for v in future_bvis_list
]
psf_list = rsexecute.compute(psf_list, sync=True)
future_psf_list = rsexecute.scatter(psf_list)
del psf_list
if use_natural:
print("Using natural weighting")
else:
print("Using uniform weighting")
vis_list = [
rsexecute.execute(convert_blockvisibility_to_visibility)(bvis)
for bvis in future_bvis_list
]
vis_list = weight_list_rsexecute_workflow(vis_list, future_psf_list)
bvis_list = [
rsexecute.execute(convert_visibility_to_blockvisibility)(vis)
for vis in vis_list
]
bvis_list = rsexecute.compute(bvis_list, sync=True)
future_bvis_list = rsexecute.scatter(bvis_list)
del bvis_list
dopsf = False
if dopsf:
print("Calculating PSF")
imaging_context = args.imaging_context
if imaging_context == "ng":
psf_list = invert_list_rsexecute_workflow(future_bvis_list,
future_psf_list,
def trial_case(results,
seed=180555,
context='wstack',
nworkers=8,
threads_per_worker=1,
memory=8,
processes=True,
order='frequency',
nfreqwin=7,
ntimes=3,
rmax=750.0,
facets=1,
wprojection_planes=1,
use_dask=True,
use_serial_imaging=True,
flux_limit=0.3,
nmajor=5,
dft_threshold=1.0,
use_serial_clean=True,
write_fits=False):
""" Single trial for performance-timings
Simulates visibilities from GLEAM including phase errors
Makes dirty image and PSF
Runs ICAL pipeline
The results are in a dictionary:
'context': input - a string describing concisely the purpose of the test
'time overall', overall execution time (s)
'time predict', time to execute GLEAM prediction graph
'time invert', time to make dirty image
'time invert graph', time to make dirty image graph
'time ICAL graph', time to create ICAL graph
'time ICAL', time to execute ICAL graph
'context', type of imaging e.g. 'wstack'
'nworkers', number of workers to create
'threads_per_worker',
'nnodes', Number of nodes,
'processes', 'order', Ordering of data_models
'nfreqwin', Number of frequency windows in simulation
'ntimes', Number of hour angles in simulation
'rmax', Maximum radius of stations used in simulation (m)
'facets', Number of facets in deconvolution and imaging
'wprojection_planes', Number of wprojection planes
'vis_slices', Number of visibility slices (per Visibbility)
'npixel', Number of pixels in image
'cellsize', Cellsize in radians
'seed', Random number seed
'dirty_max', Maximum in dirty image
'dirty_min', Minimum in dirty image
'restored_max',
'restored_min',
'deconvolved_max',
'deconvolved_min',
'residual_max',
'residual_min',
'git_info', GIT hash (not definitive since local mods are possible)
:param results: Initial state
:param seed: Random number seed (used in gain simulations)
:param context: imaging context
:param context: Type of context: '2d'|'timeslice'|'wstack'
:param nworkers: Number of dask workers to use
:param threads_per_worker: Number of threads per worker
:param processes: Use processes instead of threads 'processes'|'threads'
:param order: See simulate_list_list_rsexecute_workflow_workflowkflow
:param nfreqwin: See simulate_list_list_rsexecute_workflow_workflowkflow
:param ntimes: See simulate_list_list_rsexecute_workflow_workflowkflow
:param rmax: See simulate_list_list_rsexecute_workflow_workflowkflow
:param facets: Number of facets to use
:param wprojection_planes: Number of wprojection planes to use
:param use_dask: Use dask or immediate evaluation
:return: results dictionary
"""
if use_dask:
scheduler = os.getenv('RASCIL_DASK_SCHEDULER', None)
if scheduler is not None:
client = get_dask_client(n_workers=nworkers,
memory_limit=memory * 1024 * 1024 * 1024,
threads_per_worker=threads_per_worker)
rsexecute.set_client(client=client)
else:
rsexecute.set_client(threads_per_worker=threads_per_worker,
processes=threads_per_worker == 1,
memory_limit=memory * 1024 * 1024 * 1024,
n_workers=nworkers)
print("Defined %d workers" % (nworkers))
else:
rsexecute.set_client(use_dask=use_dask)
results['nnodes'] = 1
def init_logging():
logging.basicConfig(
filename='pipelines_rsexecute_timings.log',
filemode='w',
format='%(asctime)s,%(msecs)d %(name)s %(levelname)s %(message)s',
datefmt='%H:%M:%S',
level=logging.INFO)
init_logging()
log = logging.getLogger()
