Wannier center production of AIMD with SCAN functional

[luke419]

Hello,

I'd like to perform liquid water AIMD with SCAN functional for the output of the positions of Wannier centers together with the O and H positions of water molecules during the simulation.
The same work has been done in the paper of "First-Principle Analysis of the IR Stretching Band of Liquid Water" J. Phys. Chem. Lett. 2010, 1, 1398–1402 in which Qbox was used for the Wannier center production of liquid water AIMD.
I found that the example file of liquid water AIMD is given at
viewtopic.php?t=270&sid=605686a32954f42 ... de9394187a

To change the functional to SCAN functional and add the command for Wannier centers, what should I modify and/or add in the input file?
Would you let me know where I can get the Pseudopotential files for SCAN functional?

Thank you for you answer in advance.
Best regards,

Young

[fgygi]

Hello Young,
To replace the PBE functional by the SCAN function you only have to replace set xc PBE by set xc SCAN in the input file.
For the computation of MLWFs, it can be done using the compute_mlwf command. Since it is typically not necessary to compute the MLWFs at every step during an MD simulation, you can choose to do so e.g. only every 10 steps using

Code: Select all

set iter_cmd compute_mlwf
set iter_cmd_period 10

This will compute the MLWFs every 10 steps of the MD simulation. It will also print the total dipole. In water, it is usually sufficient to sample the MLWF positions every 100 (a.u.) (2.42 fs) so if you use a time step of 10 (a.u.) and iter_cmd_period = 10 that should work.
Details about the above variables can be found at http://qboxcode.org/doc/html/usage/variables.html
and the compute_mlwf command is described at http://qboxcode.org/doc/html/usage/comm ... mpute-mlwf
Best,
Francois

[luke419]

Dear Francois,

Many thanks for your answers.

1. It seems that a.u. is the unit of physical quantities in the qbox program.
Can we change it to other unit(e.g. a.u. to fs)?
Otherwise, in order to use 0.5 fs for time step, I may use 0.5*(1 a.u./0.0242 fs) = 20.6611570247934 a.u. for time step.
I think that not a neat decimal number would require a heavier comutation cost and I'd better use a.u. for AIMD. Is it right?

2. May I ask how the total dipole is calculated in qbox in detail?
In some paper(https://doi.org/10.1021/jz100232z) for IR spectrum analysis, the molecular dipole moment of water was estimated by the molecular dipole moment sum of myu_i (i: the index of each water molecule D2O)
which is defined by

myu_i = r_D1_i + r_D2_i + r_O_i -2*sum(c=1~4)(r_Wc_i)

where r is vector and r_D1, r_D2, r_O are the nuclear coordinates and r_Wc is the position of four corresponding Wannier centers.
I suppose that the method of total dipole in qbox would be different from the above equation for liquid water AIMD. Will they give the similar result?


Best regards,
Young

[fgygi]

Hello Young,

1. You can use any number to define the dt value without any additional cost.

2. When using the compute_mlwf command, the total dipole is computed by adding the coordinates of all nuclei (with weight Z) and subtracting the coordinates of all the Wannier centers (with weight 2.0, for 2 electrons per orbital). Note that the molecular dipoles cannot be defined unambiguously in the liquid since there is some overlap of the charge density between molecules. Molecular dipoles are however sometimes approximated by summing the coordinates of the nuclei of the molecule (with weight Z) and subtracting the coordinates of the 4 nearest MLWF centers (with weight 2). Note also that the total dipole is only defined up to a multiple of the unit cell size in a periodic system.

The total dipole can also be computed by Qbox using the polarization variable that can be set before a calculation. See the description of this variable in the documentation at http://qboxcode.org/doc/html/usage/vari ... larization .

[luke419]

Thank you for your answers, Francois.

Would you let me know for what reason "the total dipole is only defined up to a multiple of the unit cell size in a periodic system." in detail? How large supercell should be considered (e.g., 3x3x3)?

Best regards,
Young

[fgygi]

Hello Young,
The total polarization of a system is well defined for an isolated molecule or cluster. However in a periodic extended system, the definition is not unique. This is nicely explained in this paper by N. Spaldin https://doi.org/10.1016/j.jssc.2012.05.010
Best,
Francois