[ADMB Users] df3_one_variable

dave fournier davef at otter-rsch.com
Tue Jul 12 06:47:13 PDT 2011

What is a df3_one_variable?

To understand this you need to understand the difference between forward and reverse model automatic differentiation.

This type is for forward model AD with one independent variable.  The normal ADMB stuff for random effects involves
one forward sweep of AD followed by 2 reverse sweeps.  If you have a fairly complicated function of one variable
it is more efficient to get its derivatives contribution by forward mode AD.  The forward model stuff gets
embedded into the code with some "glue" by declaring the correct types.  There must be a good example somewhere.

More information about the Users mailing list