input vector as point to numeric function

So I have this function where I pass a multi-variable function, for example lets regards f(x,y) = x^2+y^2.

Inside the function I do not know in advance how many variables will be (I get that number with nargin)

I am trying to get a random point, considef p = rand(1,2) and then get f(p).

Is there a way to generate random points and then get f(p) for multivariable functions?

Most stuff I have seen here is to do save two vectors as x =[....] , y=[...] and then calculate f(x,y), how can I handle this when my function is n-variable because I cant make n vectors of size 1 just to get a point

Thank you

Answer 1

[Matt J]

It's not clear from your description where the random generation of p is supposed to happen. If it happens prior to calling f(), then the natural thing would be to not split p up into separate arguments. Just define f() to take a single, vector argument:

p=rand(1,n);

f=@(p) norm(p).^2

[Elad Goldenberg]

thank you for the answer but this isn't the answer for me :/

I'll explain more, I make a function that gets a multi variable convex function. it can be f(x,y) = x^2+y^2

it can be f(x,y) = x^3-y^10 and also f(x,y,z,t,s)= x +y +z +t +s and etc. inside the function I'm calculating the gradient and then I want to generate points and evaluate them.

I searched and found that if I send the function f = @(x)x(1)^2+x(2)^2+...+x(n)^2 then I can take a vector [1,2,...,n] and use it as a point and calculate f([1,2,...,n]) now the problem is I cant find the gradients of such function nor knowing how many variable it got using the nargin(f) method..

I don't know if this is the correct way of handling this

[Matt J]

now the problem is I cant find the gradients of such function nor knowing how many variable it got using the nargin(f) method..

Extending my example above,

p=rand(1,n);

function [value,gradient] = f(p)

 N=numel(p); %number of elements

 value=norm(p)^2/N;
 
 gradient=(2/N)*p;

end