Fig. 2: GPR-guided search for mechanically strong alloys.
From: Probing multi-dimensional composition spaces in search of strong metallic alloys
a The Gaussian process regression (GPR) search path in the Fe-Ta-W composition space: starting from the last of five simulated compositions selected based on intuition, four white arrows trace the sequence of four subsequent GPR predictions followed by MD simulations, eventually leading to Fe38W62 as the strongest composition. Colors over the composition triangle and the color map at the top show ranges of alloy strength predicted by a GPR surrogate model fitted to all alloy compositions for which MD simulations were performed. b Strengthening, defined as strength minus the V-baseline, as a function of alloy composition predicted by a GPR surrogate model fitted to nine MD simulations of Fe-Ta-W alloys. c Strengthening of Fe-Ta-W alloys predicted by the Maresca-Curtin edge (MCE) model. Color maps at the top of (b) and (c) show the range of predicted strengthening. Black dots mark compositions of maximum predicted strengthening.
