[ADMB Users] adjoint code

dave fournier davef at otter-rsch.com
Thu Sep 26 11:01:45 PDT 2013

I was perusing the web site and cam across the adjoint code discussion 
in the manual.
While showing examples for a small number of independent variables is good
for illustrating the point, it should be noted that for a function of a 
small number of independent
variables it is better to use forward mode AD.  Classes are included for 
this.  The
inverse of the cumulative distribution of the beta function is a good 
example which shows
how to use forward model AD combined with implicit differentiation to 
calculate the
derivatvies up to order 3 for an implicitly defined function.

More information about the Users mailing list