fminsearch for multi-input and single output

I have unknown function with two inputs ('x1' and 'x2') and one output ('Y') and a constant parameter 'p'. Each input is a vector of order [3000 x 3] and a output vector with order [3000 x 3]. Constant is [1 x 1]

best guess of unknown function is Y = (1-p)* 0.003* x1 + 275*p* x2.^4

i want to find a optimum value of parameter 'p' for which the error between the (Ymes - ) is minimum.

Ymes , X1 and X2 are known.

i tried fminsearch but it is not working

Ymes = rand(3000,3);
x1= rand(3000,3);
x2 = rand(3000,3);
fun = @(p) sum((Ymes - (1-p)*0.003* x1 + 275*p*x2.^4)).^2;

pguess = 1500;

[p,fminres] = fminsearch(fun,pguess)

Answer 1

[Torsten]

Ymes = rand(3000,3);
x1= rand(3000,3);
x2 = rand(3000,3);
Ymes = Ymes();
x1 = x1();
x2 = x2();
fun = @(p) sum((Ymes - ((1-p)*0.003* x1 + 275*p*x2.^4)).^2);
%starting guess
pguess = [0.15];
%optimise
[p,fminres] = fminsearch(fun,pguess)

or simply:

Ymes = rand(3000,3);
x1= rand(3000,3);
x2 = rand(3000,3);
Ymes = Ymes();
x1 = x1();
x2 = x2();
p = (-0.003*x1+275*x2.^4)\(Ymes-0.003*x1)

[Adan91h]

Thankyou for your prompt answer.

Still the error is same.

Assignment has more non-singleton rhs dimensions than non-singleton subscripts Error in fminsearch (line 189) fv(:,1) = funfcn(x,varargin{:});

[p,fminres] = fminsearch(fun,pguess)

[Torsten]

Did you see the change I made to your fun-definition ?

fun = @(p) sum((Ymes - ((1-p)*0.003* x1 + 275*p*x2.^4)).^2);

instead of

fun = @(p) sum((Ymes - (1-p)*0.003* x1 + 275*p*x2.^4)).^2;

[Adan91h]

with the following approach

Ymes = rand(3000,3);
x1= rand(3000,3);
x2 = rand(3000,3);
Ymes = Ymes();
x1 = x1();
x2 = x2();
p = (-0.003*x1+275*x2.^4)\(Ymes-0.003*x1)

p is 3x3 matrix.

P should be 1x1.

[Adan91h]

Yes i noticed the change in the function definition.

but the error is in line

[p,fminres] = fminsearch(fun,pguess)

and it is the same error that i mentioned above.

[Torsten]

In both cases, replace

Ymes = Ymes();
x1 = x1();
x2 = x2();

by

Ymes = Ymes(:);
x1 = x1(:);
x2 = x2(:);

[Adan91h]

Thank you for your answer.

following commands changed the vectors from [3000 x 3] to [9000 x 1] Ymes = Ymes(:); x1 = x1(:); x2 = x2(:);

I can not do this thing in my case. because the entries of 2nd and 3rd columns are not independent.

I cannot assume that first columns of both inputs are only related to first column of output and so on.

I think fminsearch will not work for my case.

Do you know any other command?

Thanks in Advance!

[Torsten]

My guess was that you want to fit

Ymes(i,j)

against

(1-p)*0.003* x1(i,j) + 275*p*x2(i,j).^4

for 1<=i<=3000 and 1<=j<=3.

This is what the code I suggested does.

If you want to do something else, you will have to clarify.

Best wishes

Torsten.