[Trilinos-Users] evalModel, dfdp with several parameters

[Bartlett, Roscoe A]

Nico,

I also read in the documentation:
    > Parameters are grouped together into subvectors p(l) implies that certain derivatives will be supplied for all the parameters in the subvector or none.

How does that fit?

[Bartlett, Roscoe A]

This just means that if p(0) has five components, then if you say that you support dfdp(0), that means that you have to support the computation of the derivatives for all five of the scalar parameters in p(0).

Does that make sense?

I am going to write a formal unit testing class call ModelEvaluatorTester soon that will better formalize the expectations of the ModelEvaluator interface.  That will likely happen within the next few months.

Cheers,

- Ross

-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://software.sandia.gov/pipermail/trilinos-users/attachments/20100527/a8e997cc/attachment.html