Limits on reconstruction of dynamics in networks

Area of Science:

• Network dynamics
• Systems theory
• Control theory

Background:

• Physical systems are often modeled using network dynamics.
• Observability is crucial for understanding and controlling these systems.
• Limited or noisy observations pose challenges for state reconstruction.

Purpose of the Study:

• To define and analyze an observability condition number for network dynamics.
• To assess the feasibility of trajectory reconstruction from partial, noisy observations.
• To guide the selection of optimal observation subsets for network monitoring.

Main Methods:

• Definition of an observability condition number based on network structure and dynamics.
• Calculation of the expected distance to the nearest correct trajectory under varying noise levels.
• Analysis of how this distance varies across unobserved nodes.

Main Results:

• The observability condition number quantifies the difficulty of trajectory reconstruction.
• Reconstruction becomes infeasible when the condition number is sufficiently large.
• The condition number's behavior highlights vulnerabilities in unobserved network parts.

Conclusions:

• The proposed condition number provides a quantitative measure for network observability.
• It enables informed decisions on sensor placement for effective system monitoring.
• Understanding observability limitations is key for reliable state estimation in complex networks.