self._nfreq = parameters['nfreq']
#self._f0 = parameters['freq0']
nrecords_block = int(
math.ceil(self._time_block /
(parameters['ntime_record'] * parameters['delta_t'])))
#initialize sim object, if there are to be simulated events
if SIMULATE:
self._sim_source = simulate.RandSource(
alpha=alpha,
f_m=f_m,
f_sd=f_sd,
bw_m=bw_m,
bw_sd=bw_sd,
t_m=t_m,
t_sd=t_sd,
s_m=s_m,
s_sd=s_sd,
dm_m=dm_m,
dm_sd=dm_sd,
event_rate=sim_rate,
file_params=self._parameters,
t_overlap=self._overlap,
nrecords_block=nrecords_block)
self._cal_spec = 1.
self._dedispersed_out_group = None
self.set_search_method()
self.set_trigger_action()
def set_cal_spectrum(self, cal_spec):
"""Set spectrum of the noise-cal for band-pass calibration.
Parameters
----------
cal_spec : 1D record array with records 'freq' and 'cal_T'.
"""
nfreq = self._parameters['nfreq']
# TODO Should really check that the frequency axes match exactly, or
# interpolate/extrapolate.
if len(cal_spec) != nfreq:
msg = "Noise cal spectrum frequncy axis does not match the data."
raise ValueError(msg)
spec = cal_spec["cal_T"]
spec[np.logical_not(np.isfinite(spec))] = 0
self._cal_spec = spec
def set_search_method(self, method='basic', **kwargs):
if method == 'basic':
self._search = lambda dm_data: search.basic(
dm_data,
THRESH_SNR,
self._min_search_dm,
int(HPF_WIDTH / 2 / self._parameters['delta_t']),
)
else:
msg = "Unrecognized search method."
raise ValueError(msg)
def set_trigger_action(self, action='print', **kwargs):
actions = [
self._get_trigger_action(s.strip()) for s in action.split(',')
]
def action_fun(triggers):
for a in actions:
a(triggers)
self._action = action_fun
def _get_trigger_action(self, action):
if action == 'print':
def action_fun(triggers):
print("%d:" % self._scrunch), triggers
return action_fun
self._action = action_fun
elif action == 'show_plot_dm':
def action_fun(triggers):
for t in triggers:
plt.figure()
t.plot_dm()
plt.show()
return action_fun
elif action == 'show_plot_time':
def action_fun(triggers):
for t in triggers:
plt.figure()
t.plot_time()
plt.show()
return action_fun
elif action == 'save_plot_dm':
def action_fun(triggers):
for t in triggers:
parameters = self._parameters
t_offset = t.time
f = plt.figure(1)
plt.subplot(411)
t.plot_dm()
plt.subplot(412)
t.plot_freq()
plt.subplot(413)
t.plot_time()
plt.subplot(414)
t.plot_spec()
t_dm_value = t.centre[0] * t.data.delta_dm
if t_dm_value < 5:
out_filename = "DM0-5_"
elif 5 <= t_dm_value < 20:
out_filename = "DM5-20_"
elif 20 <= t_dm_value < 100:
out_filename = "DM20-100_"
elif 100 <= t_dm_value < 300:
out_filename = "DM100-300_"
else:
out_filename = "DM300-2000_"
out_filename += path.splitext(path.basename(
self._filename))[0]
if not t.data.spec_ind is None:
out_filename += "+a=%02.f" % t.data.spec_ind
out_filename += "+%06.2fs.png" % t_offset
plt.savefig(out_filename, bbox_inches='tight')
plt.close(f)
return action_fun
elif action == 'link_squirrel':
def action_fun(triggers):
for t in triggers:
t.link_baseband(SQUIRREL)
return action_fun
elif action == 'link_chime':
def action_fun(triggers):
for t in triggers:
t.link_baseband(CHIME)
return action_fun
else:
msg = "Unrecognized trigger action: " + action
raise ValueError(msg)
def set_dedispersed_h5(self, group=None):
"""Set h5py group to which to write dedispersed data."""
