Post-processing How-To¶
This is a list of FAQs about the usage of the AbiPy scripts. Feel free to suggest new entries!
Preliminary considerations¶
The AbiPy scripts detect the file type by looking at the file extension
so don’t change the file extension.
Also, remember that it is possible to get the documentation by just adding
the --help option to the command line.
For example:
abistruct.py --help
gives the documentation and usage examples for the abistruct.py script while:
abistruct.py COMMAND --help
prints the documentation and the options supported by COMMAND
Get information about a generic FILE¶
Use:
abiopen.py FILE --print
to print information about a file inside the terminal or
abiopen.py FILE –expose
to generate multiple matplotlib figures depending on FILE.
Use --verbose or -v to increase verbosity level.
The option can be can be supplied multiple times e.g. -vv.
Get all file extensions supported by abiopen.py¶
Use:
abiopen.py --help
$ abiopen.py --help
/usr/share/miniconda/envs/abipy/bin/abiopen.py:4: DeprecationWarning: pkg_resources is deprecated as an API. See https://setuptools.pypa.io/en/latest/pkg_resources.html
__import__('pkg_resources').require('abipy==0.9.8')
usage: abiopen.py [-h] [--loglevel LOGLEVEL] [-V] [-v] [-nb]
[--classic-notebook] [--no-browser] [--foreground] [-p]
[-pn] [-pnt PANEL_TEMPLATE] [--port PORT] [-e] [-s]
[-t SLIDE_TIMEOUT] [-sns [SEABORN]] [-mpl MPL_BACKEND] [-ew]
[-ply]
filepath
positional arguments:
filepath File to open. See table below for the list of
supported extensions.
options:
-h, --help show this help message and exit
--loglevel LOGLEVEL Set the loglevel. Possible values: CRITICAL, ERROR
(default), WARNING, INFO, DEBUG
-V, --version show program's version number and exit
-v, --verbose verbose, can be supplied multiple times to increase
verbosity
-nb, --notebook Open file in jupyter notebook
--classic-notebook, -cnb
Use classic jupyter notebook instead of jupyterlab.
--no-browser Start the jupyter server to serve the notebook but
don't open the notebook in the browser. Use this
option to connect remotely from localhost to the
machine running the kernel
--foreground Run jupyter notebook in the foreground.
-p, --print Print python object and return.
-pn, --panel Open Dashboard in web browser, requires panel package.
-pnt PANEL_TEMPLATE, --panel-template PANEL_TEMPLATE
Specify template for panel dasboard.Possible values
are: FastList, FastGrid, Golden, Bootstrap, Material,
React, Vanilla.Default: FastList
--port PORT Allows specifying a specific port when serving panel
app.
-e, --expose Open file and generate matplotlib figures
automatically by calling expose method.
-s, --slide-mode Iterate over figures. Expose all figures at once if
not given on the CLI.
-t SLIDE_TIMEOUT, --slide-timeout SLIDE_TIMEOUT
Close figure after slide-timeout seconds (only if
slide-mode). Block if not specified.
-sns [SEABORN], --seaborn [SEABORN]
Use seaborn settings. Accept value defining context in
("paper", "notebook", "talk", "poster"). Default:
paper
-mpl MPL_BACKEND, --mpl-backend MPL_BACKEND
Set matplotlib interactive backend. Possible values:
GTKAgg, GTK3Agg, GTK, GTKCairo, GTK3Cairo, WXAgg, WX,
TkAgg, Qt4Agg, Qt5Agg, macosx.See also:
https://matplotlib.org/faq/usage_faq.html#what-is-a-
backend.
-ew, --expose-web Generate matplotlib plots in $BROWSER instead of
X-server. WARNING: Not all the features are supported.
-ply, --plotly Generate plotly plots in $BROWSER instead of
matplotlib. WARNING: Not all the features are
supported.
======================================================================================================
Usage example:
abiopen.py FILE => Open file in ipython shell.
abiopen.py FILE -p => Print info on object to terminal.
abiopen.py FILE -e => Generate matplotlib figures automatically.
Use -sns to activate seaborn settings.
abiopen.py FILE -eweb => Generate matplotlib figures, show them in the $BROWSER.
abiopen.py FILE -ply => Generate plotly figures automatically. Show them in the $BROWSER.
