Prev: mat into excel
Next: Creating new variables in a loop
From: SM on 15 May 2010 16:29
Hi everybody,

I have created an algorithm to which fits a model to the set of data in a following way. My data is stored in a mXn matrix
for instance Matrix A:

A=[0,0.33,2.63,3.92,5.53,30.93,24.45,59.39,76.32,112.82,105.47,192.93,171.14,208.58,156.63,174.04,198.59,167.08,106.71,105.96,84.47,64.81,53.97,24.14,24.73,16.56,14.03,5.85,6.24,4.19,3.905;...
0.48,0.081,0.35,1.21,2.38,2.75,3.75,3.68,5.875,10.25,9.39,12.11,14.659,15.71,20.86,16.32,16.399,14.992,13.74,14.714,5.89,4.24,4.10,4.20,0.99,1.20,2.11,1.15,0,0.82,0.25] ;

I want to fit the model which is defined as myfun:

function F = myfun (xx,xdata)
F = x(1).*exp(-((xdata-x(2)).^2./(2*x(3).^2)));

to the data in following way.
1- Take data of row j (here j=1) (A(1,:));
2- Take data of certain column (assigned by block0)
3- fit the model
4- plot the data and the fitted model
5- shift the block according to the shift step value
6- repeat step 3 and 4
7- last block in row 1
8- Iterate to the next row.
9- Now, combine the data from row 1 and 2 (I used reshape to do this)
10- repeat steps 2 to 8.

To combine the data I used reshape, and since I wanna plot the data in a single plot, I also calculate a new x each time which depends on j.

The problem is when the algorithm iterates through j, it fails to fit the model. Actually, I guess the problem is due to the combining the data. but I can not fix it.
I attach my code here.
Any help is appreciated.

%%%% The code is:
A=[0,0.33,2.63,3.92,5.53,30.93,24.45,59.39,76.32,112.82,105.47,192.93,171.14,208.58,156.63,174.04,198.59,167.08,106.71,105.96,84.47,64.81,53.97,24.14,24.73,16.56,14.03,5.85,6.24,4.19,3.905;...
0.48,0.081,0.35,1.21,2.38,2.75,3.75,3.68,5.875,10.25,9.39,12.11,14.659,15.71,20.86,16.32,16.399,14.992,13.74,14.714,5.89,4.24,4.10,4.20,0.99,1.20,2.11,1.15,0,0.82,0.25] ;

t=size(A,1);
l=1:1:size(A,2);
lold=l;
block_0=5;
for j=1:t
if j==1
l0=sort(unique(lold));
lold=l0;
else
B=l+(j-1)/j;
ft0=sort(([lold B]))
lold=l0
end

xdata=lold;
Anew=reshape(A(1:j,:),1,[]);
v=ones(size(Anew,1),size(Anew,2));
v(1:j,:)=j;
block=j*block_0;
sm= size(Anew,1);
sn=size(Anew,2);
inloc = 1; %1
sigma=block/2;
WidthX =2*block;
shiftstep=j; % j
timestep=1;
intime=1;
WidthY=1;
figure (1)
plot3(xdata,v,Anew,'*')
hold

for i=inloc:shiftstep:sn-WidthX
x0 = [5, (i+10), 2];
xx=lold(1,i:i+WidthX-1);
yy=Anew(1,i:i+WidthX-1);
[x,resnorm] = lsqcurvefit(@myfun,x0,xx,yy);
cmap = colormap;
u=j.*ones(1,length(xx));
if -WidthX <=x(3)&& x(3)<=WidthX
ymodel=x(1)*exp((-(xx-x(2)).^2)/(2.*x(3).^2));
plot3(xx,u,x(1).*exp(-((xx-x(2)).^2./(2*x(3).^2))),'Color',[rand(1),rand(1),rand(1)]);
hold on
l_block=x(2)
time_block=j
model_param=[x(1) x(2) x(3)]

end
end
end
From: SM on 16 May 2010 11:11
Actually, I tried to figure out the problem, but apparently the problem is due to the value of 'block'!!!
I update the block by multiplying the value into the j, while itterating. but fitting the model is not working!

Any idea, how I could assign this value?
 | 
Pages: 1
Prev: mat into excel
Next: Creating new variables in a loop