LabVIEW
Possible bug (LV2011) with Silver type Real Matrix
It appears that the control/indicator for a Real Matrix is different in the Silver pallete than on the Modern or Classic palletes. I don't know enough about Matrix data type to be able to figure out this issue, but in the attached VI you should be able to see the problem- when wiring a matrix into the Silver version Real Matrix indicator, you get a coercion dot. If you right click and do Create Constant, you will get a different type of wire. It looks similar to the 2-D array wire type, but it isn't- it appears to be something different altogether.
This is an issue when trying to multiply matrices, as the Multiply command isn't recognizing the Silver matrix as a Matrix data type- it doesn't multiply the matrices poperly. Note in the example VI I posted that the Multiply command does NOT properly multiply Silver matrices.
That should work. I simply changed over to using all Modern matrices, as none of the matrix VI's I'm using have any UI components, but it did take me a while to track down why none of them worked 🙂 I'd still like to see this fixed, as it's not obvious that the multiplication doesn't work the way it's expected to. As you said, the Silver matrices aren't actually matrices (as Labview defines them) but also aren't true arrays, so noticing the correct data type is a bit tricky!
The issue here isn't related to storage and performance. The performance issues (I believe) come from converting data from one type to another that is stored differently on the system. If I remember correctly, a coercion dot means that a copy of the data is made, so for instance an integer getting converted to a double means that there are now two copies of that variable. When running data-intensive programs, that doubling of each data point can result in a lot of memory consumption (this is my understanding, at least).
The problem here is that math on arrays is totally different from math on matrices. When multiplying an array, it is done element-by-element, but with a matrix, it should be done using proper matrix multiplication. If the data type is incorrect, you won't get memory problems, but you will get bogus math results from the multiplication operation.
