Because for cities like Paris, the matrices are huge and almost don't fit into 32GB of memory. Looping over subsets and saving intermediate results before a final re-assembly step will be more memory-efficient, and will also allow a progress indicator.
Because for cities like Paris, the matrices are huge and almost don't fit into 32GB of memory. Looping over subsets and saving intermediate results before a final re-assembly step will be more memory-efficient, and will also allow a progress indicator.