XAD

Emphasizes low runtime overhead and minimal memory footprint, making it ideal for large-scale systems like financial Monte Carlo simulations, as stated in the philosophy section.

Supports forward, reverse (adjoint), and vector modes via operator overloading, enabling efficient derivative computation for complex models with any order derivatives, per the key features.

Works seamlessly with Eigen for linear algebra and offers external function interfaces, facilitating integration into existing numerical codebases without major rewrites.

Includes checkpointing for efficient tape memory management, crucial for handling large-scale applications without excessive memory usage, as highlighted in the key features.