Tips for python and edrixs

In the design of edrixs, we made a deliberate choice to use python for the application programming interface. This is because of its readability, easy of use and flexibility to run and combine it in many different ways.

The standard way to run a python script myscript.py is:

python myscript.py

This will generate the outputs of the script such as plots you choose to save and print statements will be returned into the terminal. You can save the print output by simply redirecting the output to a file:

python myscript.py > myoutput.txt

Python also includes an excellent plotting package matplotlib, which one can use to make publication quality plots.

While our aim is that the huge majority of tasks can be done without modifying the underlying code all the python layers of code are easy to modify if you would like to. If you want to modify, say, solvers.py. we would suggest copying solvers.py to your working directory under a different name e.g. my_solvers.py. Executing %run my_solvers.py from within your script will then load the functions from the file into your namespace. Just be sure to tell the script to load functions from edrixs (and not via a relative file import) i.e. from .soc import atom_hsoc should be from edrixs.soc import atom_hsoc.

For more exploratory usage, IPython or Jupyter can be very useful. (See edrixs and docker for invoking jupyter over docker.) After starting IPython by typing ipython at the command line you might find it useful to execute your script via:

%run -i myscript.py

The interactive flag -i means that all the variables and functions you loaded will be available to you at the command line. These can be straightforwardly printed or inspected via the ? or ?? flags, which show you the object documentation and the object code respectively.:

from edrixs import cd_cubic_d
cd_cubic_d??

Including %matplotlib widget in your script will facilitate interactive plots. All these interactive options are also available in the rich outputs possible within the juptyer lab interface.

A brute force option to look at a variable deep within the code is to use a debugger:

python3 -m pdb myscript.py

Use import pdb; pdb.set_trace() to set the place where you want to enter the debugger. See here for more details.

If you are feeling even braver, you can browse the Fortran code which does the heavyweight computation. Either in the source edrixs directory or via the online edrixs repo.