small restructuring of the routines for the AHC calculation with multiple Fermi energies and adaptive refinement

[hjunlee]

Dear all:

This commit modifies the routines in berry.F90 so as to remove unnecessary multiple calls of berry_get_imf_klist in the berry module for calculations of AHC with adaptive refinement turned on and multiple Fermi energies (nfermi>1).
The current structure results in the duplication of some calculations unnecessarily.

With this commit, we can achieve much speed-up, in particular, in the case of large nfermi and rather dense adaptive-refinement k-mesh.

I made an effort to change the current structure of the berry module as little as possible even though I can achieve more speed-up with more restructuring.

Hyungjun Lee

[codecov]

Codecov Report

Merging #289 into develop will increase coverage by 0.17%.
The diff coverage is n/a.

@@             Coverage Diff             @@
##           develop     #289      +/-   ##
===========================================
+ Coverage    65.60%   65.78%   +0.17%     
===========================================
  Files           29       29              
  Lines        17939    17957      +18     
===========================================
+ Hits         11769    11813      +44     
+ Misses        6170     6144      -26     

Impacted Files	Coverage Δ
src/parameters.F90	82.94% <0.00%> (+0.08%)	⬆️
src/postw90/berry.F90	76.27% <0.00%> (+2.84%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d0d5e4f...35a1c51. Read the comment docs.

[hjunlee]

I don't understand why this minimal modifications lead to the failure in the check of codecov/patch.
How can I pass this check?

[giovannipizzi]

Don't worry, those tests are not critical.
Anyway, the reason is that more than 50% of the lines you added are untested.
Would you mind adding a test that goes through the new code path? Thanks!

[giovannipizzi]

Hi @hjunlee - since we are planning to release v3.1 in about 1 week, if you would like to see this merged, it would be great if you could add the tests.
Do you think you would have the time to do it?
Here are the instructions if you never did it. If you have questions just let us know.
Thanks in advance!

[hjunlee]

Dear Giovanni:

I added the test for calculation of anomalous Hall conductivity for bcc Fe with adaptive mesh refinement and Fermi energy scan.

Benchmark data comes from the calculation using the most recent develop version without my modification. Since my modification just removes unnecessary multiple calls, the results should be same and this is confirmed.

Sincerely,

Hyungjun Lee

[giovannipizzi]

Thanks a lot!