def test_residual_image_bag(self):
context = 'wstack'
vis_slices = {'2d': None, 'timeslice': 'auto', 'wstack': 101}
dirty_bag = invert_bag(self.vis_bag,
self.model,
dopsf=False,
context=context,
normalize=True,
vis_slices=vis_slices[context])
psf_bag = invert_bag(self.vis_bag,
self.model,
dopsf=True,
context=context,
normalize=True,
vis_slices=vis_slices[context])
model_bag = deconvolve_bag(dirty_bag,
psf_bag,
niter=1000,
gain=0.1,
algorithm='msclean',
threshold=0.01,
window_shape=None)
residual_bag = residual_image_bag(self.vis_bag,
model_bag,
context=context,
vis_slices=vis_slices[context])
final = list(residual_bag)[0][0]
export_image_to_fits(
final,
'%s/test_bag_%s_residual.fits' % (self.results_dir, context))
qa = qa_image(final, context=context)
assert qa.data['max'] < 15.0, str(qa)
def test_deconvolve_bag(self):
context = 'wstack'
vis_slices = {'2d': None, 'timeslice': 'auto', 'wstack': 101}
dirty_bag = invert_bag(self.vis_bag,
self.model_bag,
dopsf=False,
context=context,
normalize=True,
vis_slices=vis_slices[context])
psf_bag = invert_bag(self.vis_bag,
self.model_bag,
dopsf=True,
context=context,
normalize=True,
vis_slices=vis_slices[context])
dirty_bag = reify(dirty_bag)
psf_bag = reify(psf_bag)
model_bag = deconvolve_bag(dirty_bag,
psf_bag,
self.empty_model_bag,
niter=1000,
gain=0.7,
algorithm='msclean',
threshold=0.01,
window_shape=None)
model = model_bag.compute()[0]['image']
qa = qa_image(model, context=context)
export_image_to_fits(
model, '%s/test_bags_%s_deconvolve.fits' % (self.dir, context))
assert numpy.abs(qa.data['max'] - 60.5) < 0.1, str(qa)
def test_deconvolve_bag(self):
context = 'wstack'
vis_slices = {'2d': None, 'timeslice': 'auto', 'wstack': 101}
dirty_bag = invert_bag(self.vis_bag,
self.model,
dopsf=False,
context=context,
normalize=True,
vis_slices=vis_slices[context])
psf_bag = invert_bag(self.vis_bag,
self.model,
dopsf=True,
context=context,
normalize=True,
vis_slices=vis_slices[context])
model_bag = deconvolve_bag(dirty_bag,
psf_bag,
niter=1000,
gain=0.7,
algorithm='msclean',
threshold=0.01,
window_shape=None)
model = list(model_bag)[0]
qa = qa_image(model, context=context)
export_image_to_fits(
model,
'%s/test_bag_%s_deconvolve.fits' % (self.results_dir, context))
assert numpy.abs(qa.data['max'] - 60.5293158864) < 1.0e-7, str(qa)
def test_residual_image_bag(self):
context = 'wstack'
vis_slices = {'wstack': 101}
dirty_bag = invert_bag(self.vis_bag,
self.empty_model_bag,
dopsf=False,
context=context,
normalize=True,
vis_slices=vis_slices[context])
psf_bag = invert_bag(self.vis_bag,
self.empty_model_bag,
dopsf=True,
context=context,
normalize=True,
vis_slices=vis_slices[context])
dirty_bag = reify(dirty_bag)
psf_bag = reify(psf_bag)
model_bag = deconvolve_bag(dirty_bag,
psf_bag,
self.empty_model_bag,
niter=1000,
gain=0.1,
algorithm='msclean',
threshold=0.01,
window_shape=None)
residual_bag = residual_image_bag(self.vis_bag,
model_bag,
context=context,
vis_slices=vis_slices[context])
final = residual_bag.compute()[0]['image'][0]
export_image_to_fits(
final, '%s/test_bags_%s_residual.fits' % (self.dir, context))
qa = qa_image(final, context=context)
assert qa.data['max'] < 15.0, str(qa)
