How to add values from two different arrays into a matrix

I'm trying to figure out how to add values from a 1Xn double to a mX1 double such that it populates a matrix of mXn by adding consecutive column values to the entire row. An example here would be

a = [1 2] (1xn)

and

b = [1;2;3] (mx1)

gives

c=[2 3; 3 4; 4 5](mxn).

My actual example contains much larger a's and b's. In addition, its not clear to me how MatLab views what it calls 1xn double or mx1 double. Does matlab see these as arrays or matrices? Apologies in advanced if my question isn't clear. Thank you!

[Star Strider]

For more on this see Addition of non-uniform dimensional matrices (link).

Answer 1

[Bob Thompson]

Matlab sees them as either arrays or matrices, depending on the type of operations you use them for. Unfortunately I do not do much work with matrices so I don't typically use their operations, but I know that is essentially how you define the difference for matlab.

My suggestion for adding in that manner would be to use a for loop to only add one row or column at a time. There is probably a faster way of doing this, but I am not an expert on matlab matrix operations.

for i = 1:length(b);
    results(i,:) = a + b(I);
end

[Emilio Alverio]

Wow, this was exactly what I needed. The only correction was

for i = 1:length(b); results(i,:) = a + b(i); end

Thank you very much!

Answer 2

[James Tursa]

On later versions of MATLAB:

c = a + b;

On earlier versions of MATLAB:

c = bsxfun(@plus,a,b);

In MATLAB, variables have a size that contains at least two dimensions always. It could be that one or more of those dimensions is 1, in which case there are standard names used:

scalar --> a "1 x 1" variable
vector --> a "m x 1" column vector or "1 x n" row vector
matrix --> a "m x n" variable
array  --> a "m x n x p x ..." variable

To add (+) or subtract (-) or element-wise multiply (.*) or element-wise divide (./) or element-wise take power (.^) two variables, they need to be of conforming size. In older versions of MATLAB, that meant the variables needed to be exactly the same size or at least one of them needed to be a scalar. To do the "expansion" type of operation you wanted needed a special function such as bsxfun. On later versions of MATLAB, however, this expansion happens automatically. That is, any dimension that is 1 is "virtually expanded" to match the dimension of the other operand ... it is as if you did appropriate repmat operations on the operands to get them to a minimum conforming size and then did the operation.

[Emilio Alverio]

Thank you for the very thorough answer. I think I did not provided enough information or phrased it in an unclear way. The fellow below figured out what I was after. I apologize for not being clearer. Thank you again.

[James Tursa]

"... I think I did not provided enough information ..."

No. In fact you provided all the information needed to solve the problem. The two methods I posted are the standard way of doing this in MATLAB without writing explicit loops. They give the same result as the looping method you accepted.