[R] linear constraint optim with bounds/reparametrization

Thomas Lumley tlumley at u.washington.edu
Mon Aug 9 18:56:05 CEST 2004

On Mon, 9 Aug 2004, Spencer Graves wrote:

> Hi, Tom:
>       Why is adding "a multiple of log(A*theta-c) to the objective
> function ... a really bad idea as a way of faking equality constraints"?

Because it is infinite everywhere on the feasible set: log(0)?

It's fine to add constraints to the problem, but these aren't sensible
constraints for that purpose.


More information about the R-help mailing list