Writing a function handle in a for loop

Hello everybody,

As you can see below I want to create a function handle temp_control, however x(1),x(2),x(3) and given control has different permutations in different cases. So rather than writing it manually as below; ( given_control is a number)

temp_control = @(x) [given_control,x(1),x(1),x(2),0,0,0,x(3),0]';

temp_control = @(x) [x(1),given_control,given_control,x(2),0,0,0,x(3),0]';

temp_control = @(x) [x(1), x(2), x(2), x(3), 0, 0, 0, given_control, 0]';
 
temp_control = @(x) [x(1),x(2),x(2),given_control,0,0,0,x(3),0]';

I want to put x(1), x(2) and x(3) to the function handle temp_control, in a for loop. I want this for loop to handle the permutation. The function handle is used later in fminsearch.

fun = @(x) OP_residual(aircraft,ASZ,phi,theta,temp_control(x));  
[x_out,residiual] = fminsearch(fun,x_in);

I have piece of code to try this;

id = 8; %can be 1,2,4,8
possible_givens=[1,2,4,8];

possible_givens = possible_givens(possible_givens~=id);

temp_control = cell(1,9);

for j=1:length(possible_givens)
    temp_control{possible_givens(j)} = sprintf('x(%d)',j);
end
temp_control {3} = temp_control{2};
temp_control{id} = num2str(given_control(id));

temp_control = str2func(temp_control);

Obviously str2func doesnt work. I cannot solve the problem. Would appreciate if someone can help :)

[dpb]

You would have to build a string representation of the function for each case (as you've started on an attempt to do)...however, there are some "issues" as is...first one I ran into is

id = 8;                             %can be 1,2,4,8
possible_givens=[1,2,4,8];
possible_givens = possible_givens(possible_givens~=id);
temp_control{id} = num2str(given_control(id));
...

but given_control is undefined. You say it is a number so I tried:

given_control=randi(100,1,max(possible_givens));     % arbitrary integers

and then proceeded to try to finish your code to see how far had gotten/what terms looked like and

>> temp_control{id} = num2str(given_control(id));
Index exceeds matrix dimensions. 
>> id
id =
     8
>> whos given_control
  Name               Size            Bytes  Class     Attributes

  given_control      1x4                32  double              

>> max(possible_givens)
ans =
     4
>> 

You early logic doesn't work as written -- what is the actual intent there? Who does exist for the given case; clearly you've mixed up an indexing array with the data to which it is supposed to refer to I believe.

[Mehmet Velioglu]

given_control is actually a vector of doubles of size 9. sorry for the misexplanation. but it does not actually matter, you can substitute given_control(id) with a random number in your trial.

You can try this:

given_control = ones(1,9);

id = 8; %can be 1,2,4,8
possible_givens=[1,2,4,8];

possible_givens = possible_givens(possible_givens~=id);

temp_control = cell(1,9);

for j=1:length(possible_givens)
    temp_control{possible_givens(j)} = sprintf('x(%d)',j);
end
temp_control {3} = temp_control{2};
temp_control{id} = num2str(given_control(id));

temp_control = str2func(temp_control);

The intent is to create a function handle like;

temp_control = @(x) [x(1), x(2), x(2), x(3), 0, 0, 0, given_control, 0]';

in another function I write the id values will change, so temp_control needs to be in different combinations. That is my problem, I dont want to write an if statement for each case.

[dpb]

The key is you have to have a definible logic that can be turned into an algorithm that determines how many and what is the Nth argument of the proposed anonymous function.

I can't tell what is behind the examples at the beginning from which to be able to code the elements. Given that, writing a string expression to pass to str2func would be essentially a trivial exercise.

[dpb]

OK, I studied the pattern in the initial example a little more...looks like the easiest way to implement this would be to store the pieces in a cellstr() array and use some logic to index the order of putting the pieces together based on that position. The kicker that I don't quite see the actual logic behind is which index is to be the repeated one based on the examples, but I didn't have a terribly long time to study it.

If these are the only four cases, in reality probably the best choice is to just store the precise functions in a function handle array and index into that array based on the case in hand--if there's a dynamic into the order or the index of which is the repeated element that is variable, then the need for the algorithm comes back into play.