Dov Zafrir

Dov Zafrir completed his doctorate at the School of Mathematical Sciences at Tel Aviv University, where his dissertation studied stability and synchronisation in large heterogeneous dynamical networks. The central results characterised spectral conditions on the coupling matrix under which a network of heterogeneous oscillators converges to a synchronised state — extending classical results for homogeneous systems, where Perron-Frobenius theory and Lyapunov methods give clean criteria, to settings where natural frequencies and damping coefficients vary across nodes. The analysis yielded sharp coupling-strength thresholds for synchronisation across a range of random graph models, with the critical value depending on the spectral gap of the graph Laplacian and the degree of heterogeneity in a form that admits direct computation from network observables. He subsequently held a postdoctoral fellowship at the Max Planck Institute for Mathematics in the Sciences in Leipzig, where he worked on graphon limits of coupled dynamical systems and the rigorous derivation of mean-field equations from interacting particle systems on large random graphs.

Following the postdoctoral period, Zafrir developed quantitative convergence results for interacting agent systems on large sparse and dense random graphs to their graphon mean-field limits, establishing convergence rates in terms of graph density and the Lipschitz constants of the local interaction functions. This programme connected the classical theory of propagation of chaos to the graphon framework, providing rigorous justification for mean-field approximations used in the analysis of financial contagion, epidemic spreading, and opinion dynamics on networks. A parallel line of work applied these methods to financial stability questions, characterising contagion thresholds in interbank exposure networks and identifying the spectral properties of the liability matrix that determine whether a local shock is absorbed or propagates to trigger systemic failure.

At IADU, Zafrir applies dynamical systems and network theory to the systemic risk and financial stability research that forms a central part of the Institute's policy-relevant programme. His contributions include spectral methods for the analysis of financial network fragility under heterogeneous shock distributions, graphon-based mean-field limits for large heterogeneous-agent financial models, and the connection between individual institution dynamics and aggregate systemic behaviour under stress. He supports the development of analytically tractable network models that can underpin macroprudential and regulatory policy analysis.