Note that not all FILEs support plotly.
abiopen.py FILE -pn => Generate GUI in web BROWSER to interact with FILE. Requires panel package.
abiopen.py FILE -nb => Generate jupyter-lab notebook.
abiopen.py FILE -cnb => Generate classic jupyter notebook.
where `FILE` is any file supported by abipy/pymatgen e.g. Netcdf files, Abinit input, POSCAR, xsf.
File extensions supported (including zipped files with extension in ".bz2", ".gz", ".z"):
Use `-v` to increase verbosity level (can be supplied multiple times e.g -vv).
JSON file are supported as well. In this case, abiopen.py tries to reconstruct python objects
assuming JSON document in MSONable format and then invokes ipython with the `data` object.
Use `-e` or `--notebook` or `--panel` to print the JSON dictionary without reconstructing python objects.
======================================================================================================
Table mapping file extension to AbiPy object:
Extension Class
----------------- -----------------------------------------------------
.abi <class 'abipy.abio.abivars.AbinitInputFile'>
.in <class 'abipy.abio.abivars.AbinitInputFile'>
.abo <class 'abipy.abio.outputs.AbinitOutputFile'>
.out <class 'abipy.abio.outputs.AbinitOutputFile'>
.log <class 'abipy.abio.outputs.AbinitLogFile'>
.cif <class 'abipy.core.structure.Structure'>
.abivars <class 'abipy.core.structure.Structure'>
.ucell <class 'abipy.core.structure.Structure'>
.cssr <class 'abipy.core.structure.Structure'>
.json <class 'abipy.core.mixins.JsonFile'>
.py <class 'abipy.core.mixins.TextFile'>
.sh <class 'abipy.core.mixins.TextFile'>
.stdin <class 'abipy.core.mixins.TextFile'>
.stderr <class 'abipy.core.mixins.TextFile'>
.err <class 'abipy.core.mixins.TextFile'>
.files <class 'abipy.core.mixins.TextFile'>
.stdout <class 'abipy.core.mixins.TextFile'>
.cube <class 'abipy.core.mixins.CubeFile'>
anaddb.nc <class 'abipy.dfpt.anaddbnc.AnaddbNcFile'>
DEN <class 'abipy.electrons.denpot.DensityFortranFile'>
.wout <class 'abipy.wannier90.wout.WoutFile'>
EDOS <class 'abipy.electrons.ebands.EdosFile'>
.psp8 <class 'pymatgen.io.abinit.pseudos.Pseudo'>
.pspnc <class 'pymatgen.io.abinit.pseudos.Pseudo'>
.fhi <class 'pymatgen.io.abinit.pseudos.Pseudo'>
JTH.xml <class 'pymatgen.io.abinit.pseudos.Pseudo'>
.upf <class 'pymatgen.io.abinit.pseudos.Pseudo'>
COHPCAR.lobster <class 'abipy.electrons.lobster.CoxpFile'>
COOPCAR.lobster <class 'abipy.electrons.lobster.CoxpFile'>
ICOHPLIST.lobster <class 'abipy.electrons.lobster.ICoxpFile'>
DOSCAR.lobster <class 'abipy.electrons.lobster.LobsterDoscarFile'>
POSCAR <class 'abipy.core.structure.Structure'>
.vasp <class 'abipy.core.structure.Structure'>
.xyz <class 'abipy.core.structure.Structure'>
.evp <class 'abipy.dynamics.cpx.EvpFile'>
GSR.nc <class 'abipy.electrons.gsr.GsrFile'>
ESKW.nc <class 'abipy.electrons.eskw.EskwFile'>
DEN.nc <class 'abipy.electrons.denpot.DensityNcFile'>
OUT.nc <class 'abipy.abio.outputs.OutNcFile'>
VHA.nc <class 'abipy.electrons.denpot.VhartreeNcFile'>
VXC.nc <class 'abipy.electrons.denpot.VxcNcFile'>
VHXC.nc <class 'abipy.electrons.denpot.VhxcNcFile'>
POT.nc <class 'abipy.electrons.denpot.PotNcFile'>
WFK.nc <class 'abipy.waves.wfkfile.WfkFile'>
HIST.nc <class 'abipy.dynamics.hist.HistFile'>
PSPS.nc <class 'abipy.electrons.psps.PspsFile'>
DDB <class 'abipy.dfpt.ddb.DdbFile'>
PHBST.nc <class 'abipy.dfpt.phonons.PhbstFile'>
PHDOS.nc <class 'abipy.dfpt.phonons.PhdosFile'>
