Machine Learning Prediction of Field-Free Spin-Orbit-Torque Switching

data

1069 micromagnetic simulation are performed with different disk diameter, interfacial DMI strength D and the charge current. For regression, the target is the z component of final average magentization. For classification, the target is 1 (<m_z> <-0.1), 2 (-0.1 < <m_z> < 0.1) or 3 (<m_z> > 0.1)

code

The datasets are divided into 989 training groups and testing groups. Differnt regression methods and classification methods are perfomed using Scikit-learn. The accuracy of regression is represented by R² score and the dicision tree regression model delivers the highest accuracy.