On the Analysis of a Singular Stochastic Volterra Differential Equation driven by a Wiener Noise

Emmanuel Coffie Olivier Menoukeu-Pamen Frank Proske · 2025

Read On the Analysis of a Singular Stochastic Volterra Differential Equation driven by a Wiener Noise by Emmanuel Coffie Olivier Menoukeu-Pamen Frank Proske (2025) 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-2501-01373.txt.

Introduction

Coffie, Menoukeu-Pamen, and Proske settle two questions about a noisy equation with memory and a rough push. Their stochastic Volterra equation folds the whole past into the present, runs on Wiener noise, and its drift is singular, too sharp for standard theory. The authors build solutions that are strong, tied to the given noise, and prove each starting point yields exactly one. They then show the solution varies with its starting value in a Sobolev-smooth way, a trait that underpins sensitivity results for such systems.

Catalog record

CREATOREmmanuel Coffie Olivier Menoukeu-Pamen Frank Proske
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE32 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-01373

Browse the full library · About the Tower of Babel

Full text

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

On the Analysis of a Singular Stochastic Volterra Differential Equation driven by a Wiener Noise Emmanuel Coffie Olivier Menoukeu-Pamen Frank Proske Abstract In this article, we construct unique strong solutions to a class of stochastic Volterra differential equations driven by a singular drift vector field and a Wiener noise. Further, we examine the Sobolev differentiability of the strong solution with respect to its initial value. keywords : Singular Volterra SDEs, Brownian motion, Malliavin calculus Mathematics Subject Classification (2010): 60H10, 49N60, 91G80. 1 Introduction Stochastic Volterra differential equations (SDVs) were first studied by Berger and Mizel ( [ 5 ] , [ 6 ] ) and have since been successfully applied to modeling the dynamics of various phenomena, including population growth, the spread of epidemics, and (rough) stock price volatility in finance (see, e.g., [ 1 ] , [ 3 ] ). In contrast to classical stochastic differential equations, SVDEs incorporate time‑dependent kernels which introduce long‑range dependence. These equations are particularly effective in representing dynamical systems that exhibit non-Markovian behavior. In this article, we study SVDEs with spatially singular and merely measurable drift coefficients, where classical Lipschitz or Hölder assumptions fail. In particular, we study the following stochastic Volterra differential equation (SVDE) X_{t}^{x}=x+\int_{0}^{t}b(t,s,X_{s}^{x})ds+B_{t}, (1) 0\leq t\leq T , x\in\mathbb{R}^{d} , where b:\left[0,T\right]\times\left[0,T\right]\times\mathbb{R}^{d}\longrightarrow\mathbb{R}^{d} is a Borel-measurable function and (B_{t})_{t\in[0,T]} is a d-dimensional Brownian motion (on a filtered probability space). Our primary objective is to construct unique strong solutions to the stochastic Volterra differential equation (SVDE) ( 1 ), where the driving vector field b is spatially singular, that is, it is not Lipschitz continuous and may have discontinuities with respect to the variable x (and s ). For this analysis, it is assumed that the driving vector field b in ( 1 ) admits a Volterra‑type expansion: b(t,s,x)=\sum_{m\geq 0}(t-s)^{m}g_{m}(s,x), (2) 0\leq s\leq t\leq T , x\in\mathbb{R}^{d} , where g_{m}\in L^{\infty}(\left[0,T\right]\times\mathbb{R}^{d};\mathbb{R}^{d}) , such that for all m\geq 0 \big|g_{m}\big|_{\infty}^{1-\epsilon}\leq C\frac{m^{(\frac{1}{2}-\epsilon)m}}{m!} (3) for some \epsilon\in(0,\frac{1}{2}) and some constant C<\infty . In this paper, our objective is to analyze the SVDE ( 1 ) for the class \mathbb{H} of vector fields b\in L^{\infty}(\left[0,T\right]^{2}\times\mathbb{R}^{d};\mathbb{R}^{d}) of the form ( 2 ). In fact, we prove for vector fields b\in\mathbb{H} that the SVDE ( 1 ) has a unique strong solution which is Malliavin differentiable and ( \mathbb{P}- a.e.) locally Sobolev differentiable with respect to the initial condition (see Theorem 1 ). Our approach for the construction of strong solutions is based on techniques of Malliavin calculus which was developed in [ 16 ] and [ 18 ] in the context of singular stochastic differential equations with an additive Wiener noise. We mention that in the case of a non-singular kernel the authors in [ 1 ] and [ 20 ] obtