fitting 3 sets of data simultaneously using "lsqcurvefit"

I have a set of data in triplicates (attached). I want to fit the exponential part of these 3 sets simultaneously using "lsqcurvefit" and get one x(1) and x(2). I have included "t" and "or" data in the attached file and here is my function:

function y=or(x,t)

y=((0.282*x(2))-0.24)*x(1)*exp((x(2)-0.24)*t)

I managed to fit each data set individually and average x(1) and x(2) and calculte the STdev but if I could fit 3 sets simultenausly using "lsqcurvefit" would be better.

I would be grateful if you kindly help me.

Answer 1

[Torsten]

Define

T = [t1;t2;t3] 

and

OR = [or1;or2;o3]

and treat

[T,OR]

just like you treated

[t1;or1],[t2;or2],[t3;or3]

.

What's the problem ?

[Sara]

I have one t, it is the same for or1, or2, and or3. I have defined the fit as below for each or:

rangemin = 173;

rangemax = 212;

xprime(1)=100;

xprime(2)=7;

[x(1,:)]=lsqcurvefit(@or,xprime,t(rangemin:rangemax),ordata(rangemin:rangemax,1));

[x(2,:)]=lsqcurvefit(@or,xprime,t(rangemin:rangemax),ordata(rangemin:rangemax,2));

[x(3,:)]=lsqcurvefit(@or,xprime,t(rangemin:rangemax),ordata(rangemin:rangemax,3));

y(:,1)=or(x(1,:),t);

y(:,2)=or(x(2,:),t);

y(:,3)=or(x(3,:),t);

function y=or(x,t)

y=((0.282*x(2))-0.24)*x(1)*exp((x(2)-0.24)*t);

end

…………………………………………………….

I am new in Matlab and it took me a long time to do this.

Do you mean that I write as:

[x(1,:), x(2,:), x(3,:)]=lsqcurvefit(@or(x(1,:),t);or(x(2,:),t);or(x(3,:),t),xprime,t(rangemin:rangemax),ordata(rangemin:rangemax,1,2,3));

[Matt J]

Do not name your model function or(). It conflicts with the name of a (very important) built-in Matlab function.

x=lsqcurvefit(@modelfun,xprime,t,[OR1(:),OR2(:),OR3(:)])

function y=modelfun(x,t)

  y=((0.282*x(2))-0.24)*x(1)*exp((x(2)-0.24)*t(:));
  
  y=[y,y,y];

end

[Sara]

I have x(1) and x(2), I wrote as below but it doesnt work:(

[x(1,:)]=lsqcurvefit(@modelfun,xprime,t(rangemin:rangemax),or(rangemin:rangemax,:));

y(:,1)=[modelfun(x(1,:);modelfun(x(2,:);modelfun(x(3,:),t)];

figure

plot(t,or(:,1),t(rangemin:rangemax),y(rangemin:rangemax),'o');

function modelfun(x,t)

y=((0.282*x(2))-0.24)*x(1)*exp((x(2)-0.24)*t(:));

y=[y,y,y];

end

[VBBV]

[x(1,:)]=lsqcurvefit(@modelfun,xprime,t(rangemin:rangemax),ordata(rangemin:rangemax,:));  % watch 3rd argument 

Use the 3rd argument to function as ordata instead of or and try

[Torsten]

rangemin = 173;
rangemax = 212;
xprime(1)=100;
xprime(2)=7;

TDATA = [t(rangemin:rangemax).';t(rangemin:rangemax).';t(rangemin:rangemax).'];
ORDATA = [ordata(rangemin:rangemax,1);ordata(rangemin:rangemax,2);ordata(rangemin:rangemax,3)];
x = lsqcurvefit(@modelfun,xprime,TDATA,ORDATA);
Y = modelfun(x,TDATA)
Y = reshape(Y,numel(t(tangemin:rangemax)),3);

function modelfun(x,t)
  y=((0.282*x(2))-0.24)*x(1)*exp((x(2)-0.24)*t);
end

[Sara]

Thanks very much but gives error in

TDATA = [t(rangemin:rangemax).';t(rangemin:rangemax).';t(rangemin:rangemax ).'];

[Torsten]

Is t a row or a column vector ?

If it is a column vector, use

TDATA = [t(rangemin:rangemax);t(rangemin:rangemax);t(rangemin:rangemax)];

[Sara]

column, it worked but now gives error on

Y1 = Y(1:numel(t(trangemin:rangemax)));

[Torsten]

Y is a numel(t(tangemin:rangemax)) x 3 - matrix.

So

Y1 = Y(:,1)
Y2 = Y(:,2)
Y3 = Y(:,3)

guessing you used the code I submitted.

Answer 2

[Sara]

ok, sorry many comments so was confused:(

I think this is based on your comment:

rangemin = 173;

rangemax = 212;

xprime(1)=100;

xprime(2)=7;

TDATA = [t(rangemin:rangemax);t(rangemin:rangemax);t(rangemin:rangemax)];

ORDATA = [ordata(rangemin:rangemax,1);ordata(rangemin:rangemax,2);ordata(rangemin:rangemax,3)];

x = lsqcurvefit(@or,xprime,TDATA,ORDATA);

Y = or(x,TDATA);

Y = reshape(Y,numel(t(tangemin:rangemax)),3);

function y=or(x,t)

y=((0.282*x(2))-0.24)*x(1)*exp((x(2)-0.24)*t);

end

but still gives error on Y = reshape(Y,numel(t(tangemin:rangemax)),3);

[Torsten]

Y = reshape(Y,numel(t(rangemin:rangemax)),3);

instead of

Y = reshape(Y,numel(t(tangemin:rangemax)),3);

My fault, sorry.

[Sara]

these seem the same?

[Torsten]

rangemin vs. tangemin ?