DualArrays.jl: A Julia package for optimized forward-mode AD.
DualArrays.jl is a Julia package that optimises forward-mode automatic differentiation by implementing a structure of a dual array, an array of dual numbers, whose dual part (‘Jacobian’) can be declared with sparse structures (e.g BandedMatrix, SparseArray). This enables faster operations on these arrays of dual numbers that can optimise forward mode AD for uses such as SciML, neural networks and numerically solving ODEs.