SPHINX eigen values parser

[ahmedabdelkawy]

1. With the current Pyiron sphinx parser, the eigenvalues from the eps.*.dat files are not parsed

• in pyiron_atomistics/pyiron_atomistics/sphinx/base.py The collect function gives priority to the final eigen values in the log file self.collect_sphinx_log(file_name="sphinx.log", cwd=directory) before self.collect_eps_dat()

        def collect(self, directory=os.getcwd()):
        """
        The collect function, collects all the output from a SPHInX simulation.

        Args:
            directory (str): the directory to collect the output from.
        """
        self.collect_energy_struct(file_name="energy-structOpt.dat", cwd=directory)
        self.collect_sphinx_log(file_name="sphinx.log", cwd=directory)
        self.collect_energy_dat(file_name="energy.dat", cwd=directory)
        self.collect_residue_dat(file_name="residue.dat", cwd=directory)
        if self._job._spin_enabled:
            self.collect_eps_dat(file_name="eps.dat", cwd=directory)
        else:
            self.collect_eps_dat(
                file_name=[f"eps.{i}.dat" for i in [0, 1]], cwd=directory
            )
        self.collect_spins_dat(file_name="spins.dat", cwd=directory)
        self.collect_relaxed_hist(file_name="relaxHist.sx", cwd=directory)
        self.collect_electrostatic_potential(file_name="vElStat-eV.sxb", cwd=directory)
        self.collect_charge_density(file_name="rho.sxb", cwd=directory)

• This is then followed by

    def collect_eps_dat(self, file_name="eps.dat", cwd=None):
        """

        Args:
            file_name:
            cwd:

        Returns:

        """
        if isinstance(file_name, str):
            file_name = [file_name]
        values = []
        for f in file_name:
            file_tmp = posixpath.join(cwd, f)
            if not os.path.isfile(file_tmp):
                return
            values.append(np.loadtxt(file_tmp)[..., 1:])
        values = np.stack(values, axis=0)
        if "bands_eigen_values" not in self.generic.dft.list_nodes():
            self.generic.dft.bands_eigen_values = values.reshape((-1,) + values.shape)

• Which does not import values from the eps.*.dat files because it already imported the values from the log file

2. The problem with importing eigenvalues from the logfile is that pyiron expects to have the eigenvalues of all kpoints from the 1st spin channel (in case of spin-polarized calculations) followed by the eigenvalues at all kpoint from the second spin channel. Which happens here:

    def _parse_band(self, term):
        arr = np.loadtxt(re.findall(term, self.log_main, re.MULTILINE))
        shape = (-1, len(self.k_points), arr.shape[-1])
        if self.spin_enabled:
            shape = (-1, 2, len(self.k_points), shape[-1])
        return arr.reshape(shape)

• However, in sphinx, this is true from the eps.*.dat but from the log file, the eigenvalues are printed for spin up and then spin down for each kpoint in alternating fashion.

3. This appears as a problem only in the case of ferromagnetic systems (metals) where the 2 spin channels are different (while in nonmagnetic systems you have one spin channel and in antiferromagnetic systems you two identical but opposite dos)

4. I did not change the current parser priority. However, I changed the order of the parsed eigenvalues to match what pyiron expects: arr = np.vstack((arr[::2], arr[1::2]))

[freyso]

Please add a comment in the code why this is needed. Otherwise, this will cause a lot of headache for later devels.

[samwaseda]

I'm actually a bit concerned that it may or may not work depending on the SPHInX version? I'm saying this because it was correct in an earlier version and stopped being so with the current version. Maybe we need a version check inside the parser.

[freyso]

Sphinx has not changed here, I believe the loop order in the writing routine dates back from 2002. I suspect rather that the pyiron parser has never has been validated in this point. But please do so this time, Ahmed!

…

------------------------------------------------- Stay up to date and follow us on LinkedIn, Twitter and YouTube. Max-Planck-Institut für Eisenforschung GmbH Max-Planck-Straße 1 D-40237 Düsseldorf Handelsregister B 2533 Amtsgericht Düsseldorf Geschäftsführung Prof. Dr. Gerhard Dehm Prof. Dr. Jörg Neugebauer Prof. Dr. Dierk Raabe Dr. Kai de Weldige Ust.-Id.-Nr.: DE 11 93 58 514 Steuernummer: 105 5891 1000 Please consider that invitations and e-mails of our institute are only valid if they end with ***@***.*** If you are not sure of the validity please contact ***@***.*** Bitte beachten Sie, dass Einladungen zu Veranstaltungen und E-Mails aus unserem Haus nur mit der Endung ***@***.*** gültig sind. In Zweifelsfällen wenden Sie sich bitte an ***@***.*** -------------------------------------------------

--EdoMail64fb3410_7a6d7337_241f Content-Type: text/html; charset="utf-8" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline <div id=3D=22edo-message=22><div><div style=3D=22font-size:16px=22>Sphinx= has not changed here, I believe the loop order in the writing routine da= tes back from 2002.</div><div style=3D=22font-size:16px=22>I suspect rath= er that the pyiron parser has never has been validated in this point.</di= v><div style=3D=22font-size:16px=22>But please do so this time, Ahmed=21<= /div><div style=3D=22font-size:16px=22><br></div></div></div> --EdoMail64fb3410_7a6d7337_241f--

[pmrv]

Ninja comment, but it sounds like this can be more readably solved by `arr.reshape(-1, 2)` or something related. /smokebomb From: freyso ***@***.***> To: pyiron/pyiron_atomistics ***@***.***> Cc: Subscribed ***@***.***> Sent: 9/8/2023 12:00 PM Subject: Re: [pyiron/pyiron_atomistics] SPHINX eigen values parser (PR #1159) @freyso commented on this pull request. ---------------- In pyiron_atomistics/sphinx/base.py: > @@ -2286,6 +2286,7 @@ def _parse_band(self, term): arr = np.loadtxt(re.findall(term, self.log_main, re.MULTILINE)) shape = (-1, len(self.k_points), arr.shape[-1]) if self.spin_enabled: + arr = np.vstack((arr[::2], arr[1::2])) Please add a comment why this is needed. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

…

------------------------------------------------- Stay up to date and follow us on LinkedIn, Twitter and YouTube. Max-Planck-Institut für Eisenforschung GmbH Max-Planck-Straße 1 D-40237 Düsseldorf Handelsregister B 2533 Amtsgericht Düsseldorf Geschäftsführung Prof. Dr. Gerhard Dehm Prof. Dr. Jörg Neugebauer Prof. Dr. Dierk Raabe Dr. Kai de Weldige Ust.-Id.-Nr.: DE 11 93 58 514 Steuernummer: 105 5891 1000 Please consider that invitations and e-mails of our institute are only valid if they end with ***@***.*** If you are not sure of the validity please contact ***@***.*** Bitte beachten Sie, dass Einladungen zu Veranstaltungen und E-Mails aus unserem Haus nur mit der Endung ***@***.*** gültig sind. In Zweifelsfällen wenden Sie sich bitte an ***@***.*** -------------------------------------------------

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.