Table 1 Summary of the contents of each MLP training set

From: Capturing short-range order in high-entropy alloys with machine learning potentials

Name

Chemical order

Phase fraction (%)

Total size (atoms)

 

RSS

SRO

fcc

hcp

liquid

 

TS-0a

Yes

No

68

0

32

83,970

TS-1

Yes

No

100

0

0

54,540

TS-2

Yes

Yes

100

0

0

54,540

TS-3b

Yes

Yes

50

0

50

108,684

TS-4

Yes

Yes

100

0

0

36,720

TS-5

Yes

Yes

50

50

0

73,440

TS-f

Yes

Yes

25

25

50

146,880

  1. If the data set includes at least one random solid solution (RSS) configuration, it is marked as “yes” in that column. The chemical order “SRO” column indicates that configurations extracted from DFT Monte Carlo were included. The “total size” of each training set refers to the total number of local atomic environments included. The training set for TS-0 was motivated by ref. 36. A more detailed table of each training set is included in Supplementary Section 4.
  2. aSame training set as “without chemical sampling” in ref. 23.
  3. bSame training set as “with chemical sampling” in ref. 23.
Back to article page