SCR.nc <class 'abipy.electrons.scr.ScrFile'>
SIGRES.nc <class 'abipy.electrons.gw.SigresFile'>
GWR.nc <class 'abipy.electrons.gwr.GwrFile'>
GRUNS.nc <class 'abipy.dfpt.gruneisen.GrunsNcFile'>
MDF.nc <class 'abipy.electrons.bse.MdfFile'>
FATBANDS.nc <class 'abipy.electrons.fatbands.FatBandsFile'>
FOLD2BLOCH.nc <class 'abipy.electrons.fold2bloch.Fold2BlochNcfile'>
CUT3DDENPOT.nc <class 'abipy.electrons.denpot.Cut3dDenPotNcFile'>
OPTIC.nc <class 'abipy.electrons.optic.OpticNcFile'>
A2F.nc <class 'abipy.eph.a2f.A2fFile'>
SIGEPH.nc <class 'abipy.eph.sigeph.SigEPhFile'>
GSTORE.nc <class 'abipy.eph.gstore.GstoreFile'>
GPATH.nc <class 'abipy.eph.gpath.GpathFile'>
TRANSPORT.nc <class 'abipy.eph.transportfile.TransportFile'>
RTA.nc <class 'abipy.eph.rta.RtaFile'>
V1SYM.nc <class 'abipy.eph.v1sym.V1symFile'>
GKQ.nc <class 'abipy.eph.gkq.GkqFile'>
V1QNU.nc <class 'abipy.eph.v1qnu.V1qnuFile'>
V1QAVG.nc <class 'abipy.eph.v1qavg.V1qAvgFile'>
ABIWAN.nc <class 'abipy.wannier90.abiwan.AbiwanFile'>
EPH_CUMULANT.nc <class 'abipy.eph.cumulant.CumulantEPhFile'>
VARPEQ.nc <class 'abipy.eph.varpeq.VarpeqFile'>
Convert the structure stored in FILE to a different format¶
Use:
abistruct.py convert FILE
to read the structure from FILE and generate a CIF file (default behaviour).
The majority of the netcdf files produced by Abinit contain structural information
so this command can be used with netcdf output files as well as Abinit input/output
files and all the other formats supported by pymatgen e.g. POSCAR files.
Other formats can be specified with the -f option.
For example:
abistruct.py convert FILE -f abivars
gives the Abinit input variables while:
abistruct.py convert FILE -f xsf > out.xsf
exports the structure to the xsf format (xcrysden) and save it to file.
Use:
abistruct.py convert --help
to list the formats supported.
Check if my Abinit input file is OK¶
First of all, one can use:
abiopen.py ../abipy/data/refs/si_ebands/run.abi -p
to print the crystalline structure and find the space group with the spglib library.
If the structure looks good, use the abinp.py script with the validate command as in:
abinp.py validate run.abi
to validate the input file with Abinit (requires manager.yml and, obviously, Abinit).
The script provides other options to invoke Abinit
to get space group information, the list of k-points in the IBZ.
the list of atomic perturbations for phonons or the list of autoparal configurations.
See abinp.py --help for further info.
Print the warnings in the log file¶
Use:
abiopen.py run.log -p
to get:
Events found in /Users/gmatteo/git_repos/abipy/abipy/examples/flows/develop/flow_from_files/w0/t0/run.log
[1] <AbinitWarning at m_nctk.F90:568>
netcdf lib does not support MPI-IO and: NetCDF: Parallel operation on file opened for non-parallel access
[2] <AbinitWarning at m_nctk.F90:588>
The netcdf library does not support parallel IO, see message above
Abinit won't be able to produce files in parallel e.g. when paral_kgb==1 is used.
Action: install a netcdf4+HDF5 library with MPI-IO support.
[3] <AbinitWarning at m_hdr.F90:4258>
input kptrlatt= 0 0 0 0 0 0 0 0 0 /= disk file kptrlatt=8 0 0 0 8 0 0 0 8
[4] <AbinitWarning at m_hdr.F90:4261>
input kptopt= -2 /= disk file kptopt= 1
num_errors: 0, num_warnings: 4, num_comments: 0, completed: True
A similar interface is also available via:
abiview.py log run.log
Get a quick look to a file¶
The abiview.py script is especially designed for this task.
