Skip to content

Latest commit

 

History

History
393 lines (271 loc) · 10.2 KB

DC_run.md

File metadata and controls

393 lines (271 loc) · 10.2 KB
Raw

explain DC_run (header and step details)

DC-run

automated setup of the header

load modules

Tidy up TempLattices from previous runs

os.system('rm -rf '+pathtoimage + 'TempLattice*')

Switch this off, if you run multiple casa instances/DC_runs in the same work folder ! Else you delete files from another working process

prepare concatfiles

filecheck and preparation on vis

setup general and SD tcleanparameters from DC_pars*

translate mode, inter, mscale into tclean params

naming scheme specific inputs

Translate the abbreviated clean properties used in the user clean setup into real tclean parameters. Define the filename-'infix':

   cleansetup = '.'+ mode +'_'+ specsetup +'_'+ mscale +'_'+ masking +'_'+ inter +'_n'+ str(nit)

suffix for combination method (called via combisetup later)

Here, we define the suffix for each method. The scaling factor iterator will be added in the corresponding combination loop.

intermediate products name for step 1 = gather information

In this section, intermediate file names such as for the axis-reordered SD image, the optional channel-range-cut-out of an SD cube, and regridded SD image are defined.

The chosen mask setup is translated into CASA parameters and the several mask names (threshold based, SD-INT-AM based and combined) are defined and forwarded to the different combination methods. File existence checks initiate the execution of 'step 1', if one of the here defined files does not exist and step 1 is not yet included in the list of execution steps.

The choice of the 'specsetup'-parameter sets the tclean parameters 'start', 'width', and 'nchans' either to the spectral setup of the SD image (specsetup == 'SDpar') or to the user defined spectral setup (specsetup == 'INTpar').

The SDimage used in the combination methods is the axis-reordered-(channel-cut-out-)only-image or the axis-reordered-(channel-cut-out-)regridded-image, respectively. If no threshold is given and a threshold-based mask exists from previous runs, the threshold used therein is retreived from it again, else step 1 is enforced.

combi name list

This section initiates lists in which the names of the combined images are dropped. This can be used for the assessment task to select the wanted products. In case the user wants to redefine the selection within this module, he can use dryrun=True (see above).

suffix generation for step 8

Merge stepnumber the the assesment is based on to add them to the assements file names.

execution steps/methods

step 0: concat (optional)

If you want to combine several interferometric datasets specify 'thevis', 'weightscale', 'concatms' and add 0 to your 'thestep'-list

--- Slides ---

Purpose: combine several interferometric datasets (ALMA 12m and 7m) For this, specify in pars-file:

  • pathtoconcat
  • ‘a12m’
  • ‘a7m’
  • ‘weight12m’,
  • ‘weight7m ‘
  • 'concatms'
  • add 0 to your 'thesteps'-list

what it does:

  • checks CASA version of calibration of the input ms-files
  • warnings for simulated data and calibrations prior CASA 4.3 - user has to correct/adjust weights
  • executes CASA task ‘concat’

run ONCE to create concatenated ms-file, if not yet available

step 1: SD image and tclean/mask preparation

  1. the axes of the SD image are sorted according to the tclean output standard order (reorder_axes).

  2. if in specmode='SDpar' and start and end channel are defined, the SD cube is trimmed to these channels (exclusively; channel_cutout).

  3. the channel setup of the reordered (channel-cut-out0) SD cube is read out (get_SD_cube_params). If wanted, the results are used as inputs for all following tcleans etc.

  4. a dirty image from the concatvis is created and the RMS of the entire image obtained. This is used to define a threshold and from this a clean mask (derive_threshold).

  5. the dirty image is used as template to regrid the reordered SD image (imregrid).

  6. a mask from the reordered-regridded SD image is created and combined with auto-masking results in tclean (make_SDint_mask).

  7. combine theshold and SD mask, because the theshold mask may contain more/different emission peaks than auto-masking (immath).

--- Slides ---

Purpose: put input data to shape (reorder, regrid) desired by combination methods, and prepare masking and thresholds for 'SD-INT-AM' mode For this, specify in pars-file:

  • (pathtoconcat)
  • sdimage_input
  • mode
  • masking
  • general_tclean_param
  • specsetup
  • startchan
  • endchan
  • smoothing
  • RMSfactor
  • cube_rms
  • cont_chans
  • sdmasklev

what it does:

  • reorder_axes: reorder SD image axes to standard
  • channel_cutout: if specmode='SDpar' and start and end channel defined: trim SD cube to these channels
  • get_SD_cube_params: if specmode='SDpar', get SD channel setup and use as inputs for all following tcleans etc.
  • derive_threshold: make dirty image from the concatms, get RMS, use it as threshold and make clean mask from it
  • regrid_SD: regrid reordered SD image to dirty image
  • make_SDint_mask: make mask from the reordered-regridded SD image (and automask-AM)
  • combine threshold and SD(-AM) mask

effects:

