2402.14935
Linear-Quadratic Mean Field Games in Hilbert Spaces
Salvatore Federico, Daria Ghilli, Fausto Gozzi
correctmedium confidence
- Category
- Not specified
- Journal tier
- Strong Field
- Processed
- Sep 28, 2025, 12:56 AM
- arXiv Links
- Abstract ↗PDF ↗
Audit review
The paper proves existence of an LQM mild solution by a Yosida-approximation/decoupling scheme under Assumption 3.2 (existence and uniform boundedness of strict solutions to a regularized Riccati equation for ηn), plus uniqueness either for small horizon (a contraction with the explicit constant CT) or, for all T, under dissipativity/nondegeneracy of −QS and −QTST. These are stated and proved in Proposition 3.1, Theorem 3.4, Proposition 4.1, and Theorem 4.2, with key definitions of mild solutions and the formula for s given in (2.11)–(2.14) . By contrast, the model’s Phase 2 claims unconditional (all-horizon) existence for the coupled forward–backward system via a “Volterra resolvent” fixed-point on z. This overlooks that composing the backward (t→T) and forward (0→t) Volterra resolvents produces a two-sided (non-Volterra) operator on z, so the unconditional Neumann-series argument does not apply. The paper’s careful use of Assumption 3.2 (and, alternatively, small T or dissipativity) precisely addresses this gap, and is consistent with their emphasis that the coupled system is “completely new” and must be handled delicately .
Referee report (LaTeX)
\textbf{Recommendation:} minor revisions \textbf{Journal Tier:} strong field \textbf{Justification:} This is a careful and, to my knowledge, first treatment of linear-quadratic mean field games in infinite-dimensional Hilbert spaces with coupling through the mean in the cost. The Riccati part is classical, but the coupled forward–backward system is new and handled via a decoupling/Yosida approximation with clean uniqueness results. A brief clarification on why a direct Volterra-resolvent approach fails would further strengthen the exposition.