The syntax is abiview.py COMMAND FILE where COMMAND is either
the Abinit file extension (without .nc, if any) or the AbiPy object we want to visualize.
To get a quick look at the DDB file, use:
abiview.py ddb out_DDB
This command invokes anaddb to compute phonon bands and DOS from the DDB and produces matplotlib plots.
If FILE contains electronic band energies, use e.g.:
abiview.py ebands out_GSR.nc
to plot the KS eigenvalues (the same command works for other files such as WFK.nc, DEN.nc etc.
Note that abiview.py uses a predefined logic to visualize the data. There are options to tune some parameters and/or export data in different formats but exposing the AbiPy API from the command line is not easy.
For a more flexible interface, we suggest to use:
abiopen.py FILE
to start an ipython shell that will allow you to interact with the python object directly.
If you have installed jupyter on your machine/cluster and you have a web browser, use:
abiopen.py FILE -nb
to generate automatically a predefined jupyter notebook associated to the file type.
Visualize a structure¶
The visualization of the structure is delegated to external graphical applications
that must be installed on your machine.
AbiPy will extract the structure from FILE, convert it to one of the formats
supported by the graphical application and finally invoke the executable.
If you have vesta installed in one of the standard
locations of your machine, you can simply execute:
abistruct.py visualize FILE
inside the terminal.
Other applications can be specified with the --application option.
At present, AbiPy supports vesta, ovito, xcrysden, avogadro, and v_sim.
To visualize the crystalline structure inside the jupyter notebook, you may want to try the nbjsmol jupyter extension.
Get a high-symmetry kpath for a given structure¶
Use the kpath command and pass a FILE providing structural info:
abistruct.py kpath FILE
to generate a template with the input variables defining the k-path
# Abinit Structure
natom 2
ntypat 1
typat 1 1
znucl 14
xred
0.0000000000 0.0000000000 0.0000000000
0.2500000000 0.2500000000 0.2500000000
acell 1.0 1.0 1.0
rprim
0.0000000000 5.1085000000 5.1085000000
5.1085000000 0.0000000000 5.1085000000
5.1085000000 5.1085000000 0.0000000000
# K-path in reduced coordinates:
# tolwfr 1e-20 iscf -2 getden ??
ndivsm 10
kptopt -11
kptbounds
+0.00000 +0.00000 +0.00000 # $\Gamma$
+0.50000 +0.00000 +0.50000 # X
+0.50000 +0.25000 +0.75000 # W
+0.37500 +0.37500 +0.75000 # K
+0.00000 +0.00000 +0.00000 # $\Gamma$
+0.50000 +0.50000 +0.50000 # L
+0.62500 +0.25000 +0.62500 # U
+0.50000 +0.25000 +0.75000 # W
+0.50000 +0.50000 +0.50000 # L
+0.37500 +0.37500 +0.75000 # K
+0.62500 +0.25000 +0.62500 # U
+0.50000 +0.00000 +0.50000 # X
Re-symmetrize a structure when Abinit reports less symmetries than expected¶
Crystalline structures saved in text format (e.g. CIF files downloaded from the Materials Project websites) may not have enough significant digits and Abinit may not find the same spacegroup as the one reported by the source as the default tolerance for symmetry detection in Abinit is tight (tolsym = 1e-8).
In this case, one can use the abispg option of abistruct.py to compute the spacegroup with Abinit and a tolerance larger that the default value:
abistruct.py abispg problematic.cif --tolsym=1e-6
Hopefully, the code will detect the correct spacegroup, will re-symmetrize the initial lattice vectors and atomic positions and print the new symmetrized structure to terminal.
