A bug was encountered while calculating magnetic calculations

[matphy1]

Describe the bug

I am calculating La4Ni3O10 system using the abacus v3.9.0 stable branch.
this is a log.out file.
The calculation file is in attachment.
thank you.

8-mag-sdw.zip

---

Expected behavior

I hope this bug can be fixed, thank you.

[lanshuyue]

First, I reproduced your calculation process. The program encountered an error after the SCF calculation completed (without convergence). Regardless of the scf_nmax setting (1/10/200 steps), a segmentation fault occurs after the SCF finishes, which may be a program bug.

Additionally, I noticed that the magnetic moment at the end of the 200 steps of SCF calculation converges to nearly zero, indicating that the system may be non-magnetic. You may consider using nspin=1 for the calculation, which should proceed normally and converge.

[mohanchen]

Thanks for your effort! @lanshuyue

[matphy1]

According to the experimental literature, this is a magnetic system, an incommensurable magnetic structure, so I need magnetic calculations.

thank you!

[dyzheng]

@lanshuyue The newest version https://github.com/deepmodeling/abacus-develop/releases/tag/LTSv3.10.0 fixed this bug.

[dyzheng]

@matphy1 I will test this case, I think mixing_dmr 1 and mixing_restart 1e-3 and smearing_sigma 0.01 would help for convergence.

[mohanchen]

@lanshuyue The newest version https://github.com/deepmodeling/abacus-develop/releases/tag/LTSv3.10.0 fixed this bug.

The fixed PR is: #6024

[mohanchen]

I will close this PR, reopen it if needed.