# Initialise logging on the workers. This appears to only work using the process scheduler.
rsexecute.run(init_logging)
def lprint(*args):
log.info(*args)
print(*args)
lprint("Starting pipelines_rsexecute_timings")
numpy.random.seed(seed)
results['seed'] = seed
start_all = time.time()
results['context'] = context
results['hostname'] = socket.gethostname()
results['git_hash'] = git_hash()
results['epoch'] = time.strftime("%Y-%m-%d %H:%M:%S")
lprint("Context is %s" % context)
results['nworkers'] = nworkers
results['threads_per_worker'] = threads_per_worker
results['processes'] = processes
results['memory'] = memory
results['order'] = order
results['nfreqwin'] = nfreqwin
results['ntimes'] = ntimes
results['rmax'] = rmax
results['facets'] = facets
results['wprojection_planes'] = wprojection_planes
results['dft threshold'] = dft_threshold
results['use_dask'] = use_dask
lprint("At start, configuration is:")
lprint(sort_dict(results))
# Parameters determining scale of simulation.
frequency = numpy.linspace(1.0e8, 1.2e8, nfreqwin)
centre = nfreqwin // 2
if nfreqwin > 1:
channel_bandwidth = numpy.array(nfreqwin *
[frequency[1] - frequency[0]])
else:
channel_bandwidth = numpy.array([1e6])
times = numpy.linspace(-numpy.pi / 4.0, numpy.pi / 4.0, ntimes)
phasecentre = SkyCoord(ra=+0.0 * u.deg,
dec=-40.0 * u.deg,
frame='icrs',
equinox='J2000')
lprint("****** Visibility creation ******")
# Create the empty BlockVisibility's and persist these on the cluster
tmp_bvis_list = simulate_list_rsexecute_workflow(
'LOWBD2',
frequency=frequency,
channel_bandwidth=channel_bandwidth,
times=times,
phasecentre=phasecentre,
order=order,
format='blockvis',
rmax=rmax)
tmp_vis_list = [
rsexecute.execute(convert_blockvisibility_to_visibility)(bv)
for bv in tmp_bvis_list
]
tmp_vis_list = rsexecute.client.compute(tmp_vis_list, sync=True)
vis_list = rsexecute.gather(tmp_vis_list)
import matplotlib.pyplot as plt
plt.clf()
plt.hist(vis_list[0].w, bins=100)
plt.title('Histogram of w samples: rms=%.1f (wavelengths)' %
numpy.std(vis_list[0].w))
plt.xlabel('W (wavelengths)')
#plt.show()
plt.clf()
plt.hist(vis_list[0].uvdist, bins=100)
plt.title('Histogram of uvdistance samples')
plt.xlabel('UV Distance (wavelengths)')
#plt.show()
rsexecute.client.cancel(tmp_vis_list)
future_vis_list = rsexecute.scatter(vis_list)
# Find the best imaging parameters but don't bring the vis_list back here
print("****** Finding wide field parameters ******")
future_advice = [
rsexecute.execute(advise_wide_field)(
v,
guard_band_image=6.0,
delA=0.1,
facets=facets,
wprojection_planes=wprojection_planes,
oversampling_synthesised_beam=4.0) for v in future_vis_list
]
future_advice = rsexecute.compute(future_advice)
advice = rsexecute.client.gather(future_advice)[-1]
rsexecute.client.cancel(future_advice)
# Deconvolution via sub-images requires 2^n
npixel = advice['npixels2']
results['npixel'] = npixel
cellsize = advice['cellsize']
results['cellsize'] = cellsize
lprint("Image will have %d by %d pixels, cellsize = %.6f rad" %
(npixel, npixel, cellsize))
# Create an empty model image
tmp_model_list = [
rsexecute.execute(create_image)(
npixel=npixel,
cellsize=cellsize,
frequency=[frequency[f]],
channel_bandwidth=[channel_bandwidth[f]],
phasecentre=phasecentre,
polarisation_frame=PolarisationFrame("stokesI"))
for f, freq in enumerate(frequency)
]
model_list = rsexecute.compute(tmp_model_list, sync=True)
future_model_list = rsexecute.scatter(model_list)
lprint("****** Setting up imaging parameters ******")
# Now set up the imaging parameters
template_model = create_image(
npixel=npixel,
cellsize=cellsize,
frequency=[frequency[centre]],
phasecentre=phasecentre,
channel_bandwidth=[channel_bandwidth[centre]],
polarisation_frame=PolarisationFrame("stokesI"))
gcfcf = [create_pswf_convolutionfunction(template_model)]
if context == 'timeslice':
vis_slices = ntimes
lprint("Using timeslice with %d slices" % vis_slices)
elif context == '2d':
vis_slices = 1