  • if general_tclean_param['threshold'] == '':
    • use threshold from derive_threshold
    • retrieve/modify it and its mask each DC_run without new dirty image
  • if masking == 'SD-INT-AM':
    • use threshold(RMS) or SD(-AM) + threshold(RMS) mask

run this step ONCE for a new (spec-)mode and spectral setup (specsetup, startchan, endchan, nchan, width, start) or for a change of the SD mask level full flexibitiy for combination parameters from here on (aka. everything is at hand for playing around)

step 2: tclean only

   imname = imbase + cleansetup + tcleansetup
   e.g. skymodel-b_120L.HB_AM_n0.0e0.tclean

--- Slides ---

Purpose: create (deconvolved) INT image from the concatms data For this, specify in pars-file:

  • mode
  • mscale
  • masking
  • inter
  • nit
  • specsetup
  • general_tclean_param

what it does:

  • runtclean
    • executes CASA task ‘tclean’ and ‘exportfits’

step 3: feather

Requires restoringbeam='common'! perplanebeam-problems

  imname = imbase + cleansetup + feathersetup + str(sdfac[i]) 
  e.g.     skymodel-b_120L.HB_AM_n0.0e0.feather_f1.0

--- Slides ---

Purpose: combine INT image with SD image For this, specify in pars-file:

  • mode
  • mscale
  • masking
  • inter
  • nit
  • specsetup
  • sdfac

what it does:

  • runfeather
    • immath: SD.image * INT.pb
    • feather: SD.image.pbskewed and INT.image
    • immath: pbcorr feathered image
    • exportfits

step 4: SSC

  imname = imbase + cleansetup + SSCsetup + str(SSCfac[i]) 
  e.g.     skymodel-b_120L.HB_AM_n0.0e0.SSC_f1.0

--- Slides ---

Purpose: combine INT image with SD image following S. Faridani (https://bitbucket.org/snippets/faridani/pRX6) For this, specify in pars-file:

  • mode
  • mscale
  • masking
  • inter
  • nit
  • specsetup
  • SSCfac

what it does:

  • ssc
    • immath: SD.image * INT.pb
    • imsmooth: INT.image to SD.beam
    • immath: SD.image.INTpb - INT.image.SDbeam = SD-INT.diff
    • weight = INTbeam/SDbeam
    • immath: INT.image + SD-INT.diff * weight = SSC.image
    • immath: pbcorr SSC image
    • exportfits

step 5: hybrid

  imname = imbase + cleansetup + hybridsetup + str(sdfac_h[i]) 
  e.g.     skymodel-b_120L.HB_AM_n0.0e0.hybrid_f1.0

--- Slides ---

Purpose: create (deconvolved) INT image from the concatms data using SD image as startmodel, then feather INT and SD image For this, specify in pars-file:

  • mode
  • mscale
  • masking
  • inter
  • nit
  • specsetup
  • general_tclean_param
  • sdfac_h

what it does:

  • runWSM
    • prepare SD image for startmodel use
    • runtclean with SD image as startmodel
    • runfeather

step 6: SDINT

  jointname = imbase + cleansetup + sdintsetup + str(sdg[i]) 
  e.g.     skymodel-b_120L.HB_AM_n0.0e0.sdint_g1.0

--- Slides ---

Purpose: create (deconvolved) SD+INT image from the concatms data with tclean working on feathered image For this, specify in pars-file:

  • mode
  • mscale
  • masking
  • inter
  • nit
  • specsetup
  • general_tclean_param
  • sdint_tclean_param
  • sdg

what it does:

  • runsdintimg
    • image preparation (perplanebeams in SD image)
    • sdintimaging

step 7: TP2VIS

  imname = imbase + cleansetup + TP2VISsetup + str(TPfac[i]) 
  e.g.     skymodel-b_120L.HB_AM_n0.0e0.TP2VIS_t1.0

--- Slides ---

Purpose: create SD visibilities from SD image, combine SD and INT visibilities and create deconvolved image from it. For this, specify in pars-file:

  • mode
  • mscale
  • masking
  • inter
  • nit
  • specsetup
  • general_tclean_param
  • TPfac
  • TPpointingTemplate
  • listobsOutput
  • TPpointinglist
  • TPpointinglistAlternative
  • TPnoiseRegion
  • TPnoiseChannels

what it does:

  • get pointings: user specified or listobs_ptg on 12m ms-file
  • create_TP2VIS_ms
    • imstat: get noise
    • TP2VIS: transform SD image into SD.ms
  • transform_INT_to_SD_freq_spec
    • get_SD_cube_params: get frequnecy range of SD.image
    • mstransform: tranform INT.ms to SD reference frame
  • runtclean_TP2VIS_INT
    • concat: INT.ms + SD.ms
    • runtclean: on combined *.ms
    • tp2vistweak: correct for dirty/clean beam mismatch

step 8: assessment

--- Slides ---

Purpose: analyze combination results For this, specify in pars-file:

  • momchans
  • mapchan
  • skymodel
  • assessment_thresh

what it does:

all functions on continuum and mom0, Compare_Apar/Fidelity_cubes on cubes

  • numbers of involved step appear at file name end ---> allows assessment/comparison of subselection of steps.