self._dedispersed_out_group = group
#simple method to replace nested structure
def search_records(self, start_record, end_record):
data, time = self.get_records(start_record, end_record)
parameters = self._parameters
if DEV_PLOTS:
plt.figure()
imshow_data(data)
preprocess.sys_temperature_bandpass(data)
if DEV_PLOTS:
plt.figure()
imshow_data(data)
if self._parameters['cal_period_samples']:
preprocess.remove_periodic(data,
self._parameters['cal_period_samples'])
nrecords_block = int(
math.ceil(self._time_block /
(parameters['ntime_record'] * parameters['delta_t'])))
block_ind = start_record / nrecords_block
# Preprocess.
#preprocess.sys_temperature_bandpass(data)
if SIMULATE and block_ind in self._sim_source.coarse_event_schedule():
#do simulation
data += self._sim_source.generate_events(
block_ind)[:, 0:data.shape[1]]
preprocess.remove_outliers(data, 5, 128)
data = preprocess.highpass_filter(data,
HPF_WIDTH / parameters['delta_t'])
preprocess.remove_outliers(data, 5)
preprocess.remove_noisy_freq(data, 2)
preprocess.remove_bad_times(data, 2)
preprocess.remove_continuum_v2(data)
preprocess.remove_noisy_freq(data, 2)
# Readjust time axis.
times_lost = len(time) - data.shape[-1]
time = time[times_lost // 2:times_lost // 2 + data.shape[-1]]
if DEV_PLOTS:
plt.figure()
imshow_data(data[:, :2000])
plt.show()
#from here we weight channels by spectral index
#center_f = self._f0 + (self._df*self._nfreq/2.0)
#fmin = self._f0 + self._df*self._nfreq
#fmax = self._f0
nfreq = self._parameters['nfreq']
delta_f = self._parameters['delta_f']
freq0 = self._parameters['freq0']
freq = np.arange(nfreq) * delta_f + freq0
if DO_SPEC_SEARCH:
#print "----------------------"
spec_trigger = None
complete = 1
for alpha in np.linspace(SPEC_INDEX_MIN, SPEC_INDEX_MAX,
SPEC_INDEX_SAMPLES):
#for i in xrange(0,3):
#weights = array([math.pow(f/center_f, alpha) for f in np.linspace(fmax,fmin,self._nfreq)])
#f = lambda x: weights*x
#this_dat = np.matrix(np.apply_along_axis(f, axis=0, arr=data),dtype=np.float32)
spec_weights = ((freq / freq0)**alpha).astype(data.dtype)
this_dat = data * spec_weights[:, None]
dm_data = self._Transformer(this_dat)
dm_data.spec_ind = alpha
dm_data.t0 = time[0]
del this_dat
these_triggers = self._search(dm_data)
del dm_data
#print 'complete indices: {0} of {1} ({2})'.format(complete,SPEC_INDEX_SAMPLES,alpha)
if len(these_triggers) > 0:
print("%d: alpha %2f, SNR %4.1f" %
(self._scrunch, alpha, these_triggers[0].snr))
if spec_trigger == None or these_triggers[
0].snr > spec_trigger.snr:
spec_trigger = these_triggers[0]
del these_triggers
complete += 1
#spec_triggers = [t[0] for t in spec_triggers if len(t) > 0]
#spec_triggers = sorted(spec_triggers, key= lambda x: -x.snr)
#if len(spec_triggers) > 0:
#spec_triggers = [spec_triggers[0],]
#self._action(spec_triggers, dm_data)
if spec_trigger != None:
self._action((spec_trigger, ))
else:
dm_data = self._Transformer(data)
dm_data.t0 = time[0]
triggers = self._search(dm_data)
self._action(triggers)
del triggers
def get_records(self, start_record, end_record):
#data = read_records(self._filename, start_record, end_record)
data = self._data_ring.read_records(start_record, end_record)
return data
def search_all_records(self):
time_block = self._time_block
overlap = self._overlap
parameters = self._parameters
record_length = (parameters['ntime_record'] * parameters['delta_t'])
nrecords_block = int(math.ceil(time_block / record_length))
nrecords_overlap = int(math.ceil(overlap / record_length))
nrecords = self._data_ring.current_records()[1]
for ii in xrange(0, nrecords, nrecords_block - nrecords_overlap):
# XXX
print "Block starting with record: {0} of {1}".format(ii, nrecords)
#print "Progress: {0}".format(float(ii)/float(nrecords))
self.search_records(ii, ii + nrecords_block)
def search_real_time(self):
parameters = self._parameters
time_block = self._time_block
overlap = self._overlap
record_length = (parameters['ntime_record'] * parameters['delta_t'])
nrecords_block = int(math.ceil(time_block / record_length))
nrecords_overlap = int(math.ceil(overlap / record_length))
