Re: problems with sqp constraints?

[Olaf Till]

On Mon, Apr 16, 2012 at 09:46:44PM +0000, fork wrote:
> fork <forkandwait <at> gmail.com> writes:
> 
> More info ...
> 
> > Octave error (full paths elided):
> > 
> > error: octave_base_value::array_value(): wrong type argument `complex 
> > matrix'
> > error: octave_base_value::int_value (): wrong type argument `scalar'
> > error: __qp__: invalid arguments
> > error: called from:
> > error:   ... 3.6.1\m\optimization\qp.m at line 393, column 26
> > error:   ... 3.6.1\m\optimization\sqp.m at line 414, column 32
> > error:   ... sqp_experiments.m at line 47, column 35
> 
> On line qp.m:393, H has some small complex components that creep in:

sqp does not necessarily honour your constraints and bounds for
intermediate results. So logarithms of negatives may intermediately
happen with your code and probably have caused this error (though I
don't see it when I repeat your code). You could check for p <= 0
(instead of p == 0) in your objective function to avoid the error.

BTW you should check your gradient function.

Olaf

-- 
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net

signature.asc
Description: Digital signature