ddcal¶
Perform direction-dependent calibration on the data (SHARED-RISK DEVELOPMENT MODE).
enable¶
bool
Execute the ddcal worker (i.e. carry out DD-calibration).
label_in¶
str, optional, default = corr
Label of the .MS files to process. By default uses the ‘corr’ label_in for self-calibrated dataset.
use_pb¶
bool, optional, default = False
Enable primary beam usage in making the DD-corrected DDFacet image. Note that this is EXPERIMENTAL and currently only available for MeerKAT data.
image_dd¶
Imaging parameters for DD calibration with DDFacet.
enable
bool, optional, default = True
Enable the ‘image_dd’ segment.
npix
int, optional, default = 8000
Number of pixels in the image. Note that DDFacet has its own super-special scheme to decide the actual number of pixels, so this is only an approximation.
use_mask
bool, optional, default = True
Enable clean mask for DDFacet initial imaging. Note that this doubles the imaging time since it runs DDFacet twice – once to get a preliminary image to make a mask with (mask is made by the cleanmask tool), and once to get the final image with masking. Previous WSClean masks cannot be used because pixel numbers might be different.
mask_sigma
float, optional, default = 10.0
The number of standard deviations (i.e. sigma_rms) to use when clipping the initial image for masking.
mask_boxes
int, optional, default = 9
Divide the initial image (for making the mask) into this number of boxes, then perform sigma clipping in each of these boxes.
mask_niter
int, optional, default = 20
The number of sigma-clipping iterations to perform on the image, for masking, or set to 0 to clip until convergence is achieved.
mask_overlap
float, optional, default = 0.3
Overlap region for the boxes, given as a fraction of the number of boxes.
mask_tol
float, optional, default = 0.75
Tolerance for dilating the mask. Dilation will stop when the percentage difference between dilations is smaller than this value.
cell
float, optional, default = 1.3
Pixel size in arcsec.
facets_nfacets
int, optional, default = 24
Number of facets to use, and is the same as the Facets-NFacets parameter of DDFacet.
weight_col
{“WEIGHT_SPECTRUM”, “WEIGHT”, “IMAGING_WEIGHT”}, optional, default = WEIGHT
Read data weights from the specified column. Options are WEIGHT_SPECTRUM, WEIGHT, and (for rarer occasions) IMAGING_WEIGHT.
weight_mode
{“Natural”, “Uniform”, “Robust”, “Briggs”}, optional, default = Briggs
UV weighting mode. Options are ‘Natural’, ‘Uniform’, ‘Robust’, and ‘Briggs’.
weight_robust
float, optional, default = -0.4
Briggs robustness parameter, from -2 (more uniform) to 2 (more natural).
deconv_maxminoriter
int, optional, default = 100000
Number of clean iterations.
freq_nband
int, optional, default = 10
Number of frequency bands for gridding.
freq_ndegridband
int, optional, default = 15
Number of frequency bands for degridding. 0 means degrid each channel.
deconv_rmsfactor
float, optional, default = 0.0
Set the minor-cycle stopping-threshold to X*{residual RMS}, where X is this parameter value.
deconv_peakfactor
float, optional, default = 0.25
Set the minor-cycle stopping-threshold to X*{peak residual}, where X is this parameter value.
deconv_mode
{“HMP”, “Hogbom”, “SSD”, “GAClean”}, optional, default = Hogbom
The deconvolution algorithm to use. Options are ‘HMP’ (Hybrid Matching Pursuit, aka multiscale/multifrequency), ‘Hogbom’ (Hogbom’s CLEAN algorithm), ‘SSD’ (SubSpace Deconvolution algorithm), and ‘GAClean’ (Genetic Algorithm Clean). Please direct queries to DDFacet Developers for further details.
deconv_gain
float, optional, default = 0.1
Gain setting for the deconvolution loops.
deconv_fluxthr
float, optional, default = 1.0e-6
Absolute flux-density threshold at which deconvolution is stopped, in units of Jy. Relevant for HMP and Hogbom modes.
deconv_allownegative
bool, optional, default = True
Allow negative components for cleaning (valid for HMP and Hogbom modes).
hogbom_polyfitorder
int, optional, default = 6
Order of the polynomial to be used for frequency fitting.
parallel_ncpu
int, optional, default = 0
Number of processes / threads to use in parallel mode. 0 = use all of those available. 1 = disable parallelism.
predict_colname
str, optional, default = MODEL_DATA
MS column to write the predicted visibilities corresponding to the model. Setting ‘’ will disable this parameter.
log_memory
bool, optional, default = True
Log the memory usage by DDFacet.
cache_reset
bool, optional, default = True
Reset all caches (including PSF and dirty image). Change from default at your own risk.
log_boring
bool, optional, default = True
Enable progress bars and other pretty console output. Doesn’t seem to work. But who knows, try it out.
