Indefinite linear quadratic control of mean-field backward stochastic differential equation

Wencan Wang, Huanjun Zhang · 2024

Read Indefinite linear quadratic control of mean-field backward stochastic differential equation by Wencan Wang, Huanjun Zhang (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-20650.txt.

Description

Abstract This paper is concerned with a general linear quadratic (LQ) control problem of mean-field backward stochastic differential equation (BSDE). Here, the weighting matrices in the cost functional are allowed to be indefinite. Necessary and sufficient conditions for optimality are obtained via a mean-field forward-backward stochastic differential equation (FBSDE).

Catalog record

CREATORWencan Wang, Huanjun Zhang
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE87 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-20650

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Indefinite linear quadratic control of mean-field backward stochastic differential equation Wencan Wang, Huanjun Zhang Abstract This paper is concerned with a general linear quadratic (LQ) control problem of mean-field backward stochastic differential equation (BSDE). Here, the weighting matrices in the cost functional are allowed to be indefinite. Necessary and sufficient conditions for optimality are obtained via a mean-field forward-backward stochastic differential equation (FBSDE). By investigating the connections with LQ problems of mean-field forward systems and taking some limiting procedures, we establish the solvabilities of corresponding Riccati equations in the case that cost functional is uniformly convex. Subsequently, an explicit formula of optimal control and optimal cost are derived. Moreover, some sufficient conditions for the uniform convexity of cost functional are also proposed in terms of Riccati equations, which have not been considered in existing literatures for backward systems. Some examples are provided to illustrate our results. 1 Introduction In recent years, there is an increasing interest on mean-field type stochastic control problems due to their wide applications in mathematics, engineering and finance. Different from classical stochastic control problems, a new feature of the problems is that both stochastic system and cost functional embody the state and the control along with their statistical distributions, which provides a simple but effective technique for describing individuals’ mutual interactions. In this case, the controlled system is an SDE of McKean-Vlasov type, which was first introduced by Kac Kac1956 . Since then, there have been considerable results on related topics. Interested readers may refer to BDLP2009 , BLP2009 for the theory of mean-field BSDEs, AD2011 , BLM2016 , WXX2017 , WZZ2013 , WW2022 for various versions of stochastic maximum principles for mean-field models. The past few years have also witnessed rapid development of mean-field LQ control theory, both in finite and infinite horizons Yong2013 , HLY2017 , ELN2013 , NEL2015 . Yong Yong2013 investigated a mean-field LQ problem in finite horizon systemically by using a decoupling technique. A feedback optimal control is characterized by two coupled Riccati equations. Further in HLY2017 , Huang et. al. studied an infinite horizon case, where an optimal control is expressed via two coupled algebraic Riccati equations. Elliott et. al. ELN2013 and Ni et. al. NEL2015 dealt with discrete time mean-field LQ problems for finite horizon and infinite horizon, respectively. We remark that all the aforementioned papers followed the standard assumption, that is the weighting matrices in cost functional are imposed with some positive-definite conditions. The study of indefinite LQ control problems can be traced back to Chen et. al. CLZ1998 , who first pointed out that some stochastic LQ problems with indefinite (in particular, negative) control weighting matrices may still be sensible and well-posed. An equivalent cost functional method was introduced in Yu2013 to study an indefinite LQ control problem, which was extended to mean-field case in WW2021 . Li et. al. LLY2020 studied an indefinite LQ problem of mean-field SDE by introducing relax compensators. The open-loop and closed-loop solvabilities for LQ control problems of mean-field type are discussed in sun2017 and LSY2016 , respectively. It was shown that the standard assumption is not necessary for the solvabilities of mean-field LQ control problems. Ni et. al. NLZ2015 , NZL2014 considered discrete time indefinite LQ optimal control problems of mean-field type with infinite horizon and finite horizon, respectively. Wang and Zhang WZ2021 investigated uniform stabilization and asymptotic optimality for indefinite mean-field LQ social control system with multiplicative noises. We point out that the existing literatures on indefinite LQ control problems focused on forward stochastic systems. In financial investment, a European contingent claim \zeta , which is a random variable, can be regarded as a contract to be guaranteed at maturity T . One frequently encounters the situation that funds may be injected or withdrawn from the replication process of a contingent claim so as to achieve some other goals. This naturally results in stochastic control problems of BSDEs LSX2019 , DHW2018 , FHW2021 , HWW2016 . Li et. al. LSX2019 considered an LQ control problem of mean-field BSDE, where explicit formulas of optimal control and optimal cost are derived. Mean-field games and social optimal problems of backward stochastic systems with LQ formulation are discussed in DHW2018 , FHW2021 , HWW2016 . However, the study on indefinite LQ control problems of BSDEs is quite lacking in literatures. To our best knowledge, there are only a few papers on this type of control problems, including SWX2019 and SWX2022 , which constructed the optimal control in terms of a Riccati equation, an adjoint equation and a BSDE under the uniform convexity of cost functional for homogeneous and nonhomogeneous cases, respectively. Indefinite LQ control problems of mean-field BSDEs are underdeveloped in literatures and many fundamental questions remain unsolved. This paper is concerned with an indefinite LQ control problem of mean-field BSDE. As preliminary results, we derive necessary and sufficient conditions for optimality. The optimal control is characterized via a mean-field FBSDE, together with a stationarity condition. By investigating the connections with forward mean-field LQ problems, we introduce some reductions of Problem (BLQ), which leads to an equivalent control problem with ( 13 ) holds. Combining with this transformation and some limiting procedures, we derive the solvabilities of Riccati equations associated with Problem (BLQ). With these results, we construct an explicit formulas of optimal control and optimal cost under the uniform convexity of cost functional in terms of Riccati equations, an adjoint process and a mean-field BSDE. Finally, by empolying the equivalent cost functional method, we proposed some sufficient conditions for the uniform convexity of cost functional in terms of Riccati equations. Our work distinguishes itself from existing literatures in the following aspects. (i) We develop a general procedure of constructing optimal control of Problem (BLQ) under the uniform convexity of cost functional (Assumption A3 ), which extends the results in SWX2023 , SWX2022 . With the appearance of mean-field term, we need to introduce two coupled Riccati equations ( 16 )-( 17 ), which play important roles in deriving explicit formulas of optimal control and optimal cost. Moreover, it takes more efforts to establish the connections between Problem (BLQ) and Problem (FLQ λ ) as well as simplify Problem (BLQ), due to our mean-field setting. (ii) In this paper, a general indefinite LQ control problem of mean-field BADE is considered, where the cross-product terms of control and state processes are involved in cost functional. Moreover, both the state equation and the cost functional contain the nonhomogeneous terms. As we can see in Section 3, the cross-product terms and nonhomogeneous terms bring lots of difficulties in deriving optimal control and optimal cost of Problem (BLQ). Compared with LSX2019 , the representations of solution Z and optimal control u in Theorem 3.4 deserves more complex structures. (iii) We provide some necessary and sufficient conditions for the uniform convexity of cost functional in terms of Riccati equations ( 16 ) and ( 17 ). Due to the special structure of backward systems, we adopt the equivalent cost functional method to prove the sufficiency in our paper, which is quite different from mean-field LQ control problems of forward stochastic systems. Moreover, as we will see in Example 6.2, the equivalent cost functional method also provides an alternative and

Read the complete text