I am currently working on the dp potential field of a high entropy alloy with six fixed components. Using an initial dataset of 128 initial random crystal cells, each with 128 atoms, a total of over 1300 initial data pieces were processed through ten steps of AIMD. Contains random lattice constants with random configurations such as fcc, bcc, hcp, etc. Throw it into the dpgen and run, but it turns out that no reasonable configuration can be explored. I would like to ask if the initial dataset is too small. If it is a six element alloy with an unknown configuration, how much initial random data do I need to prepare to be more suitable
I am currently working on the dp potential field of a high entropy alloy with six fixed components. Using an initial dataset of 128 initial random crystal cells, each with 128 atoms, a total of over 1300 initial data pieces were processed through ten steps of AIMD. Contains random lattice constants with random configurations such as fcc, bcc, hcp, etc. Throw it into the dpgen and run, but it turns out that no reasonable configuration can be explored. I would like to ask if the initial dataset is too small. If it is a six element alloy with an unknown configuration, how much initial random data do I need to prepare to be more suitable