Multiple Outputs of a function into a single vector

I am a new Matlab programmer, and am familiar with functional languages such as python. In my university Matlab course, we are required to write test functions similar to this one:

function [x,y] = f(a, b)
    x = a;
    y = b;
end

This function is called with a number of different parameters, and I'd like to store them in one large array. Previous answers I've found lead me to this solution:

returns = zeros(4, 2);
[returns(0,0), returns(0,1)] = f(a0, b0);
[returns(1,0), returns(1,1)] = f(a1, b1);
[returns(2,0), returns(2,1)] = f(a2, b2);
[returns(3,0), returns(3,1)] = f(a3, b3);

but when "returns" has more than two outputs, this gets ugly very fast. The only other solution I've found instead breaks this into two lines:

returns = zeros(4, 2);
[A, B] = f(a0, b0);
returns(0, :) = [A, B];

[A, B] = f(a1, b1);
returns(1, :) = [A, B];

[A, B] = f(a2, b2);
returns(2, :) = [A, B];

[A, B] = f(a3, b3);
returns(3, :) = [A, B];

But if I merge those two lines, it broadcasts A into every element of the vector. In Python, I could re-cast the output as a numpy array with almost identical syntax. Is there a similar function in Matlab? Every output has the same data type. I'm aware this can be solved with for loops, but I'm interested if a functional solution exists.

Answer 1

[Walter Roberson]

There is no functional solution in MATLAB. The only way to capture multiple outputs is an assignement statement. The assignment can be to an expansion such as

[returns{1:5}] = f(a3, b3)

but you cannot gather the outputs "in-line" like

min({f{a3,b3)})  %will not work to capture multiple outputs

[Joseph]

This works! Its unfortunate I can't use full functional syntax, but this solves my immediate problem in a very clean way.

returns = cell(4, 2);
[returns{1, 1:2}] = f(a1, b1)
[returns{2, 1:2}] = f(a2, b2)
[returns{3, 1:2}] = f(a3, b3)
[returns{4, 1:2}] = f(a4, b4)
returns = cell2mat(returns)

I want to highlight James's solution for anyone else stumbling on this answer. This one works for my workflow, but his is also a very good answer. I wish I could accept both.

Answer 2

[James Tursa]

You could modify f( ) to return a vector. Or if you didn't want to modify f( ) you could create a helper function that does this. E.g.,

function v = f2(a,b)
[A,B] = f(a,b);
v = [A,B];
end

Then just call f2( ) instead of f( ).

Another approach is to have the helper function use the nargout/varargout feature to detect how many outputs are requested by the caller. E.g.,

function varargout = f2(a,b)
[A,B] = f(a,b);
if nargout == 2
    varargout{1} = A;
    varargout{2} = B;
else
    varargout{1} = [A,B];
end
end

[Joseph]

This is extremely handy. I will definitely be using this in the future. I wish I could accept both answers, because they're both very situational.