|
|
Re: Automatic Differentiation?
Posted:
Feb 28, 2004 4:32 AM
|
|
Andrea wrote: > Hello, > > if I remember correctly, AD can be useful when your objective > function has an analytical expression. There are a lot of different > engineering/mathematical problems in which this is not possible. For > example, if your OF come from a numerical solution of a PDE (or set of > PDEs), I don't think that AD can be applied.
Have a look at the examples and publications at www.autodiff.org. you can't do everything with large scale PDE problems, but the automatic differentiation of large and complex PDE codes is one of the major areas where this approach has been very successful.
Regards. Volker Rath
|
|