[QE-users] How to decide value of TDDFPT variables

Oleksandr Motornyi oleksandr.motornyi at polytechnique.edu
Mon Dec 3 12:08:28 CET 2018


Dear M. Inukai,


I would like to add a small note to the very comprehensive response of 
Iurii and Oscar regarding the "pseudo_hermitian" option:


In my experience if the code fails with this option set to .TRUE. it 
might mean that there are some problems with your system or parameters ( 
i.e. too low kinetic energy cutoff with respect to the amount of Lanczos 
recursion iterations you want).


Best regards,

Oleksandr Motornyi


On 12/03/2018 11:11 AM, Timrov Iurii wrote:
> Dear M. Inukai,
>
> 1. num_eign = number of eigenstates (excitations) which will be
> computed. This parameter must be chosen based on how large energy range
> for the spectrum the user is interested in. The bigger the energy range,
> the larger num_eign is. A priori it is not possible to guess this
> number, because it depends on the system. Try with some reasonable
> numbers (say, 5, 10, or 15) and then check whether this is enough to
> obtain a converged spectrum in the energy range of interest.
>
> 2. 'num_basis_max' is the maximum number of basis allowed in the subspace.
> In  every Davidson iteration the number of basis used in the subspace 
> is increased
> by num_init. When num_basis_max is reached, a discharging
> routine is called (this variable determines the memory requirement). The
> larger the parameter num_eign, the larger num_basis_max would be needed.
> Try with some reasonable numbers, run test calculations, and then
> increase/decrease num_basis_max based on your observations. Use examples
> in TDDFPT to get an idea about typical values of this parameter.
>
> 3. 'ecutfock' is the kinetic energy cutoff (Ry) for the exact exchange 
> operator
> in the exact-exchange (EXX) type calculations. The meaning of this 
> parameter
> is the same as in the PW code (pw.x), hence see the documentation in 
> PW/Doc.
> The value of ecutfock cannot be smaller than ecutwfc.
>
> 4. 'itermax0' generally you can use the same number that you computed
> with turbo_laczos (itermax).
>
> 5. 'p_nbnd_occ and p_nbnd_virt' if you are interested only in energy and
> absorption spectrum, it is not necessary to use all occupied and virtual
> orbitals. Using all states is useful only if you are interested in the
> interpretation of the eigenstates.
>
> 6. 'pseudo_hermitian' When you use .FALSE., the code will be 2 times
> slower but the result must be the same. So it is always recommended to
> use .TRUE. unless it does not work for some reasons. The .FALSE. option
> was implemented first, and hence it is still kept in the code for the
> case when .TRUE. does not work.
>
> 7. 'CH4-BOND' this is obsolete and will be removed in the next release 
> of QE.
> The meaning of "project", "absorbtive", and "dispersive" is not 
> documented,
> it was introduced by Osman Baris Malcioglu around ~2011 (CPC 182, 1744 
> (2011)).
> These parameters are related to the calculation of the response charge 
> density.
> You may try either to understand what does these keywords mean by 
> looking into the code or
> try to contact Baris and ask him directly.
>
> Best regards,
> Oscar Baseggio - postdoctoral researcher, SISSA, Trieste, Italy
> Iurii Timrov - postdoctoral researcher, EPFL STI IMX THEOS, Lausanne, 
> Switzerland
>
> ------------------------------------------------------------------------
> *From:* users <users-bounces at lists.quantum-espresso.org> on behalf of 
> 犬飼 学 <inukai.manabu at nitech.ac.jp>
> *Sent:* Monday, November 26, 2018 2:17:37 AM
> *To:* users at lists.quantum-espresso.org
> *Subject:* [QE-users] How to decide value of TDDFPT variables
> Dear developers
>
> Although I have red *.html in Doc (TDDFPT),
> I do not understand same variables in detail.
>
> Would you tell me any advices ?
> ----------
> How do I decide 'num_eign', 'num_basis_max' and 'ecutfock' in *.dav-in ?
>
> How do I decide 'itermax0' in *.tddfpt_pp-in ?
>
> I set number of eigenvalue (<= EF) to p_nbnd_occ and
> set number of eigenvalue (>= EF) to p_nbnd_virt ?
> Is it best ?
>
> What do absorbtive and dispersive mean in CH4-BOND ?
>
> What does 'project' mean in CH4-BOND ?
>
> When do I use pseudo_hermitian=.false.
> --------
> Best Regards,
> M. Inukai
>
> _______________________________________________
> users mailing list
> users at lists.quantum-espresso.org
> https://secure-web.cisco.com/19GAnZJq8KAnok_kIcfsPVX2cFnQjBcHHDJ1Me3k4GB8ARlEdR_PSsXj3f6NZyAT6Br-Z1c4M6lziWmaSbgvcxL9M3XKiEgWPD1igTAi13jQZbMm5p4u0EnzwrDrY2ibWKUX5hVu18-Ts5EFmwEzzvC9nDnsbHc6YjDo5Et3aY88Vevq_-ssnwzwbgNpjHU46HYluK2AHThecFDzDpc-oJNgB2jpBPVFQO4FvpnndUJhPwEMPLCBswmGae7CEu_hB9sPsX1b11-Z8fjViDo9S1I8reS_hReQeZPxnuAbWbltmhNnGRJgokkgoaER3T_C8MQIDwENW0HZ85yYfRfg9jg/https%3A%2F%2Flists.quantum-espresso.org%2Fmailman%2Flistinfo%2Fusers
>
>
>
> _______________________________________________
> users mailing list
> users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users

-- 
Oleksandr Motornyi
PhD candidate

Laboratoire des Solides Irradies
Ecole Polytechnique (Palaiseau, France)

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20181203/05ab5069/attachment.html>


More information about the users mailing list