wait_time = float(time_block - overlap) / 5
max_wait_iterations = 10
# Enter holding loop, processing records in blocks as they become
# available.
current_start_record = 0
wait_iterations = 0
while wait_iterations < max_wait_iterations:
nrecords = self._data_ring.current_records()[1]
if nrecords - current_start_record >= nrecords_block:
print("%d: Block starting with record: %d of %d" %
(self._scrunch, current_start_record, nrecords))
try:
self.search_records(current_start_record,
current_start_record + nrecords_block)
except power_data_io.DataGone:
print "%d: MISSED A BLOCK" % self._scrunch
current_start_record += nrecords_block - nrecords_overlap
wait_iterations = 0
else:
time.sleep(wait_time)
wait_iterations += 1
# Precess any leftovers that don't fill out a whole block.
self.search_records(current_start_record,
current_start_record + nrecords_block)
class FileSearch(object):
def __init__(self, **kwargs):
self._run_parameters = kwargs
filename = kwargs['filename']
self._filename = filename
scrunch = kwargs.get('scrunch', 1)
self._scrunch = scrunch # For identifying process.
self._data_ring = MockRing(filename, scrunch)
#self._data_ring = power_data_io.MockRing(filename, scrunch)
#parameters = get_parameters(filename)
#print parameters
parameters = self._data_ring.get_parameters()
self._parameters = parameters
####
if kwargs.get('to_2_DM_diag', False):
freq0 = parameters['freq0']
delta_f = parameters['delta_f']
nfreq = parameters['nfreq']
freq_max = max(freq0, freq0 + delta_f * (nfreq - 1))
freq_min = min(freq0, freq0 + delta_f * (nfreq - 1))
dm_diag = misc.diag_dm(parameters['delta_t'], delta_f, freq_max)
self._max_dm = 0.9 * 2 * dm_diag
self._overlap = 1.5 * (misc.disp_delay(2 * dm_diag, freq_min)
- misc.disp_delay(2 * dm_diag, freq_max))
self._time_block = 3.0 * self._overlap
else:
self._time_block = kwargs.get('time_block', TIME_BLOCK)
self._overlap = kwargs.get('overlap', OVERLAP)
self._max_dm = kwargs.get('max_dm', MAX_DM)
#print self._overlap, self._time_block, dm_diag
self._min_search_dm = kwargs.get('min_search_dm', MIN_SEARCH_DM)
self._Transformer = dedisperse.DMTransform(
parameters['delta_t'],
parameters['nfreq'],
parameters['freq0'],
parameters['delta_f'],
self._max_dm,
jon=USE_JON_DD,
)
print ("Filename: %s, delta_t: %f, min_dm: %f, max_dm: %f, ndm: %d,"
" time_block: %f"
% (filename, parameters['delta_t'], self._min_search_dm,
self._max_dm, self._Transformer.ndm, self._time_block))
#self._df = parameters['delta_f']
#self._nfreq = parameters['nfreq']
#self._f0 = parameters['freq0']
nrecords_block = int(math.ceil(self._time_block / (parameters['ntime_record'] * parameters['delta_t'])))
#initialize sim object, if there are to be simulated events
if SIMULATE:
self._sim_source = simulate.RandSource(alpha=alpha, f_m=f_m,f_sd=f_sd,bw_m=bw_m,bw_sd=bw_sd,t_m=t_m,
t_sd=t_sd,s_m=s_m,s_sd=s_sd,dm_m=dm_m,dm_sd=dm_sd,
event_rate=sim_rate,file_params=self._parameters,t_overlap=self._overlap,nrecords_block=nrecords_block)
self._cal_spec = 1.
self._dedispersed_out_group = None
self.set_search_method()
self.set_trigger_action()
def set_cal_spectrum(self, cal_spec):
"""Set spectrum of the noise-cal for band-pass calibration.
Parameters
----------
cal_spec : 1D record array with records 'freq' and 'cal_T'.
