Open in Colab: https://colab.research.google.com/github/casangi/casadocs/blob/master/docs/notebooks/interactive_clean_notebook.ipynb

Interactive Clean

The CASA interactive clean application is available for use in Jupyter notebooks. The primary difference between usage in Jupyter and usage directly from a Python command line interface (CLI) is that the iclean.notebook function is used within a notebook instead of the iclean function that is used from the Python CLI.

Current status

Currently the interactive clean app works well when a single instance is run within a notebook and the Jupyter kernel is running on the same host as the browser displaying the notebook. Things begin to go wrong when the red stop sign is pressed. Doing this is necessary because this is how the iclean return dictionary is returned. The get_result function blocks until the red stop sign is clicked and confirmed. At this point, the iclean return dictionary is returned as the result of get_result. Another function called get_future is available which does not block but instead retrieved the future where the iclean return dictionary will eventually appear.

Supporting, multiple iclean instances within a single worksheet is the goal, but currently only one instance should be used, unless you are willing to risk corruption of your Jupyter session. All of the GUI display cells in this notebook are beleved to work when a single one is executed within a given notebook instance.

Which version of cubevis to use

The current version of cubevis to use is v1.0.14.

Setup

This first cell ensures that CASA 6 and cubevis are available:

[ ]:

!mkdir -p ~/.casa/data
!pip install --upgrade "bokeh>=3.8" "casatasks>=6.7.2.42" cubevis

This cell simply retrieves the test data that is required.

[ ]:

import os
import ssl
import certifi
import urllib
import tarfile
ms_path = 'refim_point_withline.ms'
ms_url = "https://casa.nrao.edu/download/devel/casavis/data/refim_point_withline-ms.tar.gz"
if not os.path.isdir(ms_path):
    try:
        context = ssl.create_default_context(cafile=certifi.where())
        tstream = urllib.request.urlopen(ms_url, context=context, timeout=400)
        tar = tarfile.open(fileobj=tstream, mode="r:gz")
        tar.extractall(filter='data')
    except urllib.error.URLError:
        print("Failed to open connection to "+ms_url)

Import

The first step is to import the iclean function:

[ ]:

from cubevis import iclean

Cleanup

iclean operates on state contained within files in the current directory, so it is always a good idea to make sure that you are starting with a clean slate. This command removes the state files that accumulate from running this example.

[ ]:

!rm -rf test.*

Create the Application Object

This cell creates the interactive clean application object. This object is used to display the GUI and to retreive the iclean return dictionary from the GUI once the user has completed cleaning.

[ ]:

ic = iclean.notebook( vis=ms_path, imagename='test',
                      imsize=512, niter=50,
                      cycleniter=10,
                      cell='12.0arcsec',
                      specmode='cube', interpolation='nearest',
                      nchan=5, start='1.0GHz', width='0.2GHz', pblimit=-1e-05,
                      deconvolver='hogbom', threshold='0.001Jy', cyclefactor=3, scales=[0,3,10] )

Display the GUI

Once the application object has been created, the GUI can be displayed:

[ ]:

ic.show( )

Retrieve the result

Once the user has completed cleaning, clicked the stop sign and approved the exiting the application, the result can be retrieved:

[ ]:

ic.get_result( )

Alternate ways to display the GUI

While this is the most obvious way to create and use the interactive clean within a Jupyter notebook, there are also some subtle variations.

Display as a result of cell evaulation

The last evaluated statement in a notebook cell is displayed and displaying the iclean application object results in the GUI being displayed.

[ ]:

!rm -rf test.*
iclean.notebook( vis=ms_path, imagename='test',
                 imsize=512, niter=50,
                 cycleniter=10,
                 cell='12.0arcsec',
                 specmode='cube', interpolation='nearest',
                 nchan=5, start='1.0GHz', width='0.2GHz', pblimit=-1e-05,
                 deconvolver='hogbom', threshold='0.001Jy', cyclefactor=3, scales=[0,3,10] )

Retrieving the result

Even though we did not retain the application object, the value of the last cell evaluation is available as _. So using this, we can retrieve the return dictionary:

[ ]:

_.get_result( )

Using Bokeh’s show function

One last slight variation is to use Bokeh’s show function to display the application object:

[ ]:

!rm -rf test.*
from bokeh.io import output_notebook
from bokeh.plotting import show
output_notebook( )
ic = iclean.notebook( vis=ms_path, imagename='test',
                      imsize=512, niter=50,
                      cycleniter=10,
                      cell='12.0arcsec',
                      specmode='cube', interpolation='nearest',
                      nchan=5, start='1.0GHz', width='0.2GHz', pblimit=-1e-05,
                      deconvolver='hogbom', threshold='0.001Jy', cyclefactor=3, scales=[0,3,10] )
show(ic)