Hear Mihal,
From my experience with CASTEP, it's defining difference are the
pseudopotentials.
Thus, if you want to replicate a CASTEP calculation - use the CASTEP
UPFs as well.
Even on-the-fly runs create UPFs in the project folder (at least via
Materials Studio).
However, I am not a fan of that potentials, and I have my reasons.
I was doing a fairly similar calculation with both CASTEP and QE (manual
finite displacement force estimation).
With CASTEP, I was getting some weird differences in energies where
there should have been none.
With QE, there was no such problem.
Just for a general impression, compare default cutoffs for lanthanides
with 4f as valence electrons:
in CASTEP, its 340 eV; in PSLibrary, it's about 1500 eV.
So, either CASTEP authors found a way to reproduce the fine structure of
4f orbitals with fewer plane waves, or their potentials are, well, not
quite good.
I am not trying to hurt anyone, but this particular aspect is
suspicious.
Not sure if that transfers to other elements as well.
Best regards.
Andrii Shyichuk, University of Wrocław.
W dniu 2020-11-14 20:51, Husak Michal napisał:
Dear Nicola Marzari
I had already done everything you suggest + more:
1) Converted my structures by the Material Cloud convertor
2) Converted my structures by BURAI GUI
3)Semi/manually converted my structures from CIF to QE by the help of
QE
internal Space Group Handling
4) Read multiple articles optimizing QE parameters exactly for the
same problem I investigate (organic molecular crystals)
+ tested the parameters they use (e.g.
https://pubs.acs.org/doi/10.1021/cg4002797 )
I know how to process my data in QE ..
I search for maximally exact CASTEP/QE calculations setup replication,
to be able to check performance/result of both codes ...
I had already made huge number of calculations in CASTEP optimizing
the setup/performance ..
So I ask how to set QE identically as CASTEP as much as possible ...
________________________________
From: Nicola Marzari <[email protected]>
Sent: Saturday, November 14, 2020 8:22:48 PM
To: Quantum ESPRESSO users Forum; Husak Michal
Subject: Re: [QE-users] CASTEP to QE setup parameters conversions
One possibility would be to upload your structure (or in principle even
your CASTEP .cell file) here:
https://www.materialscloud.org/work/tools/qeinputgenerator
This will give you fairly reliable input parameters (cutoff, sampling,
etc...) and pseudopotentials for a scf calculation,
and then try to understand from this document
https://www.quantum-espresso.org/Doc/INPUT_PW.html
what else you need
nicola
On 14/11/2020 20:13, Husak Michal wrote:
Hi all
I was working a lot with CASTEP last years ...
I want to switch some my projects to QE + do head to head comparison
with CASTEP.
I am confused by some selected corresponding setup parameters
conversion
(I had handled the rest including input formats conversion myself):
QE: etot_conv_thr (a.u.]
CASTEP: geom_energy_tol (eV/atom)
Should I simply recalculate eV to a.u. and multiply by number of
atoms in unit cell ?
Why the definition in QE/CASTEP differs ?
QE: forc_conv_thr (a.u.)
CASTEP: geom_force_tol (eV/angstrem)
I have no idea how to convert this parameter ....
QE: non existing criteria
CASTEP: geom_disp_tol (angstrem) , Maximal atom displacement during a
cycle ..
Sound like QE does not check this criteria ?
QE: ecutrho (Ry), default 4 * ecutwfc
CASTEP: non existing parameter
Any idea why CASTEP does not use this parameter ?
QE: conv_thr , units not found/missing in manual, probably (a.u.) ?
CASTEP: elec_energy_tol (eV/atom)
Should I simply recalculate eV to a.u. and multiply by number of
atoms in unit cell ?
Why the definition differs ?
Michal Husak
UCT Prague, Department of Solid State Chemistry
_______________________________________________
Quantum ESPRESSO is supported by MaX
(www.max-centre.eu<http://www.max-centre.eu>)
users mailing list [email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
users mailing list [email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
users mailing list [email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users
