Prev: plotting multiple functions in same axis, 2D.
Next: quadl error
From: Corey Smith on 27 Nov 2009 12:21
Hi:

I am have a set of raw data of signal intensity along a distance, r and am trying to fit a specified equation to this data by manipulating the factor D, while keeping the rest constant. I am having trouble coming up with the code and keep getting an error saying "Failure in initial user-supplied objective function evaluation. LSQCURVEFIT cannot continue."

Below is what I have so far. Thanks in advance!

function F = myfun(Co,a,r,D,t)

I = linspace(200,6800);
r = linspace(1,6);
Co = 6895;
t = 480;
a = 1.92;
D = 4E-4;

%output = inline('((1/2).*(Co).*(erf((a-r)./(2.*(sqrt(D.*t))))+(erf((a+r)./(2.*(sqrt(D.*t)))))))-((Co./r).*(sqrt((D.*t)./pi)).*((exp((-((a-r).^2))./(4.*D.*t))-(exp((-((a+r).^2))./(4.*D.*t))))))','Co','a','r','D','t');
F = ((1/2).*(Co).*(erf((a-r)./(2.*(sqrt(D.*t))))+(erf((a+r)./(2.*(sqrt(D.*t)))))))-((Co./r).*(sqrt((D.*t)./pi)).*((exp((-((a-r).^2))./(4.*D.*t))-(exp((-((a+r).^2))./(4.*D.*t))))));

[D,resnorm]=lsqcurvefit('myfun',Co,a,r,D,t,I);
From: Godzilla on 27 Nov 2009 20:22
"Corey Smith" <csmit26(a)uwo.ca> wrote in message <hep1q0$c8u$1(a)fred.mathworks.com>...
> Hi:
>
> I am have a set of raw data of signal intensity along a distance, r and am trying to fit a specified equation to this data by manipulating the factor D, while keeping the rest constant. I am having trouble coming up with the code and keep getting an error saying "Failure in initial user-supplied objective function evaluation. LSQCURVEFIT cannot continue."
>
> Below is what I have so far. Thanks in advance!
>
> function F = myfun(Co,a,r,D,t)
>
> I = linspace(200,6800);
> r = linspace(1,6);
> Co = 6895;
> t = 480;
> a = 1.92;
> D = 4E-4;
>
> %output = inline('((1/2).*(Co).*(erf((a-r)./(2.*(sqrt(D.*t))))+(erf((a+r)./(2.*(sqrt(D.*t)))))))-((Co./r).*(sqrt((D.*t)./pi)).*((exp((-((a-r).^2))./(4.*D.*t))-(exp((-((a+r).^2))./(4.*D.*t))))))','Co','a','r','D','t');
> F = ((1/2).*(Co).*(erf((a-r)./(2.*(sqrt(D.*t))))+(erf((a+r)./(2.*(sqrt(D.*t)))))))-((Co./r).*(sqrt((D.*t)./pi)).*((exp((-((a-r).^2))./(4.*D.*t))-(exp((-((a+r).^2))./(4.*D.*t))))));
>
> [D,resnorm]=lsqcurvefit('myfun',Co,a,r,D,t,I);

I believe the fitting program expects 'myfun' in a certain format (i.e. input and output). Check your myfun.
From: Corey on 4 Dec 2009 14:38

> I believe the fitting program expects 'myfun' in a certain format (i.e. input and output). Check your myfun.

Thanks for your response. I have played with it more and gotten a bit closer, but the problem I am having now is that F wants me to define D when I run this code, but it should be finding D for me. Where am I going wrong now?


function F = myfun(D,r)

I = load('33323_MiddleTray4_xplot.txt');
x=I(:,1);
y=I(:,2);
r = x;
I = y;
Co = 6895;
t = 480;
a = 1.92;
D0 = 4E-4;

F = ((1/2).*(Co).*(erf((a-r)./(2.*(sqrt(D(1).*t))))+(erf((a+r)./(2.*(sqrt(D(1).*t)))))))-((Co./r).*(sqrt((D(1).*t)./pi)).*((exp((-((a-r).^2))./(4.*D(1).*t))-(exp((-((a+r).^2))./(4.*D(1).*t))))));

D = lsqcurvefit('myfun',D0,r,I);
From: Corey on 4 Dec 2009 15:09
I apologize, I have just made a great stride forward and it appears to be working the way I want it to. Now just to make it work so that it returns R^2 with D.
 | 
Pages: 1
Prev: plotting multiple functions in same axis, 2D.
Next: quadl error