MAPAL: A python library for mapping features and properties of alloys over compositional spaces

Compositional machine learning (ML) models have emerged as a promising high throughput approach to probe the properties and behavior of a wide variety of materials including multi-principal element alloys (MPEAs). These models use physical and thermodynamic features that are derived from some combin...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Computational materials science Ročník 262; s. 114360
Hlavní autori: Beniwal, Dishant, Ray, Pratik K.
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Elsevier B.V 30.01.2026
Predmet:
Alloy design
Machine learning
MAPAL
python library
MAPAL
python library
Alloy design
Machine learning
ISSN:0927-0256
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Compositional machine learning (ML) models have emerged as a promising high throughput approach to probe the properties and behavior of a wide variety of materials including multi-principal element alloys (MPEAs). These models use physical and thermodynamic features that are derived from some combination of alloy composition and elemental properties. The primary goals behind the development of MAPAL are to enable: (a) easy mapping of alloy features over compositional variations in binary, ternary and MPEAs, (b) integration with machine learning frameworks that require calculation of alloy features, and (c) use of pre-trained machine learning models to map the material behavior/properties over compositional spaces. We show the potential application of MAPAL for targeted design of MPEAs through two case studies on the design of Cantor alloys and Senkov alloys wherein MAPAL was used to obtain reliable preliminary estimates of hardness and phase variations in these alloy systems. MAPAL provides avenues for integration with various types of workflows for MPEAs including ML model training, transfer learning and benchmarking. [Display omitted] •Python library for supporting design of multi-principal element alloys.•Pre-trained models for transfer learning, exploration and benchmarking.•Mapping properties over high-dimensional compositional spaces for exploratory design of alloys.
ISSN:0927-0256
DOI:10.1016/j.commatsci.2025.114360