A quick check in Prime and it works.
I wish we could use just one for everything like in SMath. Backward compatibility might be the problem here but then, maybe even not.
I have to agree. It took me a while to get my head around the difference in behavior.
Alans way of doing it is basically just the same, but he uses a range variable to link the correct elements of x and ab together rather than relying on MC doing it right or using vectorization.
Both Alan and me have used similar methods to create a vector x based on the range you provided. Of course in both sheets x can be any vector you like, as long as the dimensions of x and ab are the same. For Alans sheet you simply have to provide a range variable c to count the vector elelements.Guess its a matter of personal taste if you use a range or rather vectorization. It may also be dependent on for whom the document is made for and if that person is more likely to understand whats going on with the range or the (maybe not that obvious) vectorization.
Anyway, here is an example which uses random values for both x and a and you see both methods side by side. Note that I used a*x anmd also x*x instead of x^2, so vectorization is mandatory.
Hi Werner,
Thanks! it's clearer to me now. By the way, I think you might have forgotten the link 
Newin
Obviously wrong? Why? The answer is exactly what I would have expected (after some thinking about it, I admit ;-). Vectorizing is not recursively digging in nested vectors but would vectorize the first/outer vector only, which is the 2x1 argument vector containing the larger vectors T and sin(T) as elements. So as the vector contains 2 elements, also the output would consist of just two elements. Every element is fed into x3 and the function (which expects a vector argument anyway) is evaluated. The result is the vector (1,1), as for both, T and sin(T), the second element is smaller than 0.2 and so the function x3 returns true (=1).
Mike Armstrong wrote:
Consider that a bug?
No. What exactly do you mean that is a bug? The result we see in the example you posted is exactly as it should be IMHO. Its no bug that MC doesn't go into nested vectors. I don't think that would be practical and I wouldn't want it to do that. Vectorization is a powerful operator but it can't solve all problems. E.g. it often is necessary to evaluate a function with every combination out of 2 different vectors. We can see how the result should look by using range variables (sic!) but the result can't be assigned to a variable and so isn't useful.
To make it useable for further calculations we can't use vectorization either so we will have to build our tables using range variables for indexing the vectors or using a small program.
So, no, I don't consider the implemenation of the vectorize operator as a bug, I just wish we would not have to distinguish between ranges and vectors and could use only one (vector) for both as in SMath.
By the way, I think you might have forgotten the link
Correct, here it is: http://www.ptc.com/WCMS/files/131949/en/Range_Variables_Vs._Vectors_in_Mathcad.pdf
AFAIK Stuart had written two essays on the subject but at the moment I am not able to find the second one.
