VERY STRANGE LSQCURVEFIT RESULT

Hugh — Wed, 28 Oct 2009 16:26:02 -0400

Hi there,

Using the code below lqscurvefit produces a net error of 0.00812. However when one looks at the fitted curve and compares it with the curve being fitted it is clear that they are massively different due to the second term.

Can anyone explain this apparent error?

%[cn,error]=lsqcurvefit(@PrimePerm3,c0,xdata3,ydata3,lbn,ubn,options);
[cn,error]=lsqcurvefit(@PrimePerm3,c0,xdata,ydata,lbn,ubn,options);

y=PW(xdata2,ydata2,cn(5),cn(6),cn(7),cn(8),cn(9));
params = y';
%params= [cn(1) cn(2) cn(3) cn(4) ];
params=[params cn(5) cn(6) cn(7) cn(8) cn(9)];
y=CRM(xdata3,params);
error2=(y-ydata3).^2;
error2 = sum(error2);
%Res(z,:) = [cn params error error2];
Res(1,:) = [cn params error error2];
%end

optim=find(Res(:,20)==min(Res(:,20)));
toc;
% cn=zeros(n,3); error=zeros(n,1);
%
% for i=1:n
% % add noise
% NoisyS(i,:) = noise(y,noiselevel,I0);
%
% % digitizing the signal
% DigiS = digi(NoisyS(i,:),bits);
%
% % error introduced by both noise and digitization
% substraction(i,:) = DigiS - y;
%
% % datafitting
% [cn(i,:),error(i)]=lsqcurvefit(@singleexp,c0,x,DigiS,lbn,ubn,options);
% end

ydata: 4.30000000000000 4.57000000000000 4.78000000000000 4.94000000000000 5.09000000000000 5.23000000000000 5.36000000000000 5.46000000000000 5.52000000000000 5.64000000000000 5.80000000000000 5.92000000000000 5.96000000000000 5.95000000000000

y fitted:
4.30000000000000 -21107.6721741101 4.46147667519652 4.94000000000000 5.13432544641994 5.26732355159549 5.36979229528067 5.45219792241475 5.52000000000000 5.62464324616620 5.73203476147917 5.84087976409200 5.90636658873406 5.95000000000000

MATLAB Central Newsreader - VERY STRANGE LSQCURVEFIT RESULT

VERY STRANGE LSQCURVEFIT RESULT