Prev: Dicomread changes pixel values!
Next: Testing for block diagonal form
From: JOHN pip on 6 Jun 2010 17:49
"Richard Crozier" <r.crozier(a)ed.ac.uk> wrote in message <hugjcb$24n$1(a)fred.mathworks.com>...
> "JOHN pip" <johny-pip(a)hotmail.com> wrote in message <hugg8v$8b3$1(a)fred.mathworks.com>...
> > "Matt J " <mattjacREMOVE(a)THISieee.spam> wrote in message <hugdvr$99d$1(a)fred.mathworks.com>...
> > > "JOHN pip" <johny-pip(a)hotmail.com> wrote in message <hugd5l$fb2$1(a)fred.mathworks.com>...
> > >
> > > > So when i run fmincon using exactly all the parametres above and if the initial start for the gi is between lb and ub,fmincon returns as a result for the gi exactly the values of the initial start. Is this propper to happen somethin like this
> > > ==========================
> > >
> > > Only if the gi are already optimal or stationary.
> > >
> > > and if not what should i do?Shall i try another matlab function?
> > > =============
> > >
> > > I've already advised you to stick with exhaustive search. If you insist on using fmincon, you should probably look for a bug in your objfun.
> >
> >
> > I insist on using fmincon because in my problem i aslo have to give some specific initial start and compare this to exhaustive.I m sorry but what does exactly bug means?
>
> A 'bug' in a program is another name for an error in the program resulting in an unexpected output, if this is what you are asking.
>
> see:
>
> http://en.wikipedia.org/wiki/Software_bug
>
>


Thank's for your advise but i'm pretty sure tha there is no 'bug' if i got it right in my objfun.Is it impossible to be another mistake and mostly related to fmincon?
From: Steven Lord on 6 Jun 2010 22:09

"JOHN pip" <johny-pip(a)hotmail.com> wrote in message
news:hugd5l$fb2$1(a)fred.mathworks.com...
> "Matt J " <mattjacREMOVE(a)THISieee.spam> wrote in message
> <hu5p51$e1r$1(a)fred.mathworks.com>...
>> "JOHN pip" <johny-pip(a)hotmail.com> wrote in message
>> <hu5apj$db1$1(a)fred.mathworks.com>...
>>
>> > And can i somehow constrain gi to integers when using fmincon in order
>> > to reduce the number of feasibles solutions?Is there any option that
>> > can do this?
>> ================
>>
>> Why not just stick with exhaustive search? The number of feasible
>> solutions numbers at just a few thousand at most, for any given set of
>> parameters p1,p2,p3. If you efficiently code the computation of your
>> nonlinear constraints and objective function, you should eb able to solve
>> it in the blink of an eye.
>
>
> First of all i would like to say that i highly appreciate your help,and by
> the way i would like to ask you one more thing.
> So when i run fmincon using exactly all the parametres above and if the
> initial start for the gi is between lb and ub,fmincon returns as a result
> for the gi exactly the values of the initial start.Is this propper to
> happen somethin like this and if not what should i do?Shall i try another
> matlab function?

If your function "looks" flat, yes that's proper. If you starting point
satisfies your constraints, and FMINCON can't find a direction in which it
can reduce the value of your objective function significantly (which would
be the case if your start point is in the middle of a flat region) then it
is allowed to return the starting point. Since you indicate you're trying
to solve a problem where some or all of the decision variables are integers,
I suspect you're using ROUND or another rounding function in your objective
function -- if so that could very well make the objective look flat.

--
Steve Lord
slord(a)mathworks.com
comp.soft-sys.matlab (CSSM) FAQ: http://matlabwiki.mathworks.com/MATLAB_FAQ
To contact Technical Support use the Contact Us link on
http://www.mathworks.com


First  |  Prev  | 
Pages: 1 2 3
Prev: Dicomread changes pixel values!
Next: Testing for block diagonal form