Tall Matrix Index Extraction

To All:

I have a varying size model of the dimensions (L x nx)

L = number of data points

nx = number of variables per data point

In the algorithm that I am using, it moves the data around to support a search of the data space.

A small example could be this:

x = [1,2;3,4;5,6];
id =[1,6;2,4;1,5];
xnew = x(id);
disp(xnew); %[1,6;3,2;1,4];

Now while this isn't too difficult, my state "x" is in reality a tall matrix.

xtall = tall(x);
xnewtall = x(id);     % crashes!

Does anyone know how to do this without having to do every individual point independently?

Answer 1

[SaiDileep Kola]

Since tall matrices are challenging in terms of memory, how about parsing arrays "xtall" and "id" and using with appropriate offsets, up on that par-for would run efficiently, this could be a bit better thatn individual point independently.

Some task on similar problem is done here.

[Blake Van Winkle]

To SaiDileep:

I appreciate the thought. This approach does reduce the time required, but it is still less efficient than if there was a total dimensional indexing solution.

Thanks,

Blake