array dimensions equivalence check?

You would think I would know this by now, but I never had reason to program it before.

Is there a good idiom for efficiently checking that two arrays are the same size or are row and column vectors of the same length?

For example, this kind of check would be used to find out whether two arrays are assignment-compatible, or to find out if two arrays are compatible for plot() purposes.

The code I have come up with feels overly complicated:

szA = size(A);
szB = size(B);

if isequal(szA,szB) || ...
   (sum(szA==1) == 1 && sum(szB==1) == 1 && ... 
    numel(szA)==2 && numel(szB)==2) && ...
    prod(szA) == prod(szB))

Which is to say, the sizes can be identical, or they can both be 2 dimensional with exactly one of the dimensions of each being 1, and the number of elements the same between the two.

This kind of check must be programmed all over the place. Perhaps there is a utility routine that I overlooked?

[Paulo Abelha]

Hello Walter,

I have implemented a function that might help you:

https://uk.mathworks.com/matlabcentral/fileexchange/59168-checknumericarraysize--a--dim-sizes--

Best regards,

[Bruno Luong]

Your test return false for

A = [];
B = zeros(1,0);

however plot works just fine

plot(A,B)

or Assignment

X = rand(); 
X(1,[])=X([],[]) % B on lhs, A on rhs

X = 0.6365

Answer 1

[David Young]

One alternative:

isequal(size(A), size(B)) || (isvector(A) && isvector(B) && numel(A) == numel(B))

[Walter Roberson]

Thanks, that looks better than what I had.

[Adam Wyatt]

This will return true if A is a row vector and B is a column vector (or vice versa). If you care about that:

isequal(size(A), size(B)) || ( ...
    ((iscolumn(A) && iscolumn(B)) || (isrow(A) && isrow(B))) ...
    && isequal(numel(A), numel(B)))

[Ahmed]

This seems not working for 3-D arrays or higher. Therefore, I implemented a more general function below.