[Wannier] Centre of Wannier functions
李滕辉
lith at zju.edu.cn
Fri Jun 7 15:39:00 CEST 2024
Dear Developers:
Thanks for your reply to my previous inquiry.
After encountering this issue, I manually keep tracking on the WF center to restore the continuity of the indices. However, I am still curious if there is a more permanent solution to address this matter. Because when I review historical data generated by VASP v5.4.1 combining with Wannier90 v2.1.0, I observed that the WF center could remain close to the atom center even when the fraction coordinates exceed 0.5. For instance, the atomic position section in the .win file is:
Begin Projections
Cd:sp3:z=-1,-1,-1,x=1,0,-1
Se:p:z=-1,-1,-1,x=1,0,-1
End Projections
...
begin unit_cell_cart
6.0500002 0.0000000 0.0000000
0.0000000 6.0500002 0.0000000
0.0000000 0.0000000 6.0500002
end unit_cell_cart
begin atoms_cart
Cd 1.5125000 1.5125000 1.5125000
Cd 4.5375001 4.5375001 1.5125000
Cd 4.5375001 1.5125000 4.5375001
Cd 1.5125000 4.5375001 4.5375001
Se 0.0000000 0.0000000 0.0000000
Se 0.0000000 3.0250001 3.0250001
Se 3.0250001 0.0000000 3.0250001
Se 3.0250001 3.0250001 0.0000000
end atoms_cart
...
And the .wout file indicates:
...
Final State
WF centre and spread 1 ( 2.073500, 0.951423, 0.951533 ) 2.87731075
WF centre and spread 2 ( 2.074079, 2.074121, 2.074076 ) 2.87774052
WF centre and spread 3 ( 0.951538, 0.951427, 2.073501 ) 2.87730840
WF centre and spread 4 ( 0.950970, 2.073917, 0.950972 ) 2.87765217
WF centre and spread 5 ( 5.098500, 3.976423, 0.951533 ) 2.87731073
WF centre and spread 6 ( 5.099079, 5.099121, 2.074076 ) 2.87774049
WF centre and spread 7 ( 3.976538, 3.976427, 2.073501 ) 2.87730839
WF centre and spread 8 ( 3.975970, 5.098917, 0.950972 ) 2.87765219
WF centre and spread 9 ( 5.098500, 0.951423, 3.976534 ) 2.87731074
WF centre and spread 10 ( 5.099079, 2.074121, 5.099076 ) 2.87774051
WF centre and spread 11 ( 3.976538, 0.951427, 5.098501 ) 2.87730836
WF centre and spread 12 ( 3.975970, 2.073917, 3.975972 ) 2.87765218
WF centre and spread 13 ( 2.073500, 3.976423, 3.976534 ) 2.87731076
WF centre and spread 14 ( 2.074079, 5.099121, 5.099076 ) 2.87774049
WF centre and spread 15 ( 0.951538, 3.976427, 5.098501 ) 2.87730838
WF centre and spread 16 ( 0.950970, 5.098917, 3.975972 ) 2.87765217
WF centre and spread 17 ( -0.142515, -0.115089, -0.142653 ) 1.86289944
WF centre and spread 18 ( -0.000300, 0.202594, 0.000294 ) 1.87615749
WF centre and spread 19 ( 0.142692, -0.088809, 0.142242 ) 1.86859312
WF centre and spread 20 ( -0.142515, 2.909912, 2.882347 ) 1.86289944
WF centre and spread 21 ( -0.000300, 3.227594, 3.025294 ) 1.87615749
WF centre and spread 22 ( 0.142692, 2.936191, 3.167242 ) 1.86859313
WF centre and spread 23 ( 2.882485, -0.115089, 2.882347 ) 1.86289944
WF centre and spread 24 ( 3.024700, 0.202594, 3.025294 ) 1.87615749
WF centre and spread 25 ( 3.167693, -0.088809, 3.167242 ) 1.86859313
WF centre and spread 26 ( 2.882485, 2.909912, -0.142653 ) 1.86289944
WF centre and spread 27 ( 3.024700, 3.227594, 0.000294 ) 1.87615749
WF centre and spread 28 ( 3.167693, 2.936191, 0.142242 ) 1.86859312
Sum of centres and spreads ( 66.549857, 66.548337, 66.549865 ) 68.47064743
Currently, the center of WF5 is positioned at (5.10, 3.98, 0.95) within the home cell. Given that the previous version maintained the WF center within the home cell and ensured continuity, I am hopeful that the latest version can achieve the same outcome. I am eager to understand if there are any ignored things on my end or if manual intervention is always required in Wannier v3.1.0.
I appreciate your insights on this matter and look forward to your guidance.
Thank you for your attention and assistance.
Best regards,
Tenghui Li
Department of Chemistry, Zhejiang University, China
-----原始邮件-----
发件人:"Giovanni Pizzi" <giovanni.pizzi at epfl.ch>
发送时间:2024-06-07 20:21:18 (星期五)
收件人: 李滕辉 <lith at zju.edu.cn>
抄送: "wannier at lists.quantum-espresso.org" <wannier at lists.quantum-espresso.org>
主题: Re: [Wannier] Centre of Wannier functions
Hi, I don't think there is a very simple solution by just changing some code flags.
My suggestion would be, assuming (as it should be) that you have as small tilmestep in your MD, that you post-process the data and shift at every step all WFs by integer lattice vectors, so as to keep their position as "continuous" as possible (essentially, given a WF center at step i, you look for all possible replicas by shifting by lattice vectors e.g. in a small supercell, and find the closest one to the corresponding WF in MD step (i-1), and consider that as the WF position).
Best,
Giovanni
On 7 Jun 2024, at 10:58, 李滕辉 <lith at zju.edu.cn> wrote:
Dear Wannier90 developers:
I meet an issue while using Wannier90 v3.1.0, together with VASP v6.4.1. I've done some AIMD calculations to generate a continuous trajectory. Then I proceeded with Wannier analysis and subsequently examined the Hamiltonian. Within the input file, I have specified the atom centers, aiming to obtain WFs centered on these atom locations, or at least in close proximity to them.
...
Begin Projections
Cd:sp3:z=-1,-1,-1,x=1,0,-1
Se:p:z=-1,-1,-1,x=1,0,-1
End Projections
...
begin unit_cell_cart
6.0500000 0.0000000 0.0000000
0.0000000 6.0500000 0.0000000
0.0000000 0.0000000 6.0500000
end unit_cell_cart
begin atoms_cart
Cd 1.5178271 1.7957993 1.4240287
Cd 4.4307325 4.4894561 1.7053815
Cd 4.5160066 1.2418392 4.5483831
Cd 1.5803075 4.4594387 4.5691762
Se 0.0050033 0.0332671 5.9680866
Se 6.0433290 3.0636897 3.0562890
Se 2.8807786 0.0724863 3.0164791
Se 3.2493695 3.0420933 5.8999131
end atoms_cart
...
However, I have observer that the WFs are not aligning with the atom centers of the home cell. Instead, they are centered on atoms within a unit cell whose center is located at(0.0, 0.0, 0.0). For instance, since I use sp3 orbitals for Cd, the center of WF5 corresponds to the second Cd atoms, while the atom center positioned at (4.43. 4.49. 1.71) and the WF center at (-1.01, -2.12, 1.14) in the .wout file.
...
Final State
WF centre and spread 1 ( 2.085684, 1.260598, 0.798043 ) 3.47876053
WF centre and spread 2 ( 2.049201, 2.344944, 1.921311 ) 3.44168714
WF centre and spread 3 ( 0.904277, 1.241577, 1.954697 ) 3.43610592
WF centre and spread 4 ( 1.042490, 2.326357, 0.905988 ) 3.58232243
WF centre and spread 5 ( -1.010143, -2.123145, 1.141211 ) 3.49051492
WF centre and spread 6 ( -1.128292, -0.972610, 2.258101 ) 3.52243681
WF centre and spread 7 ( -2.121837, -2.038340, 2.220324 ) 3.46813098
WF centre and spread 8 ( -2.218949, -1.012121, 1.103897 ) 3.51827962
WF centre and spread 9 ( -1.015362, 0.681962, -2.070276 ) 3.60904653
WF centre and spread 10 ( -0.890789, 1.882241, -0.973578 ) 3.63289332
WF centre and spread 11 ( -2.108611, 0.719542, -0.996879 ) 3.46637567
WF centre and spread 12 ( -2.115134, 1.897953, -2.162619 ) 3.66683836
WF centre and spread 13 ( 2.150342, -2.128650, -2.068309 ) 3.51187251
WF centre and spread 14 ( 2.156870, -1.008390, -0.926044 ) 3.53284787
WF centre and spread 15 ( 1.050800, -2.157773, -0.945039 ) 3.46859217
WF centre and spread 16 ( 0.958142, -1.003749, -2.060564 ) 3.59315863
WF centre and spread 17 ( 0.037244, 0.001632, -0.049092 ) 2.50422447
WF centre and spread 18 ( 0.015322, 0.080868, -0.134619 ) 2.21017329
WF centre and spread 19 ( -0.059586, 0.030045, -0.095360 ) 2.28365022
WF centre and spread 20 ( 0.037435, -3.029794, -2.977580 ) 2.46391151
WF centre and spread 21 ( 0.000015, -2.932250, -3.012771 ) 2.31121349
WF centre and spread 22 ( -0.056325, -2.986181, -2.999162 ) 2.38773168
WF centre and spread 23 ( 2.893882, 0.030034, 3.037144 ) 2.38000522
WF centre and spread 24 ( 2.928631, 0.123782, 2.984530 ) 2.38126096
WF centre and spread 25 ( 2.821335, 0.075610, 3.032140 ) 2.43241519
WF centre and spread 26 ( -2.791587, -3.058092, -0.116545 ) 2.45592149
WF centre and spread 27 ( -2.744926, -2.945582, -0.212516 ) 2.32087924
WF centre and spread 28 ( -2.851460, -3.004112, -0.153946 ) 2.45045905
Sum of centres and spreads ( 0.018668,-17.703645, -0.597512 ) 85.00170923
...
While this discrepancy does not present an issue in static calculations, it becomes problematic in MD trajectories. For instance, When an atom moves from (0, 0, 0.49) to (0, 0, 0.51), the center shifts from (0, 0, 0.49) to (0, 0, -0.49), resulting in a discontinuity in the Hamiltonian non diagonal element indices.
I've made attempts to address this issue by adjusting parameters in the .win file, such as setting translate_home_cell = true, or explicitly setting the center of each orbital. But none of them works. It seems that the WF centres are predetermined in ab initial calculations. Therefore, I am reaching out to inquire if there exist specific parameters or solutions to solve this problem.
Thank you for your attention and assistance.
Best regards,
Tenghui Li
Department of Chemistry, Zhejiang University, China
_______________________________________________
Wannier mailing list
Wannier at lists.quantum-espresso.org
https://lists.quantum-espresso.org/mailman/listinfo/wannier
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/wannier/attachments/20240607/8ce87974/attachment-0001.html>
More information about the Wannier
mailing list