The rapid technical progress in statistical physics in the last few decades broadens its applications to various fields other than traditional physics. In this thesis, we apply the tool of statistical physics to two different problems....[ Read more ]

The rapid technical progress in statistical physics in the last few decades broadens its applications to various fields other than traditional physics. In this thesis, we apply the tool of statistical physics to two different problems.

In the first problem, we examine a multi-agent system in which epochs of alternating winning groups are being generated by the self-segregation of agents. To study the lifetime distribution of epochs, we set up theoretical approaches which have an excellent agreement with simulation results up to 8 decades of probability in the lifetime distribution. We find evidence showing that epochs are stabilized by intelligent agents aggregating in the winning groups.

In the second problem, we investigate resource allocation problems in random networks with finite bandwidths. Algorithms to find optimal solutions are obtained by the chemical potential approach and the message passing approach. A decrease in the bandwidth of links surprisingly leads to an increase of idle links. Both theoretical analyses and simulations show that an increase of connectivity leads to a more homogeneous redistribution of resources, even with vanishing bandwidths of the links.