csnlp.core.derivatives#

A collection of two methods for computing higher-order sensitivities (i.e., Jacobian and Hessian) w.r.t. CasADi symbolic variables. Natively, CasADi does not support jacobian or hessian for matrices (or at least, they will be flattened). These “higher-order” functions allows to compute the jacobian and hessian of a matrix w.r.t. another matrix.

Functions

hohessian(ex, x[, y])

Computes hessian on higher-order matrices, similar to hojacobian.

hojacobian(ex, x)

Computes jacobian on higher-order matrices, not just vectors.