def plotnoisedist(noisepkl, plot_width=450, plot_height=400):
""" """
# plot noise and flag distributions
scans, segments, noiseperbl, flagfrac, imstd = read_noise(noisepkl)
pl = Figure(plot_width=plot_width,
plot_height=plot_height,
toolbar_location="above",
x_axis_label='Flag fraction',
y_axis_label='Image noise (sys)',
tools='pan, wheel_zoom, reset')
pl.cross(flagfrac, imstd, line_alpha=0.2, color='blue')
# find medians
flagfrac_med = np.median(flagfrac)
imstd_med = np.median(imstd)
logger.info('Median image noise (sys) = {0}'.format(imstd_med))
logger.info('Median flag fraction = {0}'.format(flagfrac_med))
pl.cross(flagfrac_med, imstd_med, line_alpha=1, size=40, color='red')
# estimate number of zero noise images
zeronoisefrac = float(len(np.where(imstd == 0.)[0])) / len(imstd)
logger.info('{0:.0%} of noise images are zeros'.format(zeronoisefrac))
return pl, imstd_med, flagfrac_med
def plotnoisecum(noisepkl, fluxscale=1, plot_width=450, plot_height=400):
""" Merged noise pkl converted to interactive cumulative histogram
noisepkl is standard noise pickle file.
fluxscale is scaling applied by gain calibrator. telcal solutions have fluxscale=1.
also returns corrected imnoise values if non-unity fluxscale provided
"""
# noise histogram
noises = read_noise(noisepkl)
imnoise = np.sort(fluxscale * noises[4])
frac = [
float(count) / len(imnoise)
for count in reversed(range(1,
len(imnoise) + 1))
]
noiseplot = Figure(
plot_width=plot_width,
plot_height=plot_height,
toolbar_location="above",
x_axis_label='Image noise (Jy; cal scaling {0:.3})'.format(fluxscale),
y_axis_label='Cumulative fraction',
tools='pan, wheel_zoom, reset')
noiseplot.line(imnoise, frac)
if fluxscale != 1:
return noiseplot, imnoise
else:
return noiseplot
def plotnoisedist(noisepkl, plot_width=450, plot_height=400):
""" """
# plot noise and flag distributions
scans, segments, noiseperbl, flagfrac, imstd = read_noise(noisepkl)
pl = Figure(plot_width=plot_width, plot_height=plot_height, toolbar_location="above",
x_axis_label='Flag fraction', y_axis_label='Image noise (sys)', tools='pan, wheel_zoom, reset')
pl.cross(flagfrac, imstd, line_alpha=0.2, color='blue')
# find medians
flagfrac_med = np.median(flagfrac)
imstd_med = np.median(imstd)
logger.info('Median image noise (sys) = {0}'.format(imstd_med))
logger.info('Median flag fraction = {0}'.format(flagfrac_med))
pl.cross(flagfrac_med, imstd_med, line_alpha=1, size=40, color='red')
# estimate number of zero noise images
zeronoisefrac = float(len(np.where(imstd == 0.)[0]))/len(imstd)
logger.info('{0:.0%} of noise images are zeros'.format(zeronoisefrac))
return pl, imstd_med, flagfrac_med
def plotnoisecum(noisepkl, fluxscale=1, plot_width=450, plot_height=400):
""" Merged noise pkl converted to interactive cumulative histogram
noisepkl is standard noise pickle file.
fluxscale is scaling applied by gain calibrator. telcal solutions have fluxscale=1.
also returns corrected imnoise values if non-unity fluxscale provided
"""
# noise histogram
noises = read_noise(noisepkl)
imnoise = np.sort(fluxscale*noises[4])
frac = [float(count)/len(imnoise) for count in reversed(range(1, len(imnoise)+1))]
noiseplot = Figure(plot_width=plot_width, plot_height=plot_height, toolbar_location="above",
x_axis_label='Image noise (Jy; cal scaling {0:.3})'.format(fluxscale),
y_axis_label='Cumulative fraction', tools='pan, wheel_zoom, reset')
noiseplot.line(imnoise, frac)
if fluxscale != 1:
return noiseplot, imnoise
else:
return noiseplot
