Efficient quantum algorithm for the design of complex materials: quantum circuit learning.

Area of Science:

• Materials Science
• Quantum Computing
• Machine Learning

Background:

• Predicting physical properties of complex materials like high entropy alloys is crucial for materials design.
• Conventional machine learning methods have limitations in accuracy and data requirements for complex material property prediction.

Purpose of the Study:

• To evaluate the validity of the quantum circuit learning (QCL) method for predicting Vickers hardness in high entropy alloys.
• To compare QCL's predictive performance against various conventional linear and nonlinear machine learning models.

Main Methods:

• Quantum Circuit Learning (QCL) model development and application.
• Implementation and comparison with linear models: linear regression (LR), ridge regression (Ridge), Bayesian ridge regression (BR), linear support vector regression (linear_SVR).
• Implementation and comparison with nonlinear models: Gaussian process regression (GBR), radial basis function kernel support vector regression (rbf_SVR), random forest (RF), gradient boosting decision trees (GBDT), multilayer perceptron neural networks (MLP).

Main Results:

• QCL demonstrated efficient prediction of Vickers hardness for high entropy alloys.
• QCL showed strong performance even for outside applicability domains and extrapolations.
• QCL achieved accurate predictions with a small number of datasets, outperforming conventional methods.

Conclusions:

• Quantum circuit learning is a valid and efficient method for predicting the Vickers hardness of complex materials.
• QCL's ability to perform well with limited data makes it suitable for early-stage materials design.
• The findings suggest QCL can accelerate the discovery and development of new complex materials.