Get neighbors for each atom in the unit cell out to a distance radius¶
If we are interested in the environment/nearest neighbours of the atoms in the unit cell, we can analyze the different coordinations with:
abistruct.py neighbors sic_relax_HIST.nc
Finding neighbors for each atom in the unit cell, out to a distance 2 [Angstrom]
[0] site PeriodicSite: C (0.0000, -0.0000, 0.0000) [-0.0000, 0.0000, -0.0000] has 4 neighbors:
PeriodicSite: Si (1.0836, -1.0836, -1.0836) [-0.7500, 0.2500, 0.2500] at distance 1.8767766107
PeriodicSite: Si (-1.0836, 1.0836, -1.0836) [0.2500, -0.7500, 0.2500] at distance 1.8767766107
PeriodicSite: Si (-1.0836, -1.0836, 1.0836) [0.2500, 0.2500, -0.7500] at distance 1.8767766107
PeriodicSite: Si (1.0836, 1.0836, 1.0836) [0.2500, 0.2500, 0.2500] at distance 1.8767766107
[1] site PeriodicSite: Si (1.0836, 1.0836, 1.0836) [0.2500, 0.2500, 0.2500] has 4 neighbors:
PeriodicSite: C (0.0000, 0.0000, 0.0000) [0.0000, 0.0000, 0.0000] at distance 1.8767766107
PeriodicSite: C (2.1671, 2.1671, 0.0000) [0.0000, 0.0000, 1.0000] at distance 1.8767766107
PeriodicSite: C (2.1671, 0.0000, 2.1671) [0.0000, 1.0000, 0.0000] at distance 1.8767766107
PeriodicSite: C (0.0000, 2.1671, 2.1671) [1.0000, 0.0000, 0.0000] at distance 1.8767766107
Search on the Materials Project database for structures¶
Use:
abistruct.py mp_search LiF
to search on the materials project database for structures corresponding to a
chemical system or formula e.g. Fe2O3 or Li-Fe-O or
Ir-O-* for wildcard pattern matching.
The script prints the results to terminal in tabular form:
# Found 2 structures in materials project database (use `verbose` to get full info)
pretty_formula e_above_hull energy_per_atom \
mp-1138 LiF 0.000000 -4.845142
mp-1009009 LiF 0.273111 -4.572031
formation_energy_per_atom nsites volume spacegroup.symbol \
mp-1138 -3.180439 2 17.022154 Fm-3m
mp-1009009 -2.907328 2 16.768040 Pm-3m
spacegroup.number band_gap total_magnetization material_id
mp-1138 225 8.7161 0.0 mp-1138
mp-1009009 221 7.5046 -0.0 mp-1009009
Important
The script will try to connect to the materials project server.
You need a ~/.pmgrc.yaml configuration file inside your home directory
with the authentication token PMG_MAPI_KEY.
For further info please refer to the
pymatgen documentation
The script provides other commands to get (experimental) structures from the COD database,
find matching structures on the materials project website and generate phase diagrams.
See abistruct.py --help for more examples.
Compare my structure with the Materials Project database¶
Let’s assume we have performed a structural relaxation and we want to compare our results with the Materials Project data. One can use the abicomp.py structure to extract the structure from the HIST.nc file and compare the data with the database:
abicomp.py mp_structure ../abipy/data/refs/sic_relax_HIST.nc
It’s also possible to select only the structures with the same space group number as the input structure with:
abicomp.py mp_structure ../abipy/data/refs/sic_relax_HIST.nc --same-spgnum
that produces
Lattice parameters:
formula natom angle0 angle1 angle2 a b c \
mp-8062 Si1 C1 2 60.0 60.0 60.0 3.096817 3.096817 3.096817
this Si1 C1 2 60.0 60.0 60.0 3.064763 3.064763 3.064763
volume abispg_num spglib_symb spglib_num
mp-8062 21.000596 None F-43m 216
this 20.355222 None F-43m 216
Use --verbose to print atomic positions.
Note that one can replace the HIST.nc file with any other file providing a structure object.
Important
The structures of the materials project have been obtained with the GGA-PBE functional and they might include the U term in the Hamiltonian. One should take into account these different settings when comparing structural relaxations.
Visualize the iterations of the SCF cycle¶
Use:
abiview.py abo run.abo
to plot the SCF iterations or the steps of the structural relaxations or the DFPT SCF cycles (depending on the content of run.abo).
Note that one can also use:
abiview.py log run.log
to print the warnings/comments/errors reported in the Abinit log file run.log.
Export bands to xmgrace format¶
Both abipy.electrons.ebands.ElectronBands and abipy.dfpt.phonons.PhononBands provide a to_xmgrace method to produce xmgrace files.
To export the data to xmgrace, use abiview.py with the --xmgrace option.
