Oded Zohar

Oded Zohar is a Research Associate in the Games, Dynamics and Strategic Control Division at the Institute for Advanced Dynamic Uncertainty. He holds a PhD in Computing and Mathematical Sciences from the California Institute of Technology, where his doctoral research developed new duality theory for a class of structured nonconvex optimisation problems arising in zero-sum game formulations. His thesis established conditions under which minimax duality gaps vanish in infinite-dimensional function spaces, derived sharp bounds on the duality gap for classes of bilinear saddle-point problems over conic constraints, and constructed first-order primal-dual algorithms with provable convergence rates under partial strong convexity.

Zohar's research centres on the structural theory of convex and nonconvex optimisation: subdifferential calculus in Banach spaces, conjugate duality and its failure in nonconvex settings, proximal operator theory, and the geometry of constraint sets in conic and semidefinite programmes. He is drawn in particular to optimisation problems whose structure encodes a game-theoretic interaction — where the primal and dual variables correspond to the strategies of opposing players and where strong duality is equivalent to the existence of a Nash equilibrium. This perspective connects his work directly to the mathematical theory of zero-sum differential games and to the saddle-point formulations of Hamilton-Jacobi-Isaacs equations.

At IADU, his research addresses primal-dual algorithms for mean field game systems, the convex-analytic foundations of Nash equilibrium existence in continuous-time games with infinite-dimensional strategy spaces, and the design of scalable first-order methods for large-scale HJI equations arising in strategic policy problems.