data_colname
str, optional, default = CORRECTED_DATA
Data column to use for initial imaging. Defaults to ‘CORRECTED_DATA’, the assumption being that self-calibration has already been done on the measurement set.
data_colname_postcal
str, optional, default = SUBDD_DATA
Data column to use for imaging after dd-calibration. Defaults to ‘SUBDD_DATA’, so as to not overwrite the corrected data. If data size increase is a concern, switch to ‘CORRECTED_DATA’
data_chunkhours
float, optional, default = 0.05
Chunk data into time bins of X hours to conserve memory, where X is this parameter.
output_mode
{“Dirty”, “Clean”, “Predict”, “PSF”}, optional, default = Clean
Output mode of DDFacet. Options are ‘Dirty’, ‘Clean’, ‘Predict’, and ‘PSF’. This setting defaults to ‘Clean’, since that is what we want to do in this worker.
calibrate_dd¶
Direction-dependent calibration parameters.
enable
bool, optional, default = True
Enable the ‘calibrate_dd’ segment.
sigma
float, optional, default = 4.5
Sigma threshold to use in detecting outlier regions in images, via CATDagger (which is enabled by setting ‘de_sources_mode’, below, to ‘auto’). The default value of 4.5 works well, but a lower value may be needed for some images.
min_dist_from_phcentre
int, optional, default = 1300
The radius (in number of pixels), from the centre of the image, out to which sources will not be tagged for DD-calibration. (This is because sources close to the phase centre may not have been cleaned deeply enough.) The default is kept at 1300 (which roughly corresponds to 30’).
dist_ncpu
int, optional, default = 1
The number of cpus for distributed computing.
de_sources_mode
str, optional, default = manual
Mode in which sources are tagged for DD calibration. Options are ‘auto’ (which uses CATDagger), and ‘manual’ (for which one needs to provide a list of sources). Use ‘auto’ with caution and at your own risk.
de_target_manual
list of str, optional, default = ‘ ‘
List of target fieldnames for carrying out DD calibration. The remaining fields will not undergo DD calibration.
de_sources_manual
list of str, optional, default = ‘ ‘
List of sources per target to tag for DD calibration, in the same order as the ‘de_target_manual’ list. Use ‘;’ to separate different sources per target.
sol_min_bl
float, optional, default = 100
The minimum baseline length to solve for.
madmax_enable
bool, optional, default = true
Enable madmax flagging in CubiCal.
madmax_thr
list of int, optional, default = 0, 10
Threshold for MAD flagging per baseline (specified in number of standard deviations). Residuals exceeding mad-thr*MAD/1.428 will be flagged. MAD is computed per baseline. This can be specified as a list e.g. N1,N2,N3,… The first value is used to flag residuals before a solution starts (use 0 to disable), the next value is used when the residuals are first recomputed during the solution several iterations later (see -chi-int), etc. A final pass may be done at the end of the solution. The last value in the list is reused if necessary. Using a list with gradually-decreasing values may be sensible.
madmax_global_thr
list of int, optional, default = 0, 12
Threshold for global median MAD (MMAD) flagging. MMAD is computed as the median of the per-baseline MADs. Residuals exceeding S*MMAD/1.428 will be flagged.
madmax_estimate
{“corr”, “all”, “diag”, “offdiag”}, optional, default = corr
MAD estimation mode. Use ‘corr’ for a separate estimate per baseline and correlation. Otherwise, a single estimate per baseline is computed using ‘all’ correlations, or only the ‘diag’ or ‘offdiag’ correlations.
dd_data_col
str, optional, default = CORRECTED_DATA
Column to calibrate, with the assumption that you have already run the selfcal worker.
dd_out_data_col
str, optional, default = SUBDD_DATA
Output data column. Note that the ddcal worker is currently hardcoded for this being set to ‘SUBDD_DATA’.
dd_weight_col
str, optional, default = WEIGHT
Column to read weights from, and apply them by default. Specify an empty string to disable this parameter.
dd_sol_stall_quorum
float, optional, default = 0.95
Minimum percentage of solutions that must have stalled before terminating the solver.
dd_g_type
str, optional, default = complex-2x2
Gain matrix type for the G-Jones matrix. Keep this set to ‘complex-2x2’, because DD-calibration fails otherwise.
dd_g_clip_high
float, optional, default = 1.5
Amplitude clipping – flag solutions with any amplitudes above this value for G-Jones matrix.
dd_g_clip_low
float, optional, default = 0.5
Amplitude clipping – flag solutions with any amplitudes below this value for G-Jones matrix.
dd_g_update_type
str, optional, default = phase-diag
Determines update type. This does not change the Jones solver type, but restricts the update rule to pin the solutions within a certain subspace.
dd_g_max_prior_error
float, optional, default = 0.35
Flag solution intervals where the prior error estimate is above this value for G-Jones matrix.
