Yu Liu
Yu Liu
@JiajiaLiu7451 Can you please supply the output file for this example?
You can get the same machine environment as GitHub through docker. @hongriTianqi @pxlxingliang What's your advice?
Currently, I cannot reproduce this bug.
Now this bug is solved in abacus v3.8.0.
Compared to QE, it seems to arise from the unstable process of init_after_vc for uspp. I will fix it later.
> The bug is introduced by the basis `1s1p0d`. If you change the beginning of orbital files to `1s1p`, the code works well. > > I'll fix it later. @YuLiu98...
The original FIRE is not implemented by me. I just reconstructed it during MD reconstruction. So I'm not familiar with the algorithm.
As I have mentioned in #4714, `index` is not the ionic step. `istate.info` is the result of the final ion step.
@lyb9812 Could you explain this question?
You can obtain the ks occupation of TDDFT per MD step from `running_md.log` using the develop branch.