"""
nfreq = self._parameters['nfreq']
# TODO Should really check that the frequency axes match exactly, or
# interpolate/extrapolate.
if len(cal_spec) != nfreq:
msg = "Noise cal spectrum frequncy axis does not match the data."
raise ValueError(msg)
spec = cal_spec["cal_T"]
spec[np.logical_not(np.isfinite(spec))] = 0
self._cal_spec = spec
def set_search_method(self, method='basic', **kwargs):
if method == 'basic':
self._search = lambda dm_data : search.basic(
dm_data,
THRESH_SNR,
self._min_search_dm,
int(HPF_WIDTH / 2 / self._parameters['delta_t']),
)
else:
msg = "Unrecognized search method."
raise ValueError(msg)
def set_trigger_action(self, action='print', **kwargs):
actions = [self._get_trigger_action(s.strip()) for s in action.split(',')]
def action_fun(triggers):
for a in actions:
a(triggers)
self._action = action_fun
def _get_trigger_action(self,action):
if action == 'print':
def action_fun(triggers):
print ("%d:" % self._scrunch), triggers
return action_fun
self._action = action_fun
elif action == 'show_plot_dm':
def action_fun(triggers):
for t in triggers:
plt.figure()
t.plot_dm()
plt.show()
return action_fun
elif action == 'show_plot_time':
def action_fun(triggers):
for t in triggers:
plt.figure()
t.plot_time()
plt.show()
return action_fun
elif action == 'save_plot_dm':
def action_fun(triggers):
for t in triggers:
parameters = self._parameters
t_offset = t.time
f = plt.figure(1)
plt.subplot(411)
t.plot_dm()
plt.subplot(412)
t.plot_freq()
plt.subplot(413)
t.plot_time()
plt.subplot(414)
t.plot_spec()
t_dm_value = t.centre[0] * t.data.delta_dm
if t_dm_value < 5:
out_filename = "DM0-5_"
elif 5 <= t_dm_value < 20:
out_filename = "DM5-20_"
elif 20 <= t_dm_value < 100:
out_filename = "DM20-100_"
elif 100 <= t_dm_value <300:
out_filename = "DM100-300_"
else:
out_filename = "DM300-2000_"
out_filename += path.splitext(path.basename(self._filename))[0]
if not t.data.spec_ind is None:
out_filename += "+a=%02.f" % t.data.spec_ind
out_filename += "+%06.2fs.png" % t_offset
plt.savefig(out_filename, bbox_inches='tight')
plt.close(f)
return action_fun
elif action == 'link_squirrel':
def action_fun(triggers):
for t in triggers:
t.link_baseband(SQUIRREL)
return action_fun
elif action == 'link_chime':
def action_fun(triggers):
for t in triggers:
t.link_baseband(CHIME)
return action_fun
else:
msg = "Unrecognized trigger action: " + action
raise ValueError(msg)
def set_dedispersed_h5(self, group=None):
"""Set h5py group to which to write dedispersed data."""
self._dedispersed_out_group = group
#simple method to replace nested structure
def search_records(self, start_record, end_record):
data, time = self.get_records(start_record, end_record)
parameters = self._parameters
if DEV_PLOTS:
plt.figure()
imshow_data(data)
preprocess.sys_temperature_bandpass(data)
if DEV_PLOTS:
plt.figure()
imshow_data(data)
if self._parameters['cal_period_samples']:
preprocess.remove_periodic(data,
self._parameters['cal_period_samples'])
nrecords_block = int(math.ceil(
self._time_block / (parameters['ntime_record'] * parameters['delta_t'])))
block_ind = start_record/nrecords_block
# Preprocess.
#preprocess.sys_temperature_bandpass(data)
if SIMULATE and block_ind in self._sim_source.coarse_event_schedule():
#do simulation
data += self._sim_source.generate_events(block_ind)[:,0:data.shape[1]]
preprocess.remove_outliers(data, 5, 128)
data = preprocess.highpass_filter(data, HPF_WIDTH / parameters['delta_t'])
preprocess.remove_outliers(data, 5)
preprocess.remove_noisy_freq(data, 2)
preprocess.remove_bad_times(data, 2)
preprocess.remove_continuum_v2(data)
preprocess.remove_noisy_freq(data, 2)
