Calling functions inside another function

Hello, I have a problem with calling a function inside another function. I have a function PLearn that looks like this:

   function LPar = PLearn(input,desOut,Par) 
          p = perc_learn(Par{1},input,desOut,Par{2},Par{3}); 
          LPar = {p}; 
   end

Then I have a function Err which gets in a parameter Name name of the function it should call (PLearn). This is the code for Err:

   function E=Err(Name,NameL,Param,Tr,DTr,Ts,DTs)
        %Lpar=PLearn(Tr,Dtr,Param);
        Lpar=@Name(Tr,Dtr,Param); 
        dif=0;
        for i = 1:size(Ts,2)
           input=Ts(:,i); 
           %Out=@NameL(LPar,input);
           Out=PRecall(Lpar,input);
           dif=dif+abs(DTs(i)-Out);
        end
        % To make sure that err is a value between 0 and 1
        % I will divide the number of miscalculated vectors by the total number of input vectors.
        E=double(dif/size(Ts,2));
    end

I call Err with these parameters:

   Err(PLearn,PRecall,p,In,InO,TIn,TInO).

Now Err calls PLearn by passing pararametrs p,In and InO into it. It gives me this error:

In D:/FreematWorkspace/k-fold/PLearn.m(PLearn) at line 2
Error: Undefined function or variable Par

But if I run PLearn straight (by typing PLearn(In,InO,P)) with the same parameters it calculates the result no problem. So I don't get why it doesn't work through Err. I thought the problem might be with the use of name and calling the function through @ but even if I just call PLearn straight up from Err (the commented version) it still gives me the same error. I would appreciate any help on how to fix this.

Thank you so much

Answer 1

[Steven Lord]

Lpar=@Name(Tr,Dtr,Param);

That's not the right way to define an anonymous function.

 Err(PLearn,PRecall,p,In,InO,TIn,TInO).

This will try to call PLearn with 0 input arguments and pass the output from that call into Err. [Actually, it probably should throw an error about the trailing period first.]

If you want to "pass one function into another", don't pass the name of the first function into the second. Instead pass a function handle to the first function into the second. [That function handle could be an anonymous function.]

Your call would become:

Err(@PLearn, @PRecall, p, In, InO, TIn, TInO)

To evaluate that function handle inside Err:

function E=Err(fun,funL,Param,Tr,DTr,Ts,DTs)
value = fun(Tr, DTr, Param);
% etc

Search the documentation for the terms "anonymous function" and "function handle" for more information.

[Simona Ondrckova]

Ohh I see now. Thank you. I have adjusted that but it still finds an error, this time it says:

Error: Unsupported type for call to toArrayVector
at line 2 in PLearn.

And again if I just run it straight through PLearn it has no problems. My Err function now looks like this:

function E=Err(Name,NameL,Param,Tr,Dtr,Ts,DTs)
    Lpar=Name(Tr,Dtr,Param); 
    dif=0;
    for i = 1:size(Ts,2)
       input=Ts(:,i); 
       Out=NameL(Lpar,input);
       dif=dif+abs(DTs(i)-Out);
    end
    % To make sure that err is a value between 0 and 1
    % I will divide the number of miscalculated
    % vectors by the total number of input vectors.
    E=double(dif/size(Ts,2));
end

and I call it with:

Err(@PLearn,@PRecall,p,In,InO,TIn,TInO)