A simple, quantitative expression for understanding and evaluating the yield strength of amorphous alloys based on symbolic regression and dimensional calculation

High strength can be the most important and outstanding mechanical property of amorphous alloys as structural materials. To improve the limitations of current strength models, elucidate the key factors determining the yield strength ( σ _y) of amorphous alloys and formulate an explicit, quantitative expression of σ _y, herein, we successfully developed an ML-based strength model by employing symbolic regression and dimensional calculation. 10 key descriptors were initially extracted from 22 candidate descriptors using random forest model and SHAP method, and dimensional calculation was simultaneously carried out with numerical calculation to ensure the dimensional consistency of final expression. The results shown that the SR-DC model was superior to SR-GP and SR-BGP models, with its satisfactory combination of accuracy ( R ² =0.93) and simplicity. Meanwhile, molar volume ( V _m), melting point ( T _m ), Pauling electronegativity ( χ _p), and valence electron density ( VED ) were identified as the key factors and were used to explicitly express the yield strength of amorphous alloys. The final expression was well verified by further theoretical and experimental verification. This work not only provided a general benchmark for the composition design and development of high-strength amorphous alloys but also demonstrated the great potential of symbolic regression combined with dimensional calculation to develop quantitative composition-structure-property relationship in future designing novel materials.