For an implicit ode/dae solver, I want to provide automatic computation
of a sparse Jacobian, either through automatic differentiation or finite
differences.  So, for given a function

f(y, ydot)

I need to compute the Jacobian in the following form:

J = df/dy + a* df/dydot

ReverseDiffSparse seems to do most of this (although Hessians are
symmetric) but I struggle to make sense of it.  Given f and a sparsity
pattern of J, is there an incantation to get df/dy and df/dydot?

If I wanted to use the coloring of ReverseDiffSparse (or some other
package) for finite differencing, how would I go about it?  What
function calls would give me a coloring of the colums?

Thanks! Mauro

Reply via email to