The problem of initial parameters in Lsqcurvefit

I have a problem in Lsqcurvefit.

I am optimizing Boucwen Model with lsqcurvefit.

But when it runs with the initial parameters of below, it says

"Matrix dimensions must agree. Error in lsqcurvefit/objective (line 265)

F = F - YDATA; "

a = [1,1,1,1,1,1];

But if i use this initial parametes of below by looking for manually, it works well .

a = [12,5966,29,147,538,245];

Since I have 300 datas, I can't find each initial parameters.

So I read two ways of below such as exitflag, output and optimooptions.

I dont understand well which things are good for mine, So could you give me the recommendations for solving this problem?

Thank you so much.

https://www.mathworks.com/help/optim/ug/lsqcurvefit.html#buuhcjo_sep_shared-exitflag

https://www.mathworks.com/help/optim/ug/optimoptions.html

Answer 1

[Matt J]

So could you give me the recommendations for solving this problem?

Check the output of predicted(). Make sure it has the same dimensions as ydata2.

[CHOI HYUNDUK]

Thank you.

I understand that the dimesions chance depends on initial parameters, right?

And then how can I make same demension in this case?

force predicted is interpolated. But ydata is not interpolated.

So I need to interpolate ydata. or I have to get not random span by ODE function.

Is there another way for this?

[Matt J]

force predicted is interpolated. But ydata is not interpolated.

Isn't it? What is this line doing in your code?

ydata2 = interp1(time, ydata,t,'spline');

[CHOI HYUNDUK]

When it is optimized

lsqcurvefit(predicted,a,t,ydata2);

here, predicted funtion gives output interpolated by random time span.

So, it needs to fit same vector between predicted output and ydata.

I made 't function' additionaly for getting 't' number for interpoltating ydata to ydata2.

But I think now, t function just is get 'initial parameter' just one time.

therefore, ydata2 has always same vector.

However, predicted output vector is changed depends on 'parameters optimizing in real time'.

I wonder how can i fix this error.

The more initial parameters are close at parameters that i want, it works well without problem coincidently.