On 10/12/2009 5:49:05 PM, djoffe wrote:
>Can anyone point me to
>resources to answer the
>following:
>How do you get Lev-Mar to not
>ignore inequality restrictions
>in the solve block? It seems
>to do so with a mind of its
>own.
The constraints are soft constraints, not hard constraints. The error in the constraint is minimized, along with all the other errors, but not forced to zero. If you want to force it to zero then you need to weight it very heavily. So, for example, instead of "a less than b" use "(a-b)*10^15 less than 0"
>Also...in Ver14, it will give
>complex results in
>optimization even if the guess
>values are all reals.
Please post an example worksheet showing this.
>Finally...I've been using
>Mathcad for years...is there a
>book you can recommend that
>might handle these and some
>other issues not seemingly in
>the help docs? (Numerical
>Methods Using Mathcad is not
>such a book)
No. As far as I know, there isn't one. The best thing to do is ask here.
Richard
