Nat. Commun. 10, 3895 (2019)

Commuter flows in cities are regular and predictable. This should come as no surprise to anyone who has witnessed crowds jostle for space on a city-bound train platform in the morning. A widespread approach to capture these regularities mathematically considers flows in gravitational terms, where people behave as masses. However, its somewhat ad-hoc nature has made researchers look for alternatives.

Credit: Alex Segre / Alamy Stock Photo

Mattia Mazzoli and co-workers have now revealed that the gravitational picture is more rigorous than it may seem. From Twitter and census data, they divided several cities into 1 × 1 km2 cells and constructed an empirical vector field pointing in the average direction of commuter flow for each cell. Such a vector field was shown to satisfy the divergence theorem and have vanishing curl. This made it possible to rigorously define a potential field, reducing the dimensionality of the problem and highlighting basins of attraction, offering help for urban planning.