elif context == "wprojection":
wstep = advice['wstep']
nw = advice['wprojection_planes']
vis_slices = 1
support = advice['nwpixels']
results['wprojection_planes'] = nw
lprint("****** Starting W projection kernel creation ******")
lprint("Using wprojection with %d planes with wstep %.1f wavelengths" %
(nw, wstep))
lprint("Support of wprojection = %d pixels" % support)
gcfcf = [
create_awterm_convolutionfunction(template_model,
nw=nw,
wstep=wstep,
oversampling=4,
support=support,
use_aaf=True)
]
lprint("Size of W projection gcf, cf = %.2E bytes" % get_size(gcfcf))
else:
context = 'wstack'
vis_slices = advice['vis_slices']
lprint("Using wstack with %d slices" % vis_slices)
gcfcf = rsexecute.scatter(gcfcf, broadcast=True)
results['vis_slices'] = vis_slices
# Make a skymodel from gleam, with bright sources as components and weak sources in an image
lprint("****** Starting GLEAM skymodel creation ******")
future_skymodel_list = [
rsexecute.execute(create_low_test_skymodel_from_gleam)(
npixel=npixel,
cellsize=cellsize,
frequency=[frequency[f]],
phasecentre=phasecentre,
polarisation_frame=PolarisationFrame("stokesI"),
flux_limit=flux_limit,
flux_threshold=dft_threshold,
flux_max=5.0) for f, freq in enumerate(frequency)
]
# We use predict_skymodel so that we can use skycomponents as well as images
lprint("****** Starting GLEAM skymodel prediction ******")
predicted_vis_list = [
predict_skymodel_list_rsexecute_workflow(future_vis_list[f],
[future_skymodel_list[f]],
context=context,
vis_slices=vis_slices,
facets=facets,
gcfcf=gcfcf)[0]
for f, freq in enumerate(frequency)
]
# Corrupt the visibility for the GLEAM model
lprint("****** Visibility corruption ******")
tmp_corrupted_vis_list = corrupt_list_rsexecute_workflow(
predicted_vis_list, phase_error=1.0, seed=seed)
lprint("****** Weighting and tapering ******")
tmp_corrupted_vis_list = weight_list_rsexecute_workflow(
tmp_corrupted_vis_list, future_model_list)
tmp_corrupted_vis_list = taper_list_rsexecute_workflow(
tmp_corrupted_vis_list, 0.003 * 750.0 / rmax)
tmp_corrupted_vis_list = rsexecute.compute(tmp_corrupted_vis_list,
sync=True)
corrupted_vis_list = rsexecute.gather(tmp_corrupted_vis_list)
# rsexecute.client.cancel(predicted_vis_list)
rsexecute.client.cancel(tmp_corrupted_vis_list)
future_corrupted_vis_list = rsexecute.scatter(corrupted_vis_list)
# At this point the only futures are of scatter'ed data so no repeated calculations should be
# incurred.
lprint("****** Starting dirty image calculation ******")
start = time.time()
dirty_list = invert_list_rsexecute_workflow(
future_corrupted_vis_list,
future_model_list,
vis_slices=vis_slices,
context=context,
facets=facets,
use_serial_invert=use_serial_imaging,
gcfcf=gcfcf)
results['size invert graph'] = get_size(dirty_list)
lprint('Size of dirty graph is %.3E bytes' %
(results['size invert graph']))
end = time.time()
results['time invert graph'] = end - start
lprint("Construction of invert graph took %.3f seconds" % (end - start))
start = time.time()
dirty, sumwt = rsexecute.compute(dirty_list, sync=True)[centre]
end = time.time()
results['time invert'] = end - start
lprint("Dirty image invert took %.3f seconds" % (end - start))
lprint("Maximum in dirty image is %f, sumwt is %s" %
(numpy.max(numpy.abs(dirty.data)), str(sumwt)))
qa = qa_image(dirty)
results['dirty_max'] = qa.data['max']
results['dirty_min'] = qa.data['min']
if write_fits:
export_image_to_fits(
dirty, "pipelines_rsexecute_timings-%s-dirty.fits" % context)
lprint("****** Starting prediction ******")
start = time.time()
tmp_vis_list = predict_list_rsexecute_workflow(
future_corrupted_vis_list,
future_model_list,
vis_slices=vis_slices,
context=context,
facets=facets,
use_serial_predict=use_serial_imaging,
gcfcf=gcfcf)
result = rsexecute.compute(tmp_vis_list, sync=True)
# rsexecute.client.cancel(tmp_vis_list)
end = time.time()
results['time predict'] = end - start
lprint("Predict took %.3f seconds" % (end - start))
