Starting point of this MATH+ student research group is the Vahana framework for large-scale
agent-based models, i.e. models that define a (social) system at the micro level of actors and
their interactions with each other and with their environment. Vahana represents an agent-based
model as an extended synchronous graph dynamical system (SysGDS), where agents are nodes
in the graph and edges determine neighborhoods or interactions. To run large-scale agent-based
models, a simulation must be parallelized, that is, the graph must be partitioned, so that each
computational process is only responsible for updating the states of a subset of the agents. This
research group seminar aims to improve the graph partitioning strategies of Vahana.
 

In this course, small groups of students shall work on
• implementing an example ABM using Vahana; the programming language is Julia
• obtaining an overview of graph partitioning algorithms and the study of one particular algo-
rithm in view of Vahana
• adjusting an algorithm to the extended SysGDS structure and implementing it, again in Julia