Graphical interface

AbiPy provides interactive dashboards that can be used either as a standalone web applications (dashboards) with the bokeh server or inside jupyter notebooks. This document explains how to install the required dependencies and how to generate dashboards/GUIs either with the command line interface (CLI) or inside jupyter notebooks.


Please note that one needs a running python backend to execute the callbacks triggerered by the GUI widgets. This part, indeed, is implemented in HTML/CSS/JS code executed by the frontend (i.e. your browser) that sends the signal to the python server (the backend). The python server is supposed to process the data and send the results back to the frontend for visualization purposes

Don’t be surprised if you start to click buttons and nothing happens! The examples provided in this page are only meant to show how to build GUI or dashboards with AbiPy.


Install the panel package either from pip with:

pip install panel

or with conda (recommended) using:

conda install panel -c conda-forge

If you plan to use panel within JupyterLab, you will also need to install the PyViz JupyterLab extension and activate it with:

conda install -c conda-forge jupyterlab
jupyter labextension install @pyviz/jupyterlab_pyviz

Basic Usage

Several AbiPy objects provide a get_panel method returning an object that can be displayed inside the jupyter notebook or inside the browser. When running inside a jupyter notebook, remember enable the integration with the panel infrastructure by executing:

from abipy import abilab

before calling any AbiPy get_panel method.


The abipanel function is needed to load extensions and javascript packages required by AbiPy. This function is just a small wrapper around the official panel API:

import panel as pn

At this point, we can start to construct AbiPy objects. For our first example, we use the abiopen function to open a GSR file, then we call get_panel to build a set of widgets that allows us to interact with the GsrFile:

from abipy import abilab
import as abidata

filename = abidata.ref_file("")
gsr = abilab.abiopen(filename)


The summary tab provides a string representation of the file but there is no widget to interact with it. If you select the e-Bands tab, you will see several widgets and a button that activates the visualization of the KS band energies. Again, in this HTML page there is no python server running in background so if you click the Plot e-bands button nothing happens (this is not a bug!).

The advantage of this notebook-based approach is that it is possible to mix the panel GUIs with python code that can be used to perform more advanced tasks not supported by the GUI.

Obviously it is possible to have multiple panels running in the same notebook. Calling get_panel with an AbiPy structure, for instance, creates a set of widgets to facilitate common operations such as exporting the structure to a different format or generating a basic Abinit input file for e.g. GS calculations:



At present, not all the AbiPy objects support the get_panel protocol but we plan to gradually support more objects, especially the most important netcdf files produced by Abinit

To generate a notebook from the command line, use the script: -nb  # short for --notebook

that will automatically open the notebook inside jupyterlab. If you prefer classic jupyter notebooks, use the -nb --classic-notebook options

If you do not need to execute python code, you may want to generate a panel dashboard with: -pn  # short for --panel

The same approach can be used with a DDB file. In this case, we get more tabs and options because one can use the GUI to set the input parameters, invoke anaddb and visualize the results:

# Open DDB file with abiopen and invoke get_panel method.
ddb_path = abidata.ref_file("mp-1009129-9x9x10q_ebecs_DDB")

The same result can be obtained from the CLI with mp-1009129-9x9x10q_ebecs_DDB -nb

There are, however, cases in which you don’t need the interactive environment provided by jupyter notebooks as you are mainly interested in the visualization of the results. In this case, it is possible to use the command line interface to automatically generate a dashboard with widgets without having to start a notebook.

To build a dashboard for a Structure object extracted from FILE, use: panel FILE

where FILE is any file providing a Structure object e.g. netcdf files, cif files, abi, abo files etc.

To build a dashboard associated to one of the AbiPy file, use the syntax: FILE --panel

where FILE is one of the Abinit files supported by For instance, one can create a dashboard to interact with a DDB file with: out_DDB --panel


To build a dashboard for an AbiPy Flow use: FLOWDIR panel

or alternatively: FLOWDIR/__AbinitFlow__.pickle --panel