Module second_order 

Second-order optimizers: Newton’s method (Hessian-based) and L-BFGS (limited- memory quasi-Newton).

Both use curvature information to take better-scaled steps than first-order methods. Newton solves the Newton system with an inner conjugate-gradient loop (the Newton-CG strategy); L-BFGS approximates the inverse Hessian from a short history of gradient/step pairs via the two-loop recursion.

Re-exports

pub use lbfgs::lbfgs;

pub use newton::newton;

Modules

lbfgs

Limited-memory BFGS (L-BFGS) with the two-loop recursion and line search.

newton

Newton’s method with an inner conjugate-gradient solve (Newton-CG).