# Create the ICAL pipeline to run major cycles, starting selfcal at cycle 1. A global solution across all
# frequencies (i.e. Visibilities) is performed.
print("Using subimage clean")
deconvolve_facets = 8
deconvolve_overlap = 16
deconvolve_taper = 'tukey'
lprint("****** Starting ICAL graph creation ******")
controls = create_calibration_controls()
controls['T']['first_selfcal'] = 1
controls['T']['timeslice'] = 'auto'
start = time.time()
ical_list = ical_list_rsexecute_workflow(
future_corrupted_vis_list,
model_imagelist=future_model_list,
context=context,
vis_slices=vis_slices,
scales=[0, 3, 10],
algorithm='mmclean',
nmoment=3,
niter=1000,
fractional_threshold=0.1,
threshold=0.01,
nmajor=nmajor,
gain=0.25,
psf_support=64,
deconvolve_facets=deconvolve_facets,
deconvolve_overlap=deconvolve_overlap,
deconvolve_taper=deconvolve_taper,
timeslice='auto',
global_solution=True,
do_selfcal=True,
calibration_context='T',
controls=controls,
use_serial_predict=use_serial_imaging,
use_serial_invert=use_serial_imaging,
use_serial_clean=use_serial_clean,
gcfcf=gcfcf)
results['size ICAL graph'] = get_size(ical_list)
lprint('Size of ICAL graph is %.3E bytes' % results['size ICAL graph'])
end = time.time()
results['time ICAL graph'] = end - start
lprint("Construction of ICAL graph took %.3f seconds" % (end - start))
print("Current objects on cluster: ")
pp.pprint(rsexecute.client.who_has())
#
# Execute the graph
lprint("****** Executing ICAL graph ******")
start = time.time()
deconvolved, residual, restored, gaintables = rsexecute.compute(ical_list,
sync=True)
end = time.time()
results['time ICAL'] = end - start
lprint("ICAL graph execution took %.3f seconds" % (end - start))
qa = qa_image(deconvolved[centre])
results['deconvolved_max'] = qa.data['max']
results['deconvolved_min'] = qa.data['min']
deconvolved_cube = image_gather_channels(deconvolved)
if write_fits:
export_image_to_fits(
deconvolved_cube,
"pipelines_rsexecute_timings-%s-ical_deconvolved.fits" % context)
qa = qa_image(residual[centre][0])
results['residual_max'] = qa.data['max']
results['residual_min'] = qa.data['min']
residual_cube = remove_sumwt(residual)
residual_cube = image_gather_channels(residual_cube)
if write_fits:
export_image_to_fits(
residual_cube,
"pipelines_rsexecute_timings-%s-ical_residual.fits" % context)
qa = qa_image(restored[centre])
results['restored_max'] = qa.data['max']
results['restored_min'] = qa.data['min']
restored_cube = image_gather_channels(restored)
if write_fits:
export_image_to_fits(
restored_cube,
"pipelines_rsexecute_timings-%s-ical_restored.fits" % context)
#
rsexecute.close()
end_all = time.time()
results['time overall'] = end_all - start_all
lprint("At end, results are:")
results = sort_dict(results)
lprint(results)
return results
def process(bvis_list, erp_params):
""" Actually process
1. Create template images
2. Optionally weight the data
3. Run the ICAL pipeline (imaging and self-calibration)
:param bvis_list: List of BlockVis to progress in parallel (or graph)
:param erp_params:
:return: List of BlockVis (or graph), results from ical
The relevant fields of the dictionary are::
'process':
{
'cellsize': Image cellsize in radians e.g. 9e-08,
'npixel': Image size in pixels e.g. 256,
'imaging_context': Type of gridder e.g. 2d or ng,
'do_wstacking': Correct for w term in imaging_context='ng' e.g. False,
'algorithm': Clean algorithm, 'hogbom', 'mmclean', 'mfsmsclean',
'fractional_threshold': Fractional of peak to end a major cycle e.g. 0.3,
'gain': Loop gain e.g. 0.1,
'niter': Number of clean interations per major cycle e.g. 1000,
'nmajor': Number of major cycles e.g. 8,
'nmoment': Number of moments for MSMFS algorithm e.g. 2,
'scales': Scales for MSMFS algorithm e.g. [0, 3, 10],
'threshold': Absolute threshold to stop all cleaning e.g. 0.003,
'weighting_algorithm': 'natural' or 'uniform',
'window_shape': 'quarter' or 'no_edge',
'do_selfcal': No self-calibration at the end of each major cycle e.g. True,
'calibration_context': Jones terms to solve for e.g. 'TG',
'global_solution': Is the solution across all frequencies e.g. True,
'T_first_selfcal': First major cycle to perform T selfcalibration e.g. 2,
'T_phase_only': Phase only solution? e.g. True
'T_timeslice': Solution interval 'auto' or time in seconds
'G_first_selfcal': First major cycle to perform G selfcalibration e.g. 5,
'G_phase_only': False,
'G_timeslice': Solution interval 'auto' or time in seconds e.g. 1200,
'B_first_selfcal': First major cycle to perform B selfcalibration e.g. 8,
'B_phase_only': False,
'B_timeslice': Solution interval 'auto' or time in seconds e.g 100000.0,
'tol': Tolerance for gain solution e.g. 1e-8,
'verbose': False,
},
"""
log.info("Creating template images")
model_list = [
rsexecute.execute(create_image_from_visibility, nout=1)(
bvis,
npixel=erp_params["process"]["npixel"],
cellsize=erp_params["process"]['cellsize'],
nchan=1,
frequency=[numpy.mean(bvis.frequency)],
channel_bandwidth=[numpy.sum(bvis.channel_bandwidth)])
for bvis in bvis_list
]
model_list = rsexecute.persist(model_list)
log.info("Applying {} weighting".format(
erp_params['process']['weighting_algorithm']))
bvis_list = weight_list_rsexecute_workflow(
bvis_list,
model_list,
weighting=erp_params['process']['weighting_algorithm'])
bvis_list = rsexecute.persist(bvis_list)
controls = create_calibration_controls()
controls['T']['first_selfcal'] = erp_params["process"]["T_first_selfcal"]
controls['T']['phase_only'] = erp_params["process"]["T_phase_only"]
controls['T']['timeslice'] = erp_params["process"]["T_timeslice"]
controls['G']['first_selfcal'] = erp_params["process"]["G_first_selfcal"]
controls['G']['phase_only'] = erp_params["process"]["G_phase_only"]
controls['G']['timeslice'] = erp_params["process"]["G_timeslice"]
controls['B']['first_selfcal'] = erp_params["process"]["B_first_selfcal"]
controls['B']['phase_only'] = erp_params["process"]["B_phase_only"]
controls['B']['timeslice'] = erp_params["process"]["B_timeslice"]
log.info("Processing with RASCIL ICAL pipeline")
results = ical_list_rsexecute_workflow(
bvis_list,
model_list,
context=erp_params['process']['imaging_context'],
nmajor=erp_params["process"]['nmajor'],
niter=erp_params["process"]['niter'],
algorithm=erp_params["process"]['algorithm'],
nmoment=erp_params["process"]['nmoment'],
scales=erp_params["process"]['scales'],
gain=erp_params["process"]['gain'],
fractional_threshold=erp_params["process"]['fractional_threshold'],
threshold=erp_params["process"]['threshold'],
window_shape=erp_params["process"]['window_shape'],
calibration_context=erp_params["process"]['calibration_context'],
do_selfcal=erp_params["process"]['do_selfcal'],
do_wstacking=erp_params["process"]['do_wstacking'],
global_solution=erp_params["process"]['global_solution'],
tol=erp_params["process"]['tol'],
controls=controls)
return bvis_list, results
cellsize=cellsize)
for v in vis_list
]
else:
model_list = [
rsexecute.execute(import_image_from_fits)(initial_model)
for v in vis_list
]
model = rsexecute.compute(model_list[0], sync=True)
# Perform weighting. This is a collective computation, requiring all visibilities :(
log.info("\nSetup of weighting")
if weighting == 'uniform':
log.info("Will apply uniform weighting")
vis_list = weight_list_rsexecute_workflow(vis_list, model_list)
if context == 'wprojection' or context == 'wprojectwstack':
gcfcf_list = [
rsexecute.execute(create_awterm_convolutionfunction)(
m,
nw=nwplanes,
wstep=wstep,
oversampling=args.oversampling,
support=support,
maxsupport=512) for m in model_list
]
gcfcf_list = rsexecute.persist(gcfcf_list)
gcfcf = rsexecute.compute(gcfcf_list[0], sync=True)
cf = convert_convolutionfunction_to_image(gcfcf[1])
cf.data = numpy.real(cf.data)