Maintaining Ferment

Mohak Goyal, Debasish Chatterjee, Nikhil Karamchandani, D. Manjunath

Abstract

We consider maintaining the control of opinions over a social network with the objective of keeping the real valued, scalar opinion of every agent above a predetermined ferment level over a finite time horizon. The opinion dynamics on the social network follow a variation of the well known discrete-time Friedkin-Johnson model. The goal is to determine an optimal control trajectory that minimizes the total cost. Using the Pontryagin maximum principle, we first develop an algorithm to obtain an optimal control trajectory. We then argue that the optimal state-action trajectories exhibit the `turnpike' property, i.e., the state and the control variables remain close to the equilibrium for all but a finite number of time instants. We then exploit this property to develop heuristics to select the set of nodes that will receive the control inputs and demonstrate their effectiveness. Several numerical results illustrate the behavior of the costs of maintaining ferment for different settings.
BibTeX
@misc{goyal2019maintaining,
  author = {Mohak Goyal and Debasish Chatterjee and Nikhil Karamchandani and D. Manjunath},
  title = {{Maintaining Ferment}},
  year = {2019},
  eprint = {1912.06343},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1912.06343},
}
Figure