Complex Networks Dynamics and Control
Complex networked systems appear in a broad range of Engineering, Social and Biological applications. Our research aims at developing suitable dynamical models and distributed control methods.
Control of Multiagent Networks
How can we control large-scale, complex networks?
Task: Given a multiagent system, representing a complex engineering or technological network, how can we control it in an efficient way? For instance, where should control inputs be placed in order to minimize the cost for control? What kind of algorithms can be used to impose a global emerging behavior in a distributed way, i.e., allowing only local exchange of information among the agents?
How: We investigate qualitative and quantitative controllability analysis and control design based on network topology and on the optimization of Gramian-based centrality metrics. Distributed control schemes not relying on any centralized authority can be efficiently developed for most multiagent networks.
How can a community reach a collective decision?
Task: For us, a social network is a community of agents exchanging opinions. The interaction among the agents is modeled as a nonlinear dynamical system on a network. In opinion forming models, the agents can often cooperate or compete, which leads naturally to representing the community as a signed network.
How: We investigate the dynamical behaviour of the community, for instance existence and stability of the equilibria of the network (corresponding to the decisions taken by the community) using tools of nonlinear dynamics and network theory.
Can we reconstruct an intracellular network?
Task: In each cell, more than 20,000 proteins and genes work together to determine its behavior. In complex diseases like multiple sclerosis, the nominal regulatory interactions are disrupted, hence understanding the networks and their disease-induced alterations is a critical issue. To unravel these interactions we can use data from several types of experiments.
How: We aim to use tools from regularized linear regression and dimensionality reduction on multi-omics data to identify the main intracellular networks involved in the differentiation of Th-cells, a subset of the immune system playing a predominant role in multiple sclerosis.
Figure below: The Th-cells family, the balance between the expression of these cell subtypes is believed to be a key factor in multiple sclerosis.