Reverse-mode automatic differentiation for sparse Hessians
24 Stars
Updated Last
2 Years Ago
Started In
February 2014


Build Status

Reverse-mode automatic differentiation for closed-form scalar algebraic expressions, producing gradients and Hessians. Efficient sparse Hessian computation is implemented by using state-of-the-art graph coloring approaches, written in pure Julia (see src/coloring.jl). This package is used by JuMP 0.18 and earlier. The code was integrated directly into the JuMP repository in March 2018. This repository will be maintained only as long as JuMP 0.18 is supported. In either case, JuMP is the only documented interface for using this code.

Used By Packages

No packages found.