def test_predict_bag(self):
errors = {'2d': 28.0, 'timeslice': 30.0, 'wstack': 2.3}
vis_slices = {'2d': None, 'timeslice': 'auto', 'wstack': 101}
for context in ['wstack', 'timeslice']:
model_vis_bag = predict_bag(self.vis_bag,
self.model_bag,
context,
vis_slices=vis_slices[context])
model_vis_bag = reify(model_vis_bag)
error_vis_bag = self.vis_bag.map(predict_record_subtract,
model_vis_bag)
error_vis_bag = reify(error_vis_bag)
error_image_bag = invert_bag(error_vis_bag,
self.model_bag,
dopsf=False,
context=context,
normalize=True,
vis_slices=vis_slices[context])
result = error_image_bag.compute()
error_image = result[0]['image'][0]
export_image_to_fits(
error_image, '%s/test_bags_%s_predict_error_image.fits' %
(self.dir, context))
qa = qa_image(error_image, context='error image for %s' % context)
assert qa.data['max'] < errors[context], str(qa)
def test_restore_bag(self):
peaks = {'wstack': 98.8113067286}
vis_slices = {'wstack': 101}
context = 'wstack'
dirty_bag = invert_bag(self.vis_bag,
self.model_bag,
dopsf=False,
context=context,
normalize=True,
vis_slices=vis_slices[context])
psf_bag = invert_bag(self.vis_bag,
self.model_bag,
dopsf=True,
context=context,
normalize=True,
vis_slices=vis_slices[context])
dirty_bag = reify(dirty_bag)
psf_bag = reify(psf_bag)
model_bag = deconvolve_bag(dirty_bag,
psf_bag,
self.empty_model_bag,
niter=1000,
gain=0.7,
algorithm='msclean',
threshold=0.01,
window_shape=None)
model_bag = reify(model_bag)
res_image_bag = residual_image_bag(self.vis_bag,
model_bag,
context=context,
vis_slices=vis_slices[context])
res_image_bag = reify(res_image_bag)
residual = res_image_bag.compute()[0]['image'][0]
export_image_to_fits(
residual, '%s/test_bags_%s_residual.fits' % (self.dir, context))
final_bag = restore_bag(model_bag, psf_bag, res_image_bag)
final = final_bag.compute()[0]
qa = qa_image(final, context=context)
export_image_to_fits(
final, '%s/test_bags_%s_restored.fits' % (self.dir, context))
assert numpy.abs(qa.data['max'] - peaks[context]) < 0.1, str(qa)
def test_invert_bag(self):
peaks = {'2d': 65.2997439062, 'timeslice': 99.6183393299, 'wstack': 100.702701119}
vis_slices = {'2d': None, 'timeslice': 'auto', 'wstack': 101}
for context in ['wstack', '2d', 'timeslice']:
dirty_bag = invert_bag(self.vis_bag, self.empty_model, dopsf=False,
context=context, normalize=True,
vis_slices=vis_slices[context])
dirty_bag.visualize('test_invert_bag.svg')
result=dirty_bag.compute()
dirty, sumwt = list(dirty_bag)[0]
export_image_to_fits(dirty, '%s/test_bag_%s_dirty.fits' % (self.results_dir, context))
qa = qa_image(dirty, context=context)
assert numpy.abs(qa.data['max'] - peaks[context]) < 1.0e-2, str(qa)
def test_restore_bag(self):
cvis_bag = self.vis_bag.map(coalesce_visibility)
peaks = {'timeslice': 103.55068395, 'wstack': 98.790503433}
vis_slices = {'2d': None, 'timeslice': 'auto', 'wstack': 101}
for context in ['timeslice', 'wstack']:
dirty_bag = invert_bag(self.vis_bag, self.model_bag, dopsf=False,
context=context, normalize=True,
vis_slices=vis_slices[context])
psf_bag = invert_bag(self.vis_bag, self.model_bag, dopsf=True,
context=context, normalize=True,
vis_slices=vis_slices[context])
