How to call functions indirectly?

Hi,

I would like to know how to call functions indirectly. For example: lets supose I have a data vector X, the functions F1.m, F2.m and F3.m (which input is the variable X) and I put these functions names in a list:

X=rand(1,100);
Myfuns={'F1','F2','F3'};

What I do today is something like:

Y(1)=F1(X);
Y(2)=F2(X);
Y(3)=F3(X);

What I would like to do is to call these functions using the variable Myfuns. Something like

for i=1:length(Myfuns)
     Y(i)=Myfuns{i}(X);
end

Of course this doesn't work, but is there a way to do something this?

Thank you,

Bernard.

Answer 1

[Daniel Shub]

You can do it with only additions to your code

F1 = @(a)length(a);
F2 = @(b)sum(b);
F3 = @(c)min(c);

X=rand(1,100);
Myfuns={'F1','F2','F3'};

for ii = 1:length(Myfuns)
  eval(['Myfuns{ii} = @(x)', Myfuns{ii}, '(x)']);
end

for i=1:length(Myfuns)
     Y(i)=Myfuns{i}(X);
end

If the functions are contained in files that exist on the path, then you can use str2fcn instead of eval

Myfuns = cellfun(@str2func, Myfuns, 'UniformOutput', false);

The best is to define the function handles directly:

Myfuns={@(x)F1(x),@(x)F2(x),@(x)F3(x)};

[Titus Edelhofer]

Hi,

that's right. Just a side note: if your function has only one input, there is no need to add overhead using an anonymous function, e.g.

F1 = @length;

Myfuns = {F1, F2, F3};

Titus

[Daniel Shub]

That is a good point. I always add the input. I have never timed it to see if it matter. I would hope the JIT would deal with it.

[Bernard Küsel]

Thanks for the answers!

What changes if my functions have more than one input?

For example:

Y=F1(x,y,z)

[Daniel Shub]

Thanks is an upvote and/or an accepted answer. If you read up on function handles, I think it will be obvious what to change.