Mixing Techniques to Compute Derivatives of semi-numerical models: Application to Magnetic Nano Switch Optimization

Phuong Pham Quang (1, 2) , Benoît Delinchant (2) , C. Llie , E. Slusanschi , Jean-Louis Coulomb (2) , Bertrand Du Peloux (1)

1 Cedrat
2 G2ELab - Laboratoire de Génie Electrique de Grenoble

This paper is about derivatives techniques and their composition for semi-numerical models. Techniques such as symbolic derivation and automatic differentiation are addressed. All techniques are illustrated for the gradient based optimization of a magnetic nano switch.

Domaines: Energie électrique

Phuong Pham Quang, Benoît Delinchant, C. Llie, E. Slusanschi, Jean-Louis Coulomb, et al.. Mixing Techniques to Compute Derivatives of semi-numerical models: Application to Magnetic Nano Switch Optimization. COMPUMAG 2011, Jul 2011, Sydney, Australia. ⟨hal-00610867⟩