Dear vampire developers,
During test on γ-IrMn3 (disordered IrMn3), I've noticed unexpected behavior in the code
and would appreciate your help in resolving the issue.
ToolChain
RHEL8, Intel OneAPI2023.0
VAMPIRE 7.0.0 (develop branch, githash 5a859afd4f9a251fcd5bf768c829ffbaf4ec6c52)
γ-IrMn3 field cooling simulation from
Input1: vampire internal tag system generation.
Input2: UCF system genration. UCF contains all 1NN and 2NN interactions.
not equivalent
(*input1 well reproduces γ-IrMn3)
Trun-off Néel anisotropy and Ir atom alloy-fraction=0.25
Input3: vampire internal tag system generation.
Input4: UCF system genration.
not equivalent
→ Mismatch is not attributed to Neel anisotropy tag
Trun-off Néel anisotropy and Ir atom alloy-fraction=0
Input5: vampire internal tag system generation.
Input6: UCF system genration.
equivalent
→ Mismatch is not attributed to UCF exchange input.
Trun-off Néel anisotropy and Ir atom alloy-fraction=0.25
non-magnetic = remove
Input7: vampire internal tag system generation.
Input8: UCF system genration.
equivalent
Trun-off Néel anisotropy and Ir atom alloy-fraction=0.50
non-magnetic = keep
Input7: vampire internal tag system generation.
Input8: UCF system genration.
significant mismatch found (see magentization length vs. temperature)
Since input8 with non-magnetic = remove becomes input7 results, the results indicate exchange interations for (non-magnetic atom)-(magnetic atom) are not removed. Therefore I conclude this behavior as a code bug. Please review this post.
I attached input for 7 and 8. 1~6 are readily created with this input.
Sincerely,
Sungyoung Ahn
Department of Materials Science and Engineering
Gwangju Institute of Science and Technology (GIST)