# Why do I need a Helper when I know the Item?

Hi there,

Quantrix is logical und structured, and a cool thing. But there is one thing I do not understand, and I’m crossing the same thing in a lot of places, when doing SELECT statements, or in other places. I’m still not sure if my brain is twisted or if it is Quantrix ðŸ™‚

I compiled a very simple example to illustrate the point.

In a 3-dim matrix, I want to select one index on one dimension, sum over the second dimension, and do it for all elements in the third dimension.

Easy, as I know the item I want to slice – “S1” – by name.

I do not find a way to do this … not using the item name, not using the descriptor. Everybody is doing this using a helper. Why do I need to convert the address to content, and then search the cell content to get back the address of the cell?? This looks very inefficient to me.

Is there anybody out there who can fix my mind? Or show me how to solve the problem without a helper column (and yes, I know how I can do it if both matricies have the same structure and linked categories, I could sum over all “Sektor”-items, but that’s not what I want to do …

Thanks for your help, Gilbert

Thank you very much **Lyndsey** for your suggestion. This one works perfectly! So, you get the prize for the first attempt.

I admit that while building a simplified model to illustrate my case, I made it too simple.

In the real case, I tried also “SELECT” and “USING AS” in vane. Maybe I’m still a bit too confused about this three variants – with sumif as my third and last option. Sometimes, I find a solution, but there are often cases where I don’t. With neither of the three. What I’m trying to do is to advocate for these “unsolvable” cases …

So, updated example, same game. I changed the following to things, which I think nor “USING AS”, nor “SELECT” can cope with:

- Instead of using the category “X” with the item “S1”, I want to use directly a literal “S1” (this was the case in my real example, though I agree that this is debatable if using a literal is bad style anyway and should not be possible)
- The Years category are not linked (In my real example, I have two different time scales, the one when energy is sold and the one when it’s produced, so I need to match years which are on a different category).

Ok, challenge for everyone: is the problem solvable without the helper matrix also this time?

Gilbert

P.S.: The solution I have in mind, as functions which I feel are needed to solve this kind of cases, and make life much easier:

a) **slice(matrix,item1)**Â = slices out from *matrix* on *item1*, returns a matrix with dim n-1

b) **doalong(matrix, category, function)** Â => takes a *matrix*, and applies the *function* (sum, average, â€¦) along the *category*, returns a matrix with dim n-1