| Abstract Scope |
Complex materials science problems, such as plastic behavior of microstructures, glass formation, single phase solid solution formation, and alloys’ liquidus temperature, originate from a large number of atoms.
To quantify the problem, we predict the liquidus temperature of binary alloys. Our results of ~8% are comparable with previous work, which is still unsatisfactory for these ML models to be of practical use.
Our analysis reveals two major challenges in predicting complex problems through supervised ML algorithms. One challenge is representing the relevant characteristics of an alloy that determines liquidus temperature (or glass forming ability) through appropriate features.
The other fundamental challenge is the discreteness of atoms properties.
We argue that these challenges are common in complex materials science problems and constitute a fundamental challenge in applying supervised ML strategies in this context. |