We apply the Metropolis-Hastings algorithm to efficiently sample from arbitrary paths distributions in a general network. Paths can be generalized into all-day travel plans through, e.g., an appropriate network expansion. The Metropolis-Hastings algorithm creates a Markov chain of paths, which resembles DTA simulations that can also be phrased as Markov chains. A combination of both chains could lead to better understood DTA simulations that avoid the arbitrariness of current choice set generation procedures.
Giuseppe Carleo, Riccardo Rossi, Dian Wu
Fabio Nobile, Juan Pablo Madrigal Cianci