Research Avenues

Abstract

The core MATSim architecture, where agents learn utilities for plans, was originally derived from the field of Complex Adaptive Systems (CAS; e.g., Axelrod, 1984; Holland, 1992; Hraber et al., 1994; Palmer et al., 1994) (also see Section 46.1 of this book). Arthur (1994) addresses a coordination problem where agents receive a payoff only when less than 60 out of 100 go to an event. He addresses this by first generating a large number of heuristic predictors for the next round’s attendance, such as “same as in last round” or “trend from last four rounds”. He next gives each agent a