Getting Started in CASA

From CASA Guides
Jump to navigationJump to search

Getting Started

Starting CASA

Navigate to your working directory in an xterm window. Start CASA by typing:

## From a UNIX shell prompt
casapy

on the Unix command line. CASA is built in iPython, and you will see a number of initialization messages, a list of available tasks, and, finally, a Python command line prompt:

CASA <1>:

CASA will also spawn a separate logger window, which logs messages from various tasks. Logger messages are saved in a file 'casapy.log', in your local directory. Whenever you start CASA, the previous log is renamed (based on the date and time), and a new 'casapy.log' is opened. In addition, your command line history is automatically maintained and stored as 'ipython.log' in your local directory.

Note that on startup, CASA adopts current settings of environment variables, so set them before starting CASA. For example, PAGER regulates the help display. The cat option works well in both interactive and script mode. In bash, use PAGER=cat. In csh or tcsh, use setenv PAGER cat.

Listing Available Tasks and Tools

CASA processes are organized into tasks and tools. Tools are the underlying functionality of CASA, intended for expert users participating in development. New users should ignore the tools. Tasks utilize a number of tools and accomplish a specific data analysis or calibration goal. To get information about tasks while in CASA, use:

# From within CASA
tasklist()

A more exhaustive summary can be obtained as follows:

taskhelp

For a similar summary of the CASA tools, type:

toolhelp

Getting Help on a Task

Let's get some exhaustive help on the CASA imaging task clean.

# From within CASA
help clean

Task Execution

There are two basic ways to execute a task in CASA.

1) The first is to list the task parameters and set them one by one. Let's say we would like to run the task importasdm. To load this task's parameters, type:

default('importasdm')

and then type

inp

to view this task's default parameters. They'll look something like this:

XXXX INPUT SCREEN SHOT HERE XXXXX

You can now set individual parameters from the command line, simply using a 'CASA_parameter_name = User_setting' syntax:

asdm = 'TOSR0005_sb1475761_1.55351.74409018519'
vis = 'awesome_data.ms'

Do another inp to see the updated parameters and make sure everything looks ok. Now you're ready to run the task! Type

go

Note that you can also summon a task's parameters by typing:

tget importasdm

instead of default('importasdm'). However, in this case, if you have run importasdm before, the parameters you used last time will be recalled.


2) The second strategy for executing tasks is critical for scripting, because it can be non-interactive. You'll type:

importasdm(asdm = 'TOSR0005_sb1475761_1.55351.74409018519', vis = 'awesome_data.ms')

and then simply press enter. This will both set the parameters for importasdm and run the task, all in one go. Note that any parameters you have not explicitly set in the parentheses will be set to their defaults.

Setting Parameters

Each task has a set of parameters. If a task is successfully executed, then the successful parameter set is stored in a '.last' file in the working directory. Other ways to save and restore parameters are available:

  • tget <taskname>: retrieves parameters from <taskname>.last;
  • default('<taskname>'): restores the default parameters;
  • saveinputs(<taskname>,<filename>): saves the current set of parameters into <filename>;
  • execfile '<filename>': retrieves a set of parameters.

Parameters are set using the Python <parameter>=<value> syntax. Some parameters have sub-parameters: an inp listing will show these on a gray background. If a gray parameter is set to a value that allows sub-parameters, then the sub-parameters are listed in the next instance of inp. Otherwise, the sub-parameters remain hidden. CASA uses multiple font colors and highlighting to give more information about the parameters and their current values. Pay particular attention if the text color is red: the value of the parameter is invalid, and the task will not run.

CASA has two distinct ways of running tasks, one sets parameters globally (so that all tasks will use the same values), and the other does not. This is helpful when writing scripts that you may wish to use more than once. To set global parameters, set each parameter, and then invoke the task:

#from within CASA
CASA <1>: default('plotxy')

CASA <2>: vis='ngc5921.ms'

CASA <3>: xaxis='channel'

CASA <4>: yaxis='amp'

CASA <5>: go

To set parameters for a single instance of a single task, call it as a function:

#from within CASA
CASA <1>: plotxy(vis='ngc5921.ms',xaxis='channel',yaxis='amp')

Other tasks will revert to global values of the parameters (these may be the defaults). When a task is invoked in this way, unspecified values use the defaults, not the global values!

The go command may also be used to invoke a task without changing the current task. For example, you may be setting parameters for plotcal, and wish to verify an antenna name. Typing go listobs will list the observations in the logger, but will not change the current task to listobs. Typing inp after this will give the inputs list for plotcal, since the current task has not been changed.

Quitting CASA

You can exit CASA by typing:

# From within CASA
quit

or

# From within CASA
exit

or CTRL-D at the casapy prompt.

If you don't want to see the question "Do you really want to exit [y]/n?", then just type

# From within CASA
Exit

and CASA will stop right then and there.

What If Something Goes Wrong?

First, always check that your inputs are correct; use

inp <taskname>

Next, check the inline documentation using the help facility:

help <taskname>

or

help par.<parameter_name>

to review the allowed parameters. For example,

help par.selectdata

will give guidance on the use of this parameter (and in this case its sub-parameters). One common source of confusion is parameter format---should it be entered as a string or as a float/integer? For example, in flagdata, antenna diameter should be entered as a float (diam = 28.0 in mode = 'shadow'. Note there are no quotes around 28.0); antenna name, on the other hand, is a string (even though it may look like a simple number---antenna = '3'. Note the quotes around '3').

If something has gone wrong and you want to stop what is executing, then typing CTRL-C will usually cleanly abort the application. Alert: you may lose contact with the logger GUI if you have one, in which case you will want to restart CASA.

If CASA crashes, then usually restarting it is sufficient to get you going again. Note that there may be spawned subprocesses still running, such as the casaviewer or the logger. These can be dismissed manually in the usual manner. After a crash, there may also be hidden processes. You can find these by listing processes, e.g. in linux:

ps -elf | grep casa

or on MacOSX (or other BSD Unix):

ps -aux | grep casa

You can then kill these, for example using the Unix kill or killall commands. This may be necessary if you are running remotely using ssh, as you cannot logout until all your background processes are terminated. For example,

killall ipcontroller

or

killall Python

will terminate the most common post-crash zombies.

NRAO Access to Stable Version (Not Release)

At NRAO you will have access to the current stable version of CASA. This will exceed the capabilities of the release but it will see changes on faster timescales. This version can be started via:

## From a UNIX shell prompt
casapy-test

Python Basics

Typing help at the CASA prompt with no arguments will bring up the native Python help, which has information about Python commands and keywords. Hitting <RETURN> at the Python help> prompt will return you to CASA.

Command-line Completion and Recall

Hitting the <TAB> key completes unambiguous commands or variable names and shows you a list of the possible completions, if there are more than one. It also completes filenames from the working directory if no CASA or Python commands match. Once you have typed enough of a command to make it unique, <TAB> will complete it. This will help you avoid errors due to typos.

You can also use up-and-down-arrow command-line recall in the CASA interface. Begin typing a command, and then use the up arrow; you will navigate back and edit commands matching what you typed.

Setting Variables

Set Python variables using <parameter>=<value> syntax. Python assigns the type dynamically as you set the value, and will allow you to assign an integer value to a string parameter. However, CASA will check types before running a task, and alert you if you have made a mistake. Python variable names are case sensitive. Also note that mis-spelling a variable assignment will not be noticed by the interface. For example, you might wish to set correlation='RR', but instead type corellation='RR'. You will find correlation unset, and a new variable, corellation will exist, with value 'RR'. Using command-line completion and recall helps avoid these errors.

Lists and Ranges

The Python range function give a task a list of indices. For example,

iflist=range(4,8)

will set

iflist = [4, 5, 6, 7]

Indexes

Python indices are 0-based. The first element in the list antlist is antlist[0]. CASA also uses 0-based indexing. Field or antenna IDs, for example, start at 0. The first field in a measurement set has FIELD_ID==0.

Indentation

Python uses indentation to determine the level of nesting in loops. Be careful when cutting and pasting, you may get an error (or an erroneous result).

System shell access

You can access most system commands (ls, rm, mv, pwd, less) directly from the CASA command line; no need for a bang (!). However, for some, like cp, you will need to use a !:

!cp vla evla

The cd command must not use the !.

Note: to access a Unix environment variable from CASA, you will need to use $$ instead of $.

Running Scripts

You may wish to run scripts which contain series of commands for data calibration, etc. To execute the script contained in the text file 'myscript.py', type

execfile('myscript.py')

Additional Tips

New Startup Screen

  • When you started CASA, it gave you a lot of useful information. But now you've been mucking around for a while, and that info is lost. How do you get it back? Just rerun the startup screen.
startup

Setting Up the Available Memory

CASA's default is to use 90% of the memory available. This should be a good setting for most users.

However, if this does not suit your needs, you can set the amount of memory that CASA will use, either in absolute or relative units.

To do so, create a '~/.casarc' file in your home directory (if you still have an '~/.aipsrc' in your home directory, it would be wise to remove it).

Within '~/.casarc' add the following variable:

system.resources.memory: <absolute_number>

or

system.resources.memfrac: <relative_number>

where <absolute_number> stands for the allocated memory in units of MB. e.g. for 16GB of memory the number would be 16000. <relative_number> stands for the percentage of memory that CASA can access, e.g. a value of 80 corresponds to 80% of the memory installed.

Now restart CASA.

Starting CASA Without the Logger

You also have the option of starting CASA without the logger, for example if you are running remotely in a terminal window without an X11 connection, or if you just do not want to see the logger GUI. In this case use the --nologger option:

casapy --nologger

The logger messages will still go into the 'casapy.log' file. If you want to see these at your terminal, use the --log2term option also:

casapy --nologger --log2term

Capturing Return Values

Some tasks return information to the interface. For example, the imstat task returns a Python dictionary with the image statistics in it. To catch these return statistics, you must assign a variable to the task call. For example:

xstat=imstat('ngc5921.clean.image')

To see what is in the dictionary, type the variable name (in this case, xstat) at the command line.

NRAO Access to Stable Version (Not Release)

At NRAO you will have access to the current stable version of CASA. This will exceed the capabilities of the release but it will see changes on faster timescales. This version can be started via:

## From a UNIX shell prompt
casapy-test

--Jack Gallimore 14:42, 30 October 2009 (UTC)