Maximizing Activity in Ising Networks via the TAP Approximation

This so-called Ising influence maximization problem and humans interacting in social networks (Lynn et al. 2017; was originally proposed in the context of social networks Galam 2008), among an array of other social and biological (Lynn and Lee 2016), where it has a clear practical applications (Kapur 1989; Phillips, Anderson, and Schapire interpretation: If a telephone company or an online service 2006; Mora et al. 2010; Lezon et al. 2006). Broadly speaking, wants to maximize user activity, how should it distribute its the maximum entropy principle allows scientists to formalize limited marketing resources among its customers? However, the hypothesis that large-scale patterns in complex we emphasize that the broader goal--to develop a unifying systems emerge organically from an aggregation of simple control theory for understanding the effects of external influence fine-scale interactions between individual elements (Jaynes in complex systems--could prove to have other important 1957). Indeed, intelligence itself, either naturally-occurring applications, from guiding healthy brain development in the human brain and groups of animals (Hillis 1988) (Goddard, McIntyre, and Leech 1969) and intervening to alleviate or artificially constructed in learning algorithms and autonomous diseased brain states (Goddard 1967) to anticipating systems (Mataric 1993; Namatame, Kurihara, and trends in financial markets (Mantegna and Stanley 1999) and Nakashima 2007), is increasingly viewed as an emergent preventing viral epidemics (Pastor-Satorras and Vespignani phenomenon (Lévy 1997), the result of repeated underlying 2001).