dis_spheres_first_wann larger error

[al7]

Hi,
Whenever I try running my calculation, it gets stuck with the following error message until timeout in the cluster

Wannierization on 6 x 6 x 6 electronic grid
-------------------------------------------------------------------
Spin CASE ( default = unpolarized )
Initializing Wannier90

Error: dis_spheres_first_wann is larger than num_bands-num_wann+1
Error: examine the output/error file for details

The errror message in the .err file is the following (I believe this could be ignored?): "Note: The following floating-point exceptions are signalling: IEEE_DENORMAL"

I saw someone else also asked this question a few years ago and the solution was to use a newer version of QE and EPW. viewtopic.php?t=235

My EPW version is 5.3.1 and QE version is 6.6

[hlee]

Dear al7:

Could you show me your epw.in, epw.out and generated *.win and *.wout files?

Sincerely,

H. Lee

[al7]

EPW.IN files

epw_in_1.PNG (36.6 KiB) Viewed 4847 times

epw_in_2.PNG (29.28 KiB) Viewed 4847 times

epw_in_3.PNG (10.05 KiB) Viewed 4847 times

[al7]

EPW.OUT

epw_out1.PNG (39.14 KiB) Viewed 4847 times

epw_out2.PNG (48.1 KiB) Viewed 4847 times

epw_out5.PNG (45.19 KiB) Viewed 4847 times

[al7]

.WIN file

win.PNG (12.61 KiB) Viewed 4847 times

.WOUT file

wout.PNG (59.63 KiB) Viewed 4847 times

[hlee]

Dear al7:

First, I assume that you are performing spin-unpolarized calculation.

I think that your chosen projections are not consistent with the number of Wannier functions.

(from your *.win file)

begin projections
S:sp
end projections
num_wann = 18

Do you really want to use the sp hybrid orbital for the projection? In this case, you have two Wannier functions (WFs) per S atom, leading to 4 WFs in total since your system has two S atoms.

But, I guess that you want to use the projections with s- and p-orbital characters.
In this case,
(1) you need to change the line of "S:sp" with "S:s;p"
(2) you need to change the line of "num_wann = 18" with "num_wann = 8" since the total number of Wannier functions is 2 * (1+3) = 8.

Sincerely,

H. Lee