Build support for multi-dimensional arrays

[SergeStinckwich]

Requirements

The transition rates may have various dimensions.
Examples:

• mu is the transition rate between R and empty and empty and S.
  What is the dimension of this rate?
  mu is a scalar or more generally, a tensor (multi-dimensional array) depending on the number of concerns and heterogeneity. For example, when concerns are status and species, mu maybe be a 2-rank tensor.
• beta is the transition rate between S and I
  What is the dimension of this rate?
  beta is more complex than the other transition rate, because it can be heterogeneous on the state of 2 individuals: the S one and the I one. If there are n concerns, the dimension of beta may be up to 2*n.

Implementation and tests

One solution is to use TensorFlow: https://github.com/PolyMathOrg/libtensorflow-pharo-bindings

Operations to define

• Hadamard product
• regular matrix product