model_bag = deconvolve_bag(dirty_bag, psf_bag, self.empty_model_bag, niter=1000, gain=0.7, algorithm='msclean',
threshold=0.01, window_shape=None)
residual_bag = residual_image_bag(cvis_bag, model_bag, context=context, vis_slices=vis_slices[context])
residual = list(residual_bag)[0][0]
export_image_to_fits(residual, '%s/test_bag_%s_residual.fits' % (self.results_dir, context))
final_bag = restore_bag(model_bag, psf_bag, residual_bag)
final = list(final_bag)[0]
qa = qa_image(final, context=context)
export_image_to_fits(final, '%s/test_bag_%s_restored.fits' % (self.results_dir, context))
assert numpy.abs(qa.data['max'] - peaks[context]) < 1.0e-2, str(qa)
def test_predict_bag(self):
cvis_bag = self.vis_bag.map(coalesce_visibility)
errors = {'2d': 28.0, 'timeslice': 30.0, 'wstack': 2.3}
vis_slices = {'2d': None, 'timeslice': 'auto', 'wstack': 101}
for context in ['timeslice', 'wstack']:
model_vis_bag = predict_bag(cvis_bag, self.model_bag, context, vis_slices=vis_slices[context])\
.map(decoalesce_visibility, self.vis_bag)
model_vis_bag = bag.from_sequence(model_vis_bag.compute())
error_vis_bag = self.vis_bag.map(subtract_visibility, model_vis_bag).compute()
error_vis_bag = bag.from_sequence(error_vis_bag)
error_image_bag = invert_bag(error_vis_bag, self.model_bag, dopsf=False,
context=context, normalize=True, vis_slices=vis_slices[context])
result = list(error_image_bag)[0]
error_image = result[0]
export_image_to_fits(error_image, '%s/test_bag_%s_predict_error_image.fits' % (self.results_dir, context))
qa = qa_image(error_image, context='error image for %s' % context)
assert qa.data['max'] < errors[context], str(qa)
def test_invert_bag(self):
peaks = {
'2d': 65.2997439062,
'timeslice': 99.6183393299,
'wstack': 101.099173809
}
vis_slices = {'2d': None, 'timeslice': 'auto', 'wstack': 101}
for context in ['2d', 'timeslice', 'wstack']:
dirty_bag = invert_bag(self.vis_bag,
self.empty_model_bag,
dopsf=False,
context=context,
normalize=True,
vis_slices=vis_slices[context])
dirty, sumwt = dirty_bag.compute()[0]['image']
export_image_to_fits(
dirty, '%s/test_bags_%s_dirty.fits' % (self.dir, context))
qa = qa_image(dirty, context=context)
assert numpy.abs(qa.data['max'] - peaks[context]) < 1.0e-2, str(qa)
def test_predict_bag(self):
errors = {'2d': 28.0, 'timeslice': 31.0, 'wstack': 2.3}
vis_slices = {'2d': None, 'timeslice': 'auto', 'wstack': 101}
for context in ['2d', 'timeslice', 'wstack']:
model_vis_bag = predict_bag(self.vis_bag,
self.model,
context,
vis_slices=vis_slices[context])
error_vis_bag = residual_vis_bag(self.vis_bag, model_vis_bag)
error_image_bag = invert_bag(error_vis_bag,
self.model,
dopsf=False,
context=context,
normalize=True,
vis_slices=vis_slices[context])
result = list(error_image_bag)[0]
error_image = result[0]
export_image_to_fits(
error_image, '%s/test_bag_%s_predict_error_image.fits' %
(self.results_dir, context))
qa = qa_image(error_image, context='error image for %s' % context)
assert qa.data['max'] < errors[context], str(qa)
def test_invert_bag(self):
peaks = {
'2d': 65.440798589,
'timeslice': 99.7403479215,
'wstack': 100.654001673
}
vis_slices = {'2d': None, 'timeslice': 'auto', 'wstack': 101}
