Pathwise Regularisation of Singular Interacting Particle Systems and their Mean Field Limits

Abstract

We investigate the regularizing effect of certain perturbations by noise in singular interacting particle systems under the mean field scaling. In particular, we show that the addition of a suitably irregular path can regularise these dynamics and we recover the McKean--Vlasov limit under very broad assumptions on the interaction kernel; only requiring it to be controlled in a possibly distributional Besov space. In the particle system we include two sources of randomness, a common noise path $Z$ which regularises the dynamics and a family of idiosyncratic noises, which we only assume to converge in mean field scaling to a representative noise in the McKean--Vlasov equation.

39 pages; main update is technical changes to the proofs of the stability estimate in Lemma 4.13 and presentation of Theorem 5.1

References in corpus (5)

• Distribution dependent SDEs driven by additive continuous noise
• C-infinity regularization of ODEs perturbed by noise
• Regularisation by regular noise
• Function approximation by neural nets in the mean-field regime: Entropic regularization and controlled McKean-Vlasov dynamics
• Entropic propagation of chaos for mean field diffusion with $L^p$ interactions via hierarchy, linear growth and fractional noise