Table 1.

Locally optimal compositions and the number of compositions needed to identify them for the two quinary HEAs.

HEA	Local optimum[a]	Predicted current density (arb. units)[b]	Idenfication success rate [%][c]	Number of samples for identification of local optimum[b,d]
Ag‐Ir‐Pd‐Pt‐Ru	Ag18Pd82	−0.203(2)	100	50(21)
	Ir9Pd64Ru27	−0.160(2)	100	28(28)
	Ir48Pt52	−0.147(2)	100	25(10)
	Ag78Ru22	−0.063(3)	69	93(27)
	Ir46Ru54	−0.003(0)	2	73(20)
	Ir10Ru90	−0.002(1)	14	110(27)
	Ru	−0.001(1)	14	48(33)
Ir‐Pd‐Pt‐Rh‐Ru	Ir42Pt58	−0.165(2)	100	23(8)
	Ir12Pd56Rh4Ru28	−0.164(1)	100	19(10)
	Rh	−0.001(2)	27	48(42)

[a] Determined as the local optima of the resulting surrogate function after sampling of 150 compositions for 64 random realizations of the two initial compositions (one such realization is shown in Figure 2). The spread in these compositions is on the order of 1 at. %. [b] Given as the mean followed by the sample standard deviation on the last digit(s) in parentheses. [c] Determined as the proportion of the resulting surrogate functions after sampling of 150 compositions for 64 random initializations that identify the optimum as a local maximum. [d] Determined as the number of samples needed for those of 64 surrogate functions with random initializations that successfully identified the optimum. The optimum has been considered identified when the molar fraction is within a 10 at. % difference of the optimum, for example, Ag₂₃Pd₇₇ would be regarded as a successful discovery of the Ag₁₈Pd₈₂ optimum.