model = copy_image(self.model)
for context in ['wstack', '2d', 'timeslice']:
dirty_bag = invert_bag(self.vis_bag,
model,
dopsf=False,
context=context,
normalize=True,
vis_slices=vis_slices[context])
dirty, sumwt = list(dirty_bag)[0]
export_image_to_fits(
dirty,
'%s/test_bag_%s_dirty.fits' % (self.results_dir, context))
qa = qa_image(dirty, context=context)
assert numpy.abs(qa.data['max'] - peaks[context]) < 1.0e-7, str(qa)
def test_selfcal_global_bag(self):
self.setupVis(add_errors=True)
selfcal_vis_bag = selfcal_bag(self.vis_bag,
self.model_bag,
global_solution=True,
context='wstack',
vis_slices=51)
dirty_bag = invert_bag(selfcal_vis_bag,
self.model_bag,
dopsf=False,
normalize=True,
context='wstack',
vis_slices=101)
if self.compute:
dirty, sumwt = dirty_bag.compute()[0]['image']
export_image_to_fits(
dirty,
'%s/test_imaging_bags_global_selfcal_dirty.fits' % (self.dir))
qa = qa_image(dirty)
assert numpy.abs(qa.data['max'] - 101.7) < 1.0, str(qa)
assert numpy.abs(qa.data['min'] + 3.5) < 1.0, str(qa)
def ical_pipeline_bag(block_vis_bag,
model_bag,
context='2d',
first_selfcal=None,
**kwargs) -> bag:
"""Create bag for ICAL pipeline
:param vis_bag:
:param model_bag:
:param context: Imaging context
:param first_selfcal: First cycle for phase only selfcal
:param kwargs: Parameters for functions in bags
:return:
"""
assert isinstance(block_vis_bag, bag.Bag), block_vis_bag
assert isinstance(model_bag, bag.Bag), model_bag
vis_bag = block_vis_bag.map(map_record,
convert_blockvisibility_to_visibility)
psf_bag = invert_bag(vis_bag,
model_bag,
context=context,
dopsf=True,
**kwargs)
psf_bag = reify(psf_bag)
# Make the predicted visibilities, selfcalibrate against it correcting the gains, then
# form the residual visibility, then make the residual image
model_vis_bag = reify(
predict_bag(vis_bag, model_bag, context=context, **kwargs))
if first_selfcal is not None and first_selfcal == 0:
vis_bag = reify(
selfcal_record(block_vis_bag,
model_vis_bag=model_vis_bag,
**kwargs))
res_vis_bag = reify(residual_vis_bag(vis_bag, model_vis_bag))
res_bag = invert_bag(res_vis_bag,
model_bag,
context=context,
dopsf=False,
**kwargs)
res_bag = reify(res_bag)
deconvolve_model_bag = reify(
deconvolve_bag(res_bag, psf_bag, model_bag, **kwargs))
nmajor = get_parameter(kwargs, "nmajor", 5)
if nmajor > 1:
for cycle in range(nmajor):
model_vis_bag = reify(
predict_bag(vis_bag,
deconvolve_model_bag,
context=context,
**kwargs))
if first_selfcal is not None and cycle >= first_selfcal:
vis_bag = reify(
selfcal_record(block_vis_bag,
model_vis_bag=model_vis_bag,
**kwargs))
res_vis_bag = reify(residual_vis_bag(vis_bag, model_vis_bag))
res_bag = invert_bag(res_vis_bag,
model_bag,
context=context,
dopsf=False,
**kwargs)
res_bag = reify(res_bag)
deconvolve_model_bag = reify(
deconvolve_bag(res_bag, psf_bag, deconvolve_model_bag,
**kwargs))
res_bag = residual_image_bag(vis_bag,
deconvolve_model_bag,
context=context,
**kwargs)
res_bag = reify(res_bag)
deconvolve_model_bag = reify(deconvolve_model_bag)
rest_bag = restore_bag(deconvolve_model_bag, psf_bag, res_bag, **kwargs)
rest_bag = reify(rest_bag)
return bag.from_sequence([deconvolve_model_bag, res_bag, rest_bag])
