matlab differing argument structures

Question

I want to write a function getNewProjectionImageSize that has two different kinds of argument structures. One is [Wp, Hp] = getNewProjectionImageSize(Wo, Ho) and the other is Hp = getNewProjectionImageSize(Wp). On my research I couldn't find how to do it. Ie this link, doesn't explain it.
Is it possible with not too much effort? In standard matlab functions, ie interp2 there are different argument structures: interp(V) interp(X, Y, V, Xq, Yq) etc.
The only solution that came to my mind is a more common argument structure [Wp, Hp] = getNewProjectionImageSize(W, H) with H as an optional argument (using nargin), leaving the interpretation of W and H to the user. But I would prefer the first way if it is possible.

You can look into the source code of matlab to check how they implement it (right click > Open "interp2"; or ctrl+D shortcut). — m7913d, Sep 12 '17 at 09:40
Note that the different arguments (length) is just a matter of documentation. — m7913d, Sep 12 '17 at 09:42

Luis Mendo · Answer 1 · 2017-09-12T09:57:32.807

Use nargin and nargout to know the number of input and output arguments with which the function has been called, and then use varargin and varargout to access the inputs or define the outputs.

You also need to decide what to do if the function is called without any output arguments. In many functions that case is treated as if the call had been with one output.

Here's an example.

With 1 input and 0 or 1 outputs: the function outputs the element-wise square of the input.
With 2 inputs and 2 outputs: the function outputs the element-wise sum and product of the inputs.

Code:

function varargout = f(varargin)
if nargin==1 && nargout<2 % nargout<2 covers the 0-output case, which
    % is interpreted as 1 output
    varargout{1} = varargin{1}.^2; % compute first (and only) output
elseif nargin==2 && nargout==2
    varargout{1} = varargin{1} + varargin{2}; % compute first output
    varargout{2} = varargin{1} .* varargin{2}; % compute second output
else
    error('Incorrect number of inputs or outputs')
end

Examples:

>> y = f([10 20 30]) % 1 input, 1 output
y =
   100   400   900

>> f([10 20 30]) % 1 input, 0 outputs
ans =
   100   400   900

>> [a, b] = f([10 20 30], 4) % 2 inputs, 2 outputs
a =
    14    24    34
b =
    40    80   120

>> y = f([10 20 30], 4) % 2 inputs, 1 output
Error using f (line 9)
Incorrect number of inputs or outputs

Justin · Answer 2 · 2017-09-12T10:02:19.090

Use varargin as your input argument and varargout as the output. These allow you to accept/return variable numbers of inputs and outputs. E.g. something like this:

function varargout = getNewProjectionImageSize(varargin)

if nargin==1
    % Have passed in one input argument, Wp
    Wp = varargin{1};
    % Calculate Hp here...
    varargout{1} = Hp;
elseif nargin==2
    % Have passed in two input args, Wo, Ho
    Wo = varargin{1};
    Ho = varargin{2};
    % Calculate Wp and Hp here ...
    varargout{1} = Wp;
    varargout{2} = Hp;
else
    error('Must supply one or two input arguments');
end

end

An alternative would be to pass in named arguments with key/value pairs, so your calls would look like:

[Wp, Hp] = getNewProjectionImageSize('Wo',Wo, 'Ho',Ho)
Hp = getNewProjectionImageSize('Wp',Wp)

There's some description of how to do that here using inputParser.

matlab differing argument structures

2 Answers2