For electrons, use:
abiview.py ebands out_GSR.nc --xmgrace
and:
abiview.py phbands out_PHBST.nc -xmgrace
for phonons.
Visualize the Fermi surface¶
Use:
abiview.py ebands out_GSR.nc --bxsf
to export a set of band energies in BXSF format suitable for the visualization of the Fermi surface with xcrysden. Then use:
xcrysden --bxsf BXSF_FILE
to visualize the Fermi surface with xcrysden
abifile.ebands.to_bxsf("mgb2.bxsf")
Important
This option requires k-points in the irreducible wedge and a gamma-centered k-mesh.
Visualize phonon displacements¶
AbiPy is interfaced with the phononwebsite project
If you have already installed the python package from github
it’s possbile to export the PHBST.nc to JSON and then load the file via the web-interface.
Alternatively, it’s possible to automate the entire procedure with the abiview.py script.
Use:
abiview.py phbands out_PHBST.nc -web
to start a local web server and open the HTML page inside the default browser
(the browser can be changed with the --browser option).
It is also possible to visualize the phonon modes starting directly from a DDB file with:
abiview.py ddb -web
In this case, AbiPy will invoke anaddb to produce the PHBST.nc file on an automatically
generated q-path and then start the web server.
Visualize the results of a structural relaxation¶
The quickest way is to use:
abiview hist out_HIST.nc
to plot the results with matplotlib or:
abiopen.py out_HIST.nc -p
to print the most important results to terminal.
Note that it’s possible to generate a XDATCAR file with:
abiview hist out_HIST.nc --xdatcar
and visualize the crystalline structure with ovito:
abiview hist out_HIST.nc -appname=ovito
Important
The XDATCAR format assumes a fixed unit cell so you won’t be able to visualize the modifications of the unit cell lattice vectors in ovito.
Plot results stored in a text file in tabular format¶
Use:
abiview.py data FILE_WITH_COLUMNS
to plot with matplotlib all the columns given in the file. By default, the first column is assumed to contain the values for the x-axis but it is possible to change this behaviour and use the row index with the –use-index option. Multiple datasets i.e. multiple sets of data separated by blank lines are supported.
To compare multiple files use:
abicomp.py data FILE1 FILE2
Obviously one can use standard tools such as gnuplot and xmgrace but the AbiPy scripts are quite handy for a quick analysis of the results.
Compare multiple files¶
The abicomp.py script is explicitly designed for this kind of task. It operates on multiple files (usually files with the same extension) and either produces matplotlib plots or creates AbiPy robots providing methods to analyze the results, perform convergence studies and build pandas DataFrames.
The COMMAND defines the quantity to be compared, followed by a list of filenames.
To compare e.g. the structure given in one Abinit input file with the structure coming from a GSR.nc file, use:
abicomp.py structure run.abi out_GSR.nc
Note
In this example, we can use files of different type because they both have a Structure object. This philosophy can be applied to other commands as well: everything works as long as AbiPy is able to extract the quantity of interest from the file.
To plot multiple electronic structures on a grid, use:
abicomp.py ebands *_GSR.nc out2_WFK.nc -p
Remember that it is possible to use the shell syntax *_GSR.nc to select all files with a given extension.
If you have nested directories, use unix find to scan the directory tree for files matching a given pattern
For example:
abicomp.py ebands `find . -name *_GSR.nc`
finds all GSR.nc files contained withing the current working directory.
The output of find is then passed to the abicomp.py script.
Note
Note the backticks syntax used in the command.
Profile the scripts¶
All AbiPy script can be executed in profile mode by just prepending the prof keyword
to the command line arguments.
This option could be useful if the script seems to be slow and you need to understand what’s happening.
Use:
abiopen.py prof FILE
or:
abistruct.py prof COMMAND FILE
if the script requires a COMMAND argument.
Get the description of a variable¶
The abidoc.py script provides a simplified interface to the Abinit documentation.
Use:
abidoc.py man ecut
to print the official documentation for the ecut variable to the terminal.
To list all the variables depending on the natom dimension, use:
abidoc.py withdim natom
More options are available. See abidoc.py --help.
Avoid transfering files from the cluster to localhost just to use matplotlib¶
Use SSHFS to mount the remote file system over SSH. Now one can execute the AbiPy scripts in a terminal running on the local machine. to open/visualize the files stored on the cluster.