Séminaires doctorant à venir

Algorithmic differentiation (AD) is a technique to evaluate the (exact) derivative of computer programs automatically. In the first part of this talk, I will introduce AD and its two modes of operation: forward mode and reverse mode. I will also discuss the differentiation of iterative algorithms. In the second part of this talk, I will present the application of AD techniques in plane-wave density functional theory, one of the most widely employed simulation approaches for modeling materials with an accurate quantum-mechanical description of electrons.