lmp run error: what(): expected scalar type Float but found Double

[vantasyoscar]

I've compile lammps successfully, but encountered some issues when processing the input files:

units real atom_style atomic newton off thermo 1 read_data MgS.lmp pair_style nequip pair_coeff * * model_e8.pth Mg S mass 1 24.305 # Mg mass 2 32.06 # S neighbor 1.0 bin neigh_modify delay 5 every 1 velocity all create 300.0 23456789 fix 1 all nvt temp 300.0 300.0 0.5 thermo_style custom step pe ke etotal temp press vol thermo 100 dump 1 all custom 100 MgS.dump id type x y z dump 2 all atom 100 MgS.lammpstrj run 20000 write_data nvt1

faced with:

terminate called after throwing an instance of 'c10::Error' what(): expected scalar type Float but found Double

I’m not sure how to resolve this problem.

[vantasyoscar]

I solve this by setting default_dtype:float32 . Does it mean I must set float32 precision if I want to deploy it on lammps?

[Linux-cpp-lisp]

Should be resolved by #52: you should be able to use the defaults of nequip 0.6.0 (default_dtype: float64, model_dtype: float32 with the main branch of pair_nequip). We recommend those defaults, and do not recommend setting default_dtype:float32.