Yes, stillyslalom, it’s intentional (unavoidable) that the 2D (3xN element) array for each particular value of (j,k) could (and probably will) be a different length. That’s why I’m thinking that a tuple with (Nj)x(Nk) elements, and each element being a 3xN appendable array, might be the way to go. Do you have any ideas on the proper Julia syntax for some way to do that?
Basically, I have a huge data set that must be cut into i slices to fit into computer memory. For each “ith” piece, I then have to run (Nj)x(Nk) analyses, covering (Nj)x(Nk) parameter combinations, and for each combination I get an Nx3 (or 3xN) array. Then, I have to glue each of these Nx3 arrays to the results from the previous data slices, making sure to add the results for each value (j,k) to the results for the same value of (j,k) from the previous slice.