Minimalizing a function of two variables

Hopefully my final question for a while.

I have my final function

FullScore(P1,P2)

This takes in experimental data, and works out the mean difference between this and data that is produced using P1 and P2. Origionally i was going to try use lsqcurvefit/lsqnonlin but because of my previous question my experimental data and simulated data are both used to produce the Y values on their respecive graphs.

My goal is to find the values of P1 and P2 that produce data that is the most similar to my experimental data

i am now simply trying

fminsearch(FullScore,[15 0.5]) %P1 goes from 0-35 and P2 0.01-1

When i run the function on its own it works fine, but when i run it from above it doesnt read 15 as P1 and 0.5 as P2 as the fire time these come up in the code i get(this is basically the first line of the code)

Not enough input arguments.

Error in FullScore (line 15)

newtext = [text1, num2str(P1), 'p ', num2str(P2), text2] ;

How can i make it so that my function takes in my two startin guesses correctly?

Answer 1

[Voss]

First thing. This:

fminsearch(FullScore,[15 0.5])

calls the function FullScore with no inputs; that's why you get that error. You need to send the function handle @FullScore instead:

fminsearch(@FullScore,[15 0.5])

Then you'll get a different error, due to the fact that fminsearch takes functions of one input only. You can get around that by either (1) redefining your function FullScore to take a single input:

function out = FullScore(P1P2)
P1 = P1P2(1);
P2 = P1P2(2);
% ... your code
end

Or (2) make an anonymous function that takes a single 1-by-2 input and sends two scalar inputs to FullScore, and call fminsearch on that anonymous function:

f = @(pp)FullScore(pp(1),pp(2));
fminsearch(f,[15 0.5])

(or, same as above, but without storing the anonymous function as the variable f:)

fminsearch(@(pp)FullScore(pp(1),pp(2)),[15 0.5])

With the second approach you don't have to change the definition of FullScore.

[Charles Mitchell-Thurston]

Thanks, you answer made it run.

Sadly the output is wrong but thats a problem for tomorrow. (by wrong i mean it gives P1 and P2 suggestions but the suggestions are bad)