I know v4 is far along and probably not on the anvil for new features, but this is a small one.
Descriptors on matricies. They would be helpful. Really, really, really helpful.
They would be helpful enough if they weren’t visible in the model browser, but just hidden as a property of the matrix.
They would be especially helpful, however, if there were a right column in the Model Browser that could display them in the same way that the column (row) to the right of (below) items displays item descriptors.
Use case: In nearly every model I make, each matrix has single, defined unit or dimensionality. Keeping track of these in an organized way, and being able to use them in scripting, would be really, really huge. It would make possible things like automatic dimensionality checks.
Thanks for considering it.
I think Dom’s idea is very interesting. I agree that it would be excellent if Quantrix gained the ability to check that matrices are of appropriate dimensionality with respect to matrices that they depend upon. For instance, if Matrix1 has categories A and B, and Matrix 2 has categories B and C, and Matrix3 is formed by the following formula:
: = Matrix1:: * Matrix2::
then, in almost every conceivable circumstance, Matrix3 MUST have categories A, B, and C. If it is missing any of these, then the modeler has almost certainly made a very large mistake, because Quantrix will silently be defaulting to the first slice of the omitted category.
I see this as the single largest potential source of errors in complex models.
However, in this case, I was actually referring to a different kind of dimensionality. I was referring to physical dimensionality – things like mass, or mass per time, or distance per time squared.
Because I keep each scalar, vector, or matrix value in my model in a separate matrix, each of my /matrices/ has a single dimensionality associated with it. I would like a way to keep track of these and use them programmatically.
Just for consideration as a workaround: The unit or dimensionality of a matrix is given by its categories. So you could develop a script which creates (if the matrix has not been created yet) or updates (if this test matrix exists already) a matrix listing the
– names of all matrices in the model
– The category names in these matrices
Further, you could create a matrix including the “should” unit or dimensionality. Then you could compare the “should” matrix with the “output” above with formulaes or a script.
It’s just a rough thought. Let me know what you think.