# Readjust time axis.
times_lost = len(time) - data.shape[-1]
time = time[times_lost // 2: times_lost // 2 + data.shape[-1]]
if DEV_PLOTS:
plt.figure()
imshow_data(data[:,:2000])
plt.show()
#from here we weight channels by spectral index
#center_f = self._f0 + (self._df*self._nfreq/2.0)
#fmin = self._f0 + self._df*self._nfreq
#fmax = self._f0
nfreq = self._parameters['nfreq']
delta_f = self._parameters['delta_f']
freq0 = self._parameters['freq0']
freq = np.arange(nfreq) * delta_f + freq0
if DO_SPEC_SEARCH:
#print "----------------------"
spec_trigger = None
complete = 1
for alpha in np.linspace(SPEC_INDEX_MIN,SPEC_INDEX_MAX,SPEC_INDEX_SAMPLES):
#for i in xrange(0,3):
#weights = array([math.pow(f/center_f, alpha) for f in np.linspace(fmax,fmin,self._nfreq)])
#f = lambda x: weights*x
#this_dat = np.matrix(np.apply_along_axis(f, axis=0, arr=data),dtype=np.float32)
spec_weights = ((freq / freq0)**alpha).astype(data.dtype)
this_dat = data * spec_weights[:,None]
dm_data = self._Transformer(this_dat)
dm_data.spec_ind = alpha
dm_data.t0 = time[0]
del this_dat
these_triggers = self._search(dm_data)
del dm_data
#print 'complete indices: {0} of {1} ({2})'.format(complete,SPEC_INDEX_SAMPLES,alpha)
if len(these_triggers) > 0:
print ("%d: alpha %2f, SNR %4.1f" % (self._scrunch,
alpha, these_triggers[0].snr))
if spec_trigger == None or these_triggers[0].snr > spec_trigger.snr:
spec_trigger = these_triggers[0]
del these_triggers
complete += 1
#spec_triggers = [t[0] for t in spec_triggers if len(t) > 0]
#spec_triggers = sorted(spec_triggers, key= lambda x: -x.snr)
#if len(spec_triggers) > 0:
#spec_triggers = [spec_triggers[0],]
#self._action(spec_triggers, dm_data)
if spec_trigger != None:
self._action((spec_trigger,))
else:
dm_data = self._Transformer(data)
dm_data.t0 = time[0]
triggers = self._search(dm_data)
self._action(triggers)
del triggers
def get_records(self, start_record, end_record):
#data = read_records(self._filename, start_record, end_record)
data = self._data_ring.read_records(start_record, end_record)
return data
def search_all_records(self):
time_block = self._time_block
overlap = self._overlap
parameters = self._parameters
record_length = (parameters['ntime_record'] * parameters['delta_t'])
nrecords_block = int(math.ceil(time_block / record_length))
nrecords_overlap = int(math.ceil(overlap / record_length))
nrecords = self._data_ring.current_records()[1]
for ii in xrange(0, nrecords, nrecords_block - nrecords_overlap):
# XXX
print "Block starting with record: {0} of {1}".format(ii,nrecords)
#print "Progress: {0}".format(float(ii)/float(nrecords))
self.search_records(ii, ii + nrecords_block)
def search_real_time(self):
parameters = self._parameters
time_block = self._time_block
overlap = self._overlap
record_length = (parameters['ntime_record'] * parameters['delta_t'])
nrecords_block = int(math.ceil(time_block / record_length))
nrecords_overlap = int(math.ceil(overlap / record_length))
wait_time = float(time_block - overlap) / 5
max_wait_iterations = 10
# Enter holding loop, processing records in blocks as they become
# available.
current_start_record = 0
wait_iterations = 0
while wait_iterations < max_wait_iterations:
nrecords = self._data_ring.current_records()[1]
if nrecords - current_start_record >= nrecords_block:
print ("%d: Block starting with record: %d of %d"
% (self._scrunch, current_start_record, nrecords))
try:
self.search_records(current_start_record,
current_start_record + nrecords_block)
except power_data_io.DataGone:
print "%d: MISSED A BLOCK" % self._scrunch
current_start_record += nrecords_block - nrecords_overlap
wait_iterations = 0
else:
time.sleep(wait_time)
wait_iterations += 1
# Precess any leftovers that don't fill out a whole block.
self.search_records(current_start_record,
current_start_record + nrecords_block)
