SHG error

[Sunxl]

Dear Davide,

I performed additional tests using the fixed version of Lumen.

For the FM system, the problem seems to be fixed. The SHG calculation no longer gives NaN values, and the non-zero (yxx) component obtained with the fixed version of Lumen agrees with the result from Yambo 5.3-master.

However, for the AFM system, the problem is still present. Using the fixed version of Lumen, the non-zero (xxx) component still shows strong positive-negative oscillations between adjacent frequency points, similar to what I observed previously.

Thank you for the reminder. I have added my signature now.

Best,
sunxl

[claudio]

Dear sunxl

we fixed another bug in the code, now should be fine.
I tested on your system with smaller k-points grid (9 6 1) and few bands (71 - 78).
in attachment the plots and inputs.

Please download the bug-fixes again now it should be fine

best
Claudio

[Davide]

Dear all,
let me also clarify that this was not strictly a "code bug", but rather an issue related to the theoretical definition of the Berry phase polarization, which is defined up to a periodic term. The oscillations were due to jumps of one period in the polarization, which so far was not occurring in other systems, and the fix (by Claudio) addressed this.

Best,
D.

[Sunxl]

Dear caludio,

Thank you very much! I will re-download the new fix version for testing.

Best,
sunxl