Table 3 Model performance on the water dataset

From: Cartesian atomic moment machine learning interatomic potentials

 

BPNN39

ACE35

REANN48

DeePMD9

NequIP17

MACE16

CACE38

CAMP

Energy

2.3

1.7

0.8

2.1

0.94

0.63

0.59

0.59 (0.002)

Forces

120

99

53

92

45

36

47

34 (0.1)

Speed

   

61.8

3.9

 

9.8

14.2

  1. RMSEs of energy and forces are in the units of meV/atom and meV/Ã…, respectively. Speed is measured as completed steps per second in MD simulations of the water system of 192 atoms, using a single NVIDIA V100 GPU. All energy and forces RMSE results, except CAMP, are from ref. 38. DeePMD and NequIP running speed results are from ref. 47. Errors shown in parentheses are obtained as the standard deviations from five independently trained CAMP models.
Back to article page