Hi Steven,

I missed the tol in my first post and clarified that in my 2nd post. The
contraint is

|C(x)-1|<tol.

So it still remains an inequality constraint but I guess with a very small
interior region (my tol is 10e-2).

The derivative of the constraints are the same for the positive and
negative case of the inequalities.

Under this setup does you reasoning/explanation still valid ?

Best
Kaushik
On Feb 10, 2016 9:59 AM, "Steven G. Johnson" <[email protected]> wrote:

>
> > On Feb 3, 2016, at 9:36 PM, Kaushik Matia <[email protected]>
> wrote:
> >  I am new to nlopt and I am trying to figure out an issue with the LD
> MMA algorithm.
> >
> >  My optimization problem has the following inequality constraints.
> >
> >   |C(x) -1| < 0
>
> I assume you mean ≤ 0.   This is effectivley an equality constraint, C(x)
> = 1.
>
> However, as explained in the documentation, MMA does not support equality
> constraints in any form — technically, it requires the feasible set to have
> a non-empty interior.  So, it's not surprising that it fails.
_______________________________________________
NLopt-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss

Reply via email to