Matrix dimensions don't agree

Hello, so I am trying to find the best parameters and RSS value for the following program:

%Program HollingTypeIII.m

%Parameterization mode A1b

%Calls RSSHolling.m and fminsearch.m

global age ulna

%Load data

load ulnaestimation.txt

data = ulnaestimation

age = data(:,1);

ulna = data(:,2);

%Initialize a and b

theta = log([1 1.5]);

%Run Nelder-Mead algorithm and output best parameter to screen

[output RSSbest] = fminsearch('RSSHolling',theta);

params = exp(output);

abest = params(1)

bbest = params(2)

RSSbest

The subroutine is:

%Subroutine RSSHolling.m

%Called by HollingTypeIII.m

function RSS = RSSHolling(theta)

global age ulna K

par = exp(theta);

a = par(1);

b = par(2);

K = 13.94615385;

pred = (((K-b)*(age.^2))/((a^2)+(age.^2)))+b

%Compute RSSHolling on sqrt scale

residuals = sqrt(ulna)-sqrt(pred);

RSS = sum(residuals.^2);

end

However, when I run the program, it comes up with an error:

Error using - Matrix dimensions must agree.

Error in RSSHolling (line 12) residuals = sqrt(ulna)-sqrt(pred);

Somehow, when the pred equation is run, it results in a huge matrix, when I need a 40x1 matrix. I have tried reshaping the matrix; separately, I have tried writing "pred'" in the residuals equation to flip the matrix. This does not help. Does anyone have any answers as to what might be wrong in my program?

Answer 1

[James Tursa]

Try changing the division to element-wise:

pred = (((K-b)*(age.^2))./((a^2)+(age.^2)))+b; % <-- Changed / to ./

[Viktoria Kolpacoff]

It worked. Thank you so much!