It is currently assumed that all data sections are presented as tabular data, which is then conveniently parsed using np.genfromtxt.
This may be too restrictive for more general VRP variants. For instance, the site-dependent VRP restricts some vehicles from visiting a subset of clients. If we keep the assumption that data sections are tabular, then this would require a $N$ times $K$ matrix with 0 and 1's to denote whether client $i$ can be visited by vehicle $j$ or not.
A more compact option would allow a list of indices. For example, 1 1 2 3 4 could mean that vehicle 1 is not allowed to visit clients 1 2 3 4.
It is currently assumed that all data sections are presented as tabular data, which is then conveniently parsed using
np.genfromtxt.This may be too restrictive for more general VRP variants. For instance, the site-dependent VRP restricts some vehicles from visiting a subset of clients. If we keep the assumption that data sections are tabular, then this would require a $N$ times $K$ matrix with 0 and 1's to denote whether client $i$ can be visited by vehicle $j$ or not.
A more compact option would allow a list of indices. For example,
1 1 2 3 4could mean that vehicle 1 is not allowed to visit clients 1 2 3 4.