dd_g_max_post_error
float, optional, default = 0.35
Flag solution intervals where the posterior variance estimate is above this value for G-Jones matrix.
dd_dd_max_prior_error
float, optional, default = 0.35
Flag solution intervals where the prior error estimate is above this value for DE term.
dd_dd_max_post_error
float, optional, default = 0.35
Flag solution intervals where the posterior variance estimate is above this value for DE term.
dd_g_timeslots_int
int, optional, default = 10
Time solution interval in timeslot units for G-Jones matrix.
dd_g_chan_int
int, optional, default = 0
Frequency solution interval in channel units for G-Jones matrix.
dd_dd_timeslots_int
int, optional, default = 100
Time solution interval in timeslot units for DE-Jones matrix.
dd_dd_chan_int
int, optional, default = 100
Frequency solution interval in channel units for DE-Jones matrix.
dist_nworker
int, optional, default = 0
Number of processes.
copy_data¶
Copy DD-calibrated data to CORRECTED_DATA column. THIS IS DANGEROUS - only if you want to go ahead with line imaging.
enable
bool, optional, default = True
Enable copying of DD-calibrated data to CORRECTED_DATA column.
image_wsclean¶
WSClean imaging paramaters for ddcal worker.
enable
bool, optional, default = True
Enable WSClean imaging of the DD-calibrated data.
img_ws_npix
int, optional, default = 1800
Number of pixels in output image.
img_ws_padding
float, optional, default = 1.3
Padding in WSClean.
img_ws_mgain
float, optional, default = 0.90
Gain for the major cycle during image CLEANing.
img_ws_cell
float, optional, default = 2.
Image pixel size (in arcsec).
img_ws_weight
{“briggs”, “uniform”, “natural”}, optional, default = briggs
Image weighting type. Options are ‘briggs’, ‘uniform’, and ‘natural’. If ‘briggs’, set the img_ws_robust parameter below.
img_ws_robust
float, optional, default = 0.
Briggs robust value.
img_ws_uvtaper
str, optional, default = 0
Taper for imaging (in arcsec).
img_ws_niter
int, optional, default = 1000000
Number of cleaning iterations.
img_ws_nmiter
int, optional, default = 0
Number of major cycles.
img_ws_cleanborder
float, optional, default = 1.3
Clean border.
img_ws_nchans
int, optional, default = 3
Number of channels in output image.
img_ws_joinchans
bool, optional, default = True
Join channels to create MFS image.
img_ws_specfit_nrcoeff
int, optional, default = 2
Number of spectral polynomial terms to fit to each clean component. This is equal to the order of the polynomial plus 1.
img_ws_stokes
{“I”}, optional, default = I
Stokes image to create. For this first release of CARACal, the only option is ‘I’.
img_ws_auto_mask
float, optional, default = 7
Auto-masking threshold, given as the number of sigma_rms.
img_ws_auto_thr
float, optional, default = 0.5
Auto-clean threshold, given as the number of sigma_rms.
img_ws_col
str, optional, default = CORRECTED_DATA
Column to image.
img_ws_fits_mask
str, optional, default = catalog_mask.fits
Filename of fits mask (in output/masking folder).
img_ws_multi_scale
bool, optional, default = False
Switch on multiscale cleaning.
img_ws_multi_scale_scales
list of int, optional, default = 0, 5, 10, 20
Scales for multiscale cleaning, in pixels.
img_ws_local_rms
bool, optional, default = False
Switch on local-rms measurement for cleaning.
transfer_model_dd¶
Repredict WSClean model to the highest channel resolution.
enable
bool, optional, default = False
Enable the ‘transfer_model_dd’ segment.
dd_model
str, optional, default = auto
Name of the sky-model file. (Currently the only supported format is that of WSClean component lists.) When set to ‘auto’, the pipeline builds the file name from the input parameters of the selfcal loop. The file is assumed to be in the ‘output’ directory.
dd_row_chunks
int, optional, default = 0
Number of rows of input .MS that are processed in a single chunk.
dd_model_chunks
int, optional, default = 0
Number of sky-model components that are processed in a single chunk.
dd_within
str, optional, default = ‘ ‘
Give JS9 region file. Only sources within those regions will be included.
dd_points_only
bool, optional, default = False
Select only ‘point’ sources.
dd_num_sources
int, optional, default = 0
Select only N brightest sources.
dd_num_workers
int, optional, default = 0
Explicitly set the number of worker threads. Default is 0, meaning it uses all threads.
dd_mem_frac
float, optional, default = 0.5
Fraction of system RAM that can be used. Used when setting automatically the chunk size.
report¶
bool, optional, default = False
(Re)generate a full HTML report at the end of this worker.
cabs¶
list of map, optional, default = ‘ ‘
Specifies non-default image versions and/or tags for Stimela cabs. Running with scissors: use with extreme caution.