[SciPy-user] constrained optimization
Tue Apr 29 03:15:39 CDT 2008
On Tue, Apr 29, 2008 at 3:04 AM, dmitrey <firstname.lastname@example.org> wrote:
> I guess the real problems for using this and/or almost any other
> coordinate transformation could be possible yielding ill-condition
> problem (or increasing ill-condition, if former problem is already
> ill-conditioned), for example when coord x[-1] is close to zero either
> in optimal point or during moving along trajectory. This makes
> optimization solvers to work unstable, unpredictable and maybe stop far
> from optimum.
True. However, it should be noted that this transform and the related
ones have very nice properties (e.g., they have a Euclidean metric).
One could equally say that the original formulation is transformed
from the more natural Euclidean coordinates and could lead
optimization solvers to perform poorly, particularly those which
require gradients (and thus, a metric).
"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
-- Umberto Eco
More information about the SciPy-user