Some confusion in my script for lsqcurvefit

Hello everyone,

I have somewhat of an issue I don't quite know how to resolve. I have a very simply code:

data1=load ('Concentrations.txt');
data2=load ('Titration.txt');
J=data2(:,1)
L=data1(:,1);
P=data1(:,2);
A=P;
B=P+L;
C=L';
xdata=[P L]
fun=@(x,xdata) (B+x(1)-sqrt(((B+x(1)).^2)-4.*A.*C))./(2.*A)
x0=[1]
x = lsqcurvefit(fun,x0,xdata,J)

The error I am getting is

Function value and YDATA sizes are not equal.

Now from my understanding, what is going on here is I have some unknown x(1), and this value changes while keeping xdata constant, to find a value that matches ydata. This then goes through each data point (i.e. each row or column in your xdata file), and does the same thing.

Now what this would imply is the output of the function F(x,xdata) is not the same as my ydata (therefore F(x,xdata-ydata) would be problematic). However, I have checked the sizes of all my vectors, and they all match.

>> whos J
  Name      Size            Bytes  Class     Attributes

  J         7x1                56  double              

>> whos xdata
  Name       Size            Bytes  Class     Attributes

  xdata      7x2               112  double              

>> whos P
  Name      Size            Bytes  Class     Attributes

  P         7x1                56  double              

>> whos L
  Name      Size            Bytes  Class     Attributes

  L         7x1                56  double                      

Now in my situation, both P and L are changing each iteration (that's why it's a 7x2), so what I've attempted to do is have xdata be a matrix where one column is P and another L. I.E. for F(x,xdata(1)) it will take P(1,1) and L(1,2).

Answer 1

[Guillaume]

You know that (the badly named) P and L are both column vectors, then you have

A=P;        %how about using meaningful names for the variables
C=L';    %NOTE THE TRANSPOSE

So, C is a row vector while A is a column vector, therefore in the anonymous function

A.*C

will be a square matrix of size numel(P) x numel(P) (in your case 7x7).

[Sam Mahdi]

Thank you, didn't see that typo there. Also, the P and L are abbreviations for protein and ligand, and the A,B,C, is for the quadratic equation I have set up.