Theoretical Interdisciplinary Physics

Under the guidance of Professor Yi-Cheng Zhang, the Theoretical Interdisciplinary Physics group explores a variety of phenomena in interdisciplinary physics. According to the statistics from Google Scholar, Prof. Zhang’s publications have received more than 10,000 citations and his H-index is more than 40 (see details on his personal Google Scholar web page). The group participates in EU-supported projects: GROWTHCOM and NESS (for more information see Projects). Our recent project QLectives has made its summarizing book available for free download.

Our current research interests 

• Information Filtering

  While information used to be a scarce resource in past, with the advent of the Internet and the World Wide Web, we have more choices than ever before. Recommender systems are one of the key tools to cope with the information overload. They use data on past user preferences to predict possible future likes and interests. Building on network representations of the input data and processes well known from physics (random walk and heat diffusion in particular), we develop recommendation methods and test them on large real datasets (coming from the well known web services such as Amazon.com and Delicious.com). We also attempt to tackle the higher level problem of the long term influence of information filtering tools on information diversity.

• Complex economic systems

  After introducing the classical idealized model of trading behavior, the Minority Game, our attention has gradually shifted to models where interacting economic agents give rise to phenomena as diverse as products' quality differentiation and bankruptcy avalanches. Agents with imperfect information and competition of firms in markets of this kind is studied too.

• Investment optimization

  In 1956, J. L. Kelly proposed a new strategy for optimizing a portfolio of risky assets which uses some of the most important results of the information theory. We study this strategy, known as the Kelly or growth-optimal portfolio, as well as the influence of limitited information (about the assets), correlations (between asset returns) and other constraints (transaction fees, etc.) on its applicability.