After seeing that individual components of the sublists (i.e., triples) need to be changed, I decided to ask one of the Maple programmers about the efficiency of this approach. Turns out that changing an element of a list requires Maple to recreate the full list. Not very efficient.
The suggestion was to use an Array or Matrix of size 3x120. Each column would represent one of the triples, and the 120 columns would be the 120 "vectors" you originally asked about. It's much more efficient computationally to change an element in an Array or Matrix. It all has to do with allocation to memory locations, pointers, etc.
If you need the "linear algebra" properties of these 120 vectors, use a Matrix. If the data structure is just a storage for 360 numbers, an Array will suffice.
Create the Array with A:=Array(1..3,1..120);
Assign a value to the location row=2, column=25 with the syntax A(2,25):=x;
Access this value with A(2,25);
If instead, a Matrix is appropriate, use B:=Matrix(3,120);
Assign a value to location row=2, column=25 with the same syntax as in the Array.
A single column can be extracted from the Matrix with the command LinearAlgebra:-Column(B,25);
Of course, if a number of commands from the LinearAlgebra package are to be used, simply load the package via the syntax with(LinearAlgebra) so that calls to commands in the package don't need the prefix LinearAlgebra:-
It would be useful to know what kind of "advanced calculus functions" are to be applied to the columns of the Array or Matrix. Some of these functions might reside in the VectorCalculus package. Recall that I pointed out earlier that a VectorCalculus Vector carries as an attribute its coordinate system. This is not the case for LinearAlgebra Vectors, so a conversion might be necessary. However, if the vector operations are just dot and cross products, it is not necessary to go to the VectorCalculus package. We can give better advice if we know more specifically just what operations you need to inflict on your vectors, or triples, as the case may be.