Back to search

National Technical University of Athens (NTUA)

Topics on mean-field and McKean-Vlasov backward stochastic differential equations, and the backward propagation of chaos

Abstract

dc:description

Backward propagation of chaos is referring to the phenomenon where the behavior of interactive agents (or particles), described from a system of backward stochastic differential equations (BSDEs), progressively resembles the one as if they where independent, while the number of agents increases to infinity. This thesis aims to study backward propagation of chaos in a setting as general as possible, and also to introduce the notion of stability of backward propagation of chaos. Here stability is understood as the continuity property of backward propagation of chaos with respect to the data sets. The interaction between the different agents is expressed through their empirical measure. In order to identify the asymptotic behaviour of the mean-field systems of BSDEs we are going to use the McKean–Vlasov BSDE. We consider two instances of backward propagation of chaos, when we have path dependence in the generator and when we have the usual instantaneous dependence. So, we begin by establishing the existence and uniqueness for the solutions of the mean-field system and McKean–Vlasov BSDE, under an appropriate framework. Next, we introduce a new way of proving backward propagation of chaos which allows for asymmetric terminal conditions for the mean-field systems, and general square-integrable martingales with independent increments as drivers. Furthermore, we also show that the known convergence rates for the backward propagation of chaos extend to our general setting. Finally, we introduce the notion of stability of backward propagation of chaos with respect to data sets, and prove its validity under a natural setting, for the usual dependence case. First we establish the uniform convergence of the mean-field systems to the McKean–Vlasov BSDEs with respect to the data sets, and then we naturally expand the known stability of BSDEs to include the McKean–Vlasov BSDEs. Their conjunction gives us the stability result. Because our setting incorporates both continuous and discontinuous cases, it allows the development of numerical schemes for the backward propagation of chaos under L^2−type approximations. Additionally to the treatment of backward propagation of chaos, in chapter 1 we present a new way for proving the main section theorems of stochastic calculus. Some merits of our approach are that it requires minimum prerequisites, avoids any direct mention of capacities and works directly with the predictable section while measurable section is an immediate corollary, thus also avoids the double work that is hiddenin the background of the usual proofs. Last but not least, optional (resp. accessible) section follows from an intuitive approximation argument based on the dichotomy of predictable and totally inaccessible times that further clarifies the relationship between these concepts. The chapter closes with an interesting short proof of a disintegration theorem about measures, provided for completeness reasons.

Degree

thesis:*
Grantor dc:publisher
National Technical University of Athens (NTUA)
Year dc:date
2024

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Theodorakopoulos, Stefanos
  • Θεοδωρακόπουλος, Στέφανος

Subjects

dc:subject × 21

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/57092

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Theodorakopoulos, Stefanos; Θεοδωρακόπουλος, Στέφανος. Topics on mean-field and McKean-Vlasov backward stochastic differential equations, and the backward propagation of chaos. National Technical University of Athens (NTUA), 2024. http://hdl.handle.net/10442/hedi/57092