James Misaka

In ASE and other scripts using deepmd (even deepmd-kit itself ?) are using OMP to do parallelism, I consider that MPI will be a little hard, but it is meaningful.

Also, Largest Gradient is directly using the value of calculation result, but the proper way is to use the modulo, which is correspondent to real force and virial. This problem...

Tried replace OpenMPI 4.1.5 with MPICH 4.1.2, The error is likely the same: ``` /lustre/home/2201110432/apps/abacus/abacus-3.4.2/source/module_io/parameter_pool.cpp: In function ‘bool ModuleIO::Init(const std::string&, const std::string&, const std::string&)’: /lustre/home/2201110432/apps/abacus/abacus-3.4.2/source/module_io/parameter_pool.cpp:83:1: warning: no return statement in...

@caic99 Could you take a look ?

> Hi @QuantumMisaka , This is a known issue. Please see #2299 > > Edit: Please notice that the core problem here is that ABACUS is built with a different...

In my practice in Fe5C2 (510) surface, this problem does NOT EXIST, the TMAG and AMAG are both 143 - 145 and 154 - 156 no matter I set normal...

@mohanchen Any comments ?

Also, I cannot find any noticing information in stdout and running*.log for restart calculation.

can COHP part be `lcao -> lcao` and `pw -> lcao`? Theoretically, COHP is based by atomic orbital, the work of popular software LOBSTER is based by VASP so the...