Python

Introduction

Python is a programming language. See www.python.org for more information. Python has two flavours a 2.X and a 3.X series. The 2.X series became unsupported at the end of 2020 and should not be used for new code. For portable future-proof code, use the 3.X series. However, the default version of python on many of our operating systems is from the 2.X series, althought this is changing. Don't use the default system python, use one of the many python 3 programs installed.

A good way to get started working with python is to use the Anaconda Python distribution, this has over 200 packages preinstalled and you can add many others using the 'conda install package_name' command which does not require sudo or root privileges. If you want to install Anaconda python on an Astrophysics machine, you don't need any elevated rights, just follow the web page here.

Documentation

See docs.python.org.

Initialisation

There are many versions of python installed on the Astrophysics systems, and which one you get is determined by your PATH variable. Here are some of the ways you can use python.

Anaconda Python

You install this yourself in your home directory. There are instructions here. The quick and dirty method to use Anaconda python is to change your PATH variable to put the Anaconda bin directory first in your PATH. You can do this as shown below: 
Substitute your username for XXXXX below.

bash/sh shells:
export PATH=/home/XXXXX/anaconda3/bin:$PATH

tcsh/csh shells:
setenv PATH /home/XXXXX/anaconda3/bin:$PATH

To save on typing, make an alias in your .bashrc file to run the
command above.

bash:
alias ana3='export PATH=/home/XXXXX/anaconda3/bin:$PATH'

tcsh:
alias ana3 'setenv PATH /home/XXXXX/anaconda3/bin:$PATH'
If you are using anaconda python in a virtual environment, you must activate that environment using the conda command, eg 
conda activate my_environment

you need to have created this environment and defined the python version beforehand.

python3 command

You should be able to access a recent python 3 using the python3 command, so just use python3 instead of python. This version won't have as many packages installed as Anaconda python, but you can install any missing ones using the pip or pip3 commands. For example 
pip3 install astropy --user

Python module

Some of our more powerful servers have the module command installed, you use this as below: 
To see what is available, type:

module available

and you should see something like the below.

------------------------ /usr/share/Modules/modulefiles ------------------------
dot         module-git  module-info modules     null        use.own

------------------------------- /opt/modulefiles -------------------------------
python/3.9.3_x86_64

to use this version of python type:

module load python/3.9.3_x86_64

Verifying the Python Version

You can check this using the 'which' command as below: 
which python
/usr/bin/python
You can check which version of python that is by using the '-V' option as below: 
python -V
Python 2.6.6

Version

Lots of versions. Unless you have a very good reason not to, new code should use Python 3 syntax. Other versions can be found in /usr/local/Python-*
Also check out the parallelised Intel Python here.

Testing

Put the following in a text file called pythontest.py and make it executable with chmod +x pythontest.phy, then run it as ./pythontest.py. 
#!/usr/bin/python3

print ("Hello from Python")
The version of python named on the first line is the version that will be used to run the code that follows it.

Usage

Python can be run interactively after typing the name of whichever version of python you want, type your commands at the '>>>' prompt or you can write a script where the first line is #!/usr/blah/python_version. Don't forget to make the script executable with a chmod +x script_name.

You can also run python in a notebook with 'live code' such as Jupyter Notebook.

Related Software

Pip and Conda

If you are using anaconda python, use 'conda install' to install packages if you are using a system python, then you can use 'pip install --user' to install packages, neither command requires sudo or root access. 
pip install --user astropy
or if you are using a 3.6.X version of Python 
pip36 install --user matplotlib

or to upgrade a package to the latest available version:

pip36 install --user matplotlib --upgrade
You should also add ~/.local/bin to your PATH variable as that is where pip puts executable programs. Add this to the end of your .cshrc file, using your own username. 
setenv PATH /homeb/yourusername/.local/bin:${PATH}
Rhys Morris
Last modified: Monday, 22 March 2021