Code used the implement the mean-field Ising Influence Maximization algorithm at both the level of a full network and at the level of blocks within a graph with block structure.
Plot showing the estimate of the mean-field optimal control field on a two block SBM.
The code was validated using Python 3.7.4 on an Anaconda distribution (conda version : 4.8.3). A full list of the packages installed can be found in the environment.yml file. This can be used to construct an identical anaconda environment to that used to perform the simulations. See the Anaconda documentation for more details.
To install run: python setup.py install
If this does not work then try running: python setup.py develop
This module is used for the numerical simulations in the paper:
Garrod M., and N. S. Jones. Influencing dynamics on social networks without knowledge of network microstructure. In preparation, 2020.
This code was also used for the numerical simulations in my PhD thesis "Influence and ensemble variability in unobserved networks" (available at: https://doi.org/10.25560/83107). The simulations used an earlier version of this code base, however, the implementations of the algorithms used are essentially the same.
The implementation of the Ising influence maximisation algorithm builds upon ideas described in the paper:
Lynn, Christopher, and Daniel D. Lee. "Maximizing influence in an ising network: A mean-field optimal solution." Advances in Neural Information Processing Systems. 2016.
The code used for the projection onto the positive simplex is based on the ideas presented in the paper:
Large-scale Multiclass Support Vector Machine Training via Euclidean Projection onto the Simplex Mathieu Blondel, Akinori Fujino, and Naonori Ueda. ICPR 2014.
and can be found at: https://gist.github.com/mblondel/6f3b7aaad90606b98f71