Thread Subject: Help with lsqcurvefit

I'm newbie on matlab and I want to use lsqcurvefit to adjut 2 parameters, but this function is giving me a

error that I coudn't solve.

My function is,

VEc=Vredm.*(((1-x).*K(1,1).*vass(1,1).*(X111-A1))+(x.*K(2,1).*vass(2,1).*(Y111-A2))+(((1-x).*KAB.*VassAB.*Y11

1.*(1-K(1,1).*X111))./((V(2,1)/V(1,1)).*(1-K(2,1).*Y111)+KAB.*Y111)));

Vredm, x, X111, Y111 - 23 values each variable
K, vass, V have - 2 values each variable
A1, A2 - 1 value each variable

KAB, VassAB - adjustable parameters

When I give some values for KAB and VassAB, VEc runs OK, but when I use it in lsqcurvefit it give me this

error,

Error using ==> times
Matrix dimensions must agree.


Here is my lsqcurvefit function,
[KAB,VassAB]=lsqcurvefit(@(KAB,VassAB)

Vredm.*(((1-x).*K(1,1).*vass(1,1).*(X111-A1))+(x.*K(2,1).*vass(2,1).*(Y111-A2))+(((1-x).*KAB.*VassAB.*Y111.*(

1-K(1,1).*X111))./((V(2,1)/V(1,1)).*(1-K(2,1).*Y111)+KAB.*Y111))),[1,1],x,ve);

Please somebody could help me with this function?

Thanks in advanve.

I'm not sure, but I believe you did not construct your anonymous
function properly. As you see in the lsqcurvefit function reference page
http://www.mathworks.com/access/helpdesk/help/toolbox/optim/ug/lsqcurvefit.html
lsqcurvefit moves the values in your first argument (KAB), assuming
VassAB is a fixed set of data. Take a look at the lsqcurvefit example
http://www.mathworks.com/access/helpdesk/help/toolbox/optim/ug/brn4noo.html#brp3l6v-1

When you call lsqcurvefit you have to pass in xdata and ydata matrices.
Did you do that? I don't see them in your function call.

Good luck,

Alan Weiss
MATLAB mathematical toolbox documentation

On 7/30/2010 4:42 PM, Juca wrote:
> I'm newbie on matlab and I want to use lsqcurvefit to adjut 2
> parameters, but this function is giving me a
> error that I coudn't solve.
>
> My function is,
>
> VEc=Vredm.*(((1-x).*K(1,1).*vass(1,1).*(X111-A1))+(x.*K(2,1).*vass(2,1).*(Y111-A2))+(((1-x).*KAB.*VassAB.*Y11
>
>
> 1.*(1-K(1,1).*X111))./((V(2,1)/V(1,1)).*(1-K(2,1).*Y111)+KAB.*Y111)));
>
> Vredm, x, X111, Y111 - 23 values each variable K, vass, V have - 2
> values each variable A1, A2 - 1 value each variable
> KAB, VassAB - adjustable parameters
>
> When I give some values for KAB and VassAB, VEc runs OK, but when I use
> it in lsqcurvefit it give me this
> error,
>
> Error using ==> times
> Matrix dimensions must agree.
>
>
> Here is my lsqcurvefit function,
> [KAB,VassAB]=lsqcurvefit(@(KAB,VassAB)
> Vredm.*(((1-x).*K(1,1).*vass(1,1).*(X111-A1))+(x.*K(2,1).*vass(2,1).*(Y111-A2))+(((1-x).*KAB.*VassAB.*Y111.*(
>
>
> 1-K(1,1).*X111))./((V(2,1)/V(1,1)).*(1-K(2,1).*Y111)+KAB.*Y111))),[1,1],x,ve);
>
>
> Please somebody could help me with this function?
>
> Thanks in advanve.

Alan Weiss <aweiss@mathworks.com> wrote in message <i373br$k1u$1@fred.mathworks.com>...
> I'm not sure, but I believe you did not construct your anonymous
> function properly. As you see in the lsqcurvefit function reference page
> http://www.mathworks.com/access/helpdesk/help/toolbox/optim/ug/lsqcurvefit.html
> lsqcurvefit moves the values in your first argument (KAB), assuming
> VassAB is a fixed set of data. Take a look at the lsqcurvefit example
> http://www.mathworks.com/access/helpdesk/help/toolbox/optim/ug/brn4noo.html#brp3l6v-1
>
> When you call lsqcurvefit you have to pass in xdata and ydata matrices.
> Did you do that? I don't see them in your function call.
>
> Good luck,
>
> Alan Weiss
> MATLAB mathematical toolbox documentation
>
> On 7/30/2010 4:42 PM, Juca wrote:
> > I'm newbie on matlab and I want to use lsqcurvefit to adjut 2
> > parameters, but this function is giving me a
> > error that I coudn't solve.
> >
> > My function is,
> >
> > VEc=Vredm.*(((1-x).*K(1,1).*vass(1,1).*(X111-A1))+(x.*K(2,1).*vass(2,1).*(Y111-A2))+(((1-x).*KAB.*VassAB.*Y11
> >
> >
> > 1.*(1-K(1,1).*X111))./((V(2,1)/V(1,1)).*(1-K(2,1).*Y111)+KAB.*Y111)));
> >
> > Vredm, x, X111, Y111 - 23 values each variable K, vass, V have - 2
> > values each variable A1, A2 - 1 value each variable
> > KAB, VassAB - adjustable parameters
> >
> > When I give some values for KAB and VassAB, VEc runs OK, but when I use
> > it in lsqcurvefit it give me this
> > error,
> >
> > Error using ==> times
> > Matrix dimensions must agree.
> >
> >
> > Here is my lsqcurvefit function,
> > [KAB,VassAB]=lsqcurvefit(@(KAB,VassAB)
> > Vredm.*(((1-x).*K(1,1).*vass(1,1).*(X111-A1))+(x.*K(2,1).*vass(2,1).*(Y111-A2))+(((1-x).*KAB.*VassAB.*Y111.*(
> >
> >
> > 1-K(1,1).*X111))./((V(2,1)/V(1,1)).*(1-K(2,1).*Y111)+KAB.*Y111))),[1,1],x,ve);
> >
> >
> > Please somebody could help me with this function?
> >
> > Thanks in advanve.

Alan,

I took a look at the documentation, changed my function and now it's working. You're right, the problem was my anonymous function.

This is my new working function,
[a]=lsqcurvefit(@(a,x) Vredm.*(((1-x).*K(1,1).*vass(1,1).*(X111-A1))+(x.*K(2,1).*vass(2,1).*(Y111-A2))+(((1-x).*a(1).*a(2).*Y111.*(1-K(1,1).*X111))./((V(2,1)/V(1,1)).*(1-K(2,1).*Y111)+a(1).*Y111))),[1,-5],x,ve,[1,-500],[100,-1]);

Thanks for the help.