# [Wannier] Trouble getting symmetry adapted Wannier functions to converge

Hello Wannier90 Community,

I am trying to compute site symmetric ab initio Tight-Binding models using the
site_symmetry=TRUE flag in <seedname>.win. I have followed the online tutorials
(2020_03_Oxford/2c_symmetry_adapted) and am able to replicate results for H3S,
GaAs and extend the results to the zincblende phase of SiC using an 8 band sp3
initial projection. However, when I try to modify the system even a little, I
get either strange results or error messages.

The example systems I consider here are for a 2 atom unit cell zincblende SiC,
a 4 atom unit cell (1x1x2 supecell) zincblende SiC, and a 4 atom unit cell 2H
SiC.

Since the 2 atom zincblende converges as expected, I had assumed the 4 atom
doubled unit cell would produce similar results and converge, since it is
merely zone folded in k-space. This system does converge, but gives Wannier
function centers that are not atom centered. I suspect this has something to do
with QE not identifying glide/screw g={t|R} symmetries in systems with pure
fractional translation g={t|1} (i.e., supercells). This can be noted in the QE
output, where the 2 atom ZB unit cell has 24 symmetries, whereas the 4 atom ZB
unit cell has 6.

The case for 2H (wurtzite) is even worse. I cannot get the system to converge
for an sp3 projection, with the error suggesting something wrong with the
initial projection. I find this strange since the orbital character of the
states for all three systems specified should be nearly identical.

Attached are the shell scripts that can be run to reproduce the calculations
outlined.

Summary of results
###########################################################
Example for 3C SiC

(Note: this example worked as expected)

Final State
WF centre and spread    1  (  0.000000,  0.000000,  0.000000 )
1.57823570
WF centre and spread    2  (  0.000000,  0.000000,  0.000000 )
2.57469067
WF centre and spread    3  (  0.000000,  0.000000, -0.000001 )
2.57469307
WF centre and spread    4  ( -0.000000,  0.000002, -0.000000 )
2.57469073
WF centre and spread    5  (  1.095076,  1.095076,  1.095076 )
0.90201541
WF centre and spread    6  (  1.095076,  1.095076,  1.095076 )
1.61900198
WF centre and spread    7  (  1.095076,  1.095076,  1.095076 )
1.61900244
WF centre and spread    8  (  1.095076,  1.095076,  1.095076 )
1.61900196
Sum of centres and spreads (  4.380305,  4.380306,  4.380303 )
15.06133196

where the coordinates match those specified in the projection block
and should be the positions of atoms in the unit cell...
(Viewable in many places, but most easily near the top of *.wout)

*----------------------------------------------------------------------------*
|   Site       Fractional Coordinate          Cartesian Coordinate
(Ang)     |

+----------------------------------------------------------------------------+
| Si   1  -0.00000  -0.00000  -0.00000   |   -0.00000  -0.00000
-0.00000    |
| C    1   0.25000   0.25000   0.25000   |    1.09508   1.09508
1.09508    |

*----------------------------------------------------------------------------*

Example for 1x1x2 supercell of 3C-SiC (4 atom unit cell for ZB made by

(Note: This example "ran" and "converged" but the results are
non-sense)

Final State
WF centre and spread    1  (  0.483907,  0.483907, -0.483907 )
2.68709962
WF centre and spread    2  (  0.564237, -0.379955,  0.379955 )
2.48775520
WF centre and spread    3  ( -0.379955,  0.564237,  0.379955 )
2.48775525
WF centre and spread    4  ( -0.379955, -0.379955, -0.564238 )
2.48775592
WF centre and spread    5  (  0.817993,  0.817993,  1.372159 )
1.16964635
WF centre and spread    6  (  1.468481,  1.101956,  1.088195 )
1.60161090
WF centre and spread    7  (  1.101956,  1.468481,  1.088195 )
1.60161060
WF centre and spread    8  (  1.101956,  1.101956,  0.721671 )
1.60161275
WF centre and spread    9  (  2.674063,  2.674064, -0.483912 )
2.68716225
WF centre and spread   10  (  2.754390,  1.810196,  0.379955 )
2.48780836
WF centre and spread   11  (  1.810196,  2.754390,  0.379956 )
2.48780761
WF centre and spread   12  (  1.810196,  1.810196, -0.564238 )
2.48780878
WF centre and spread   13  (  3.008153,  3.008152,  1.372150 )
1.16977422
WF centre and spread   14  (  3.658659,  3.292089,  1.088214 )
1.60157323
WF centre and spread   15  (  3.292089,  3.658659,  1.088214 )
1.60157287
WF centre and spread   16  (  3.292090,  3.292089,  0.721644 )
1.60157170
Sum of centres and spreads ( 27.078454, 27.078454,  7.963969 )
32.24992563

*----------------------------------------------------------------------------*
|   Site       Fractional Coordinate          Cartesian Coordinate
(Ang)     |

+----------------------------------------------------------------------------+
| Si   1   0.00000   0.00000   0.00000   |    0.00000   0.00000
0.00000    |
| Si   2   0.00000   0.00000   0.50000   |    2.19015   2.19015
0.00000    |
| C    1   0.25000   0.25000   0.12500   |    1.09508   1.09508
1.09508    |
| C    2   0.25000   0.25000   0.62500   |    3.28523   3.28523
1.09508    |

*----------------------------------------------------------------------------*

Example for 2H-SiC  (4 atom primitve wurtzite unit cell for SiC)

(Note: this example did not run... specifically the final step with
wannier90.x gives a convergence error and terminates)

Error in symmetrize_u: not converged
Either eps is too small or specified irreps is not
compatible with the bands
diff,eps=    0.2910700672E+02    0.1000000000E-06
Exiting.......
symmetrize_ukirr: not converged

*----------------------------------------------------------------------------*
|   Site       Fractional Coordinate          Cartesian Coordinate
(Ang)     |

+----------------------------------------------------------------------------+
| Si   1   0.00000   0.00000   0.00000   |    0.00000   0.00000
0.00000    |
| Si   2   0.33333  -0.33333   0.50000   |    0.00000   1.78692
2.53571    |
| C    1   0.00000   0.00000   0.37677   |    0.00000   0.00000
1.91076    |
| C    2   0.33333  -0.33333   0.87677   |    0.00000   1.78692
4.44647    |

*----------------------------------------------------------------------------*

I am looking for advice on how/why I am getting error messages and how to get
quality converged results when using site_symmetry=TRUE in wannier90 (combined
with quantum espresso (QE) and pw2wannier90's flag write_dmn=True).

Thanks!

Joseph Sink, PhD
Postdoctoral Fellow
University of Iowa, Iowa City, Iowa
Department of Physics and Astronomy

2H_SiC_run.shell
Description: 2H_SiC_run.shell

3C_doubled_SiC_run.sh
Description: 3C_doubled_SiC_run.sh

3C_SiC_run.shell
Description: 3C_SiC_run.shell

_______________________________________________
Wannier mailing list
Wannier@lists.quantum-espresso.org
https://lists.quantum-espresso.org/mailman/listinfo/wannier