pyfplo

Introduction

General

The python extension pyfplo exposes some of the functionality of the FPLO package to python. Among the main features are input management (write and read of =.in), the easy creation of FPLO band/bandweight type files, the python version of faddwei and the slabify module, which allows to map Wannier function models onto larger unit cells, finite slabs and semi infinite slabs. Band structure and spectral densities can be calculated for these super structures. The Berry curvature for Weyl points can be calculated.

There are examples under ..../FPLO22.00-62/DOC/pyfplo/.

Attention

Many data objects which are returned by the pyfplo classes are copies of the underlying c++ objects, which means that:

bp=sla.BandPlot()
bp.points.append(...)

will modify the temporary object returned by pyfplo.common.BandPlot.points and then discard it. In general, the only supported operation is assignement:

l=[]
l.append(['$~G',[0,0,0]])
l.append(...)
l.append(...)

bp=sla.BandPlot()

bp.points=l

Installation

Prerequisites:

numpy must be installed on your system. You need to know how to link to lapack, if you want to use it or you use our own copy (which might not as effective).

Prepare Make:

As with the other fplo make files we need some preliminary setup. This is done by install/MMakefile together with the default setup procedure.

Make:

Go to PYTHON under your fplo source tree and run make for a python2 package and make python3 for a python3 package. Note that the example scripts need to be run via python3 script.py in the latter case. Answer all questions. As a first trial just hit enter, unless a number is required and see if the installation works. Later try to refine the installation, if needed. The result should be a new directory called pyfplo. If you want you can copy the whole pyfplo directory somewhere else.

Setup:

The standard installation procedure of python often requires admin rights, which is inconvenient. We decided to let pyfplo reside inside the FPLO source tree. To use it we need to set environment variables. If one copies pyfplo somewhere else, one need to adapt pyfplopath in the code below. This way different versions can be managed. Use export statements in your batch scripts or local shell environment. (The actual syntax for exporting environment variables depends on the shell you use. There are most certainly already examples in your local shell configuration file.) Add

# Define the pyfplo path here. The last directory in this path
# must be the one under which pyfplo resides.

pyfplopath=.../FPLO/FPLO22.00-62/PYTHON

# Now export the environment variables.
export PYTHONPATH=$pyfplopath:$PYTHONPATH

# uncomment this if you get dynamic link issues
#export LD_LIBRARY_PATH=$pyfplopath/pyfplo:$LD_LIBRARY_PATH

to your script or .bashrc (or similar config files) where ../FPLO/FPLO22.00-62/PYTHON is replaced by the actual directory your pyfplo sits in. It is important that these variables are set before the python script is executed.

If more than one pyfplo versions exist (which is often the case) it is absolutely necessary to either set the environment via .bashrc (and the like) or in job-queue scripts or to wrap pyfplo into a shell script which sets PYTHONPATH (and perhaps LD_LIBRARY_PATH) first and then executes the python script. If LD_LIBRARY_PATH is need both variables need to be set correctly otherwise python will import modules from one version and the library from another, which obviously is not good.

There is yet another possibility, which might work. You could explicitely specify the pyfplo version path in the beginning of the script before you import any pyfplo modules:

import sys
sys.path.insert(0,"PATH_UNDER_WHICH_TO_FIND_pyfplo");

where PATH_UNDER_WHICH_TO_FIND_pyfplo is a path under which your desired pyfplo version sits. This method requires to modify your python scripts, but has its own charm.

You might want to find out, which version was used. Do this:

import pyfplo.fedit as fedit
print '\npyfplo version=: {0}\nfrom: {1}\n'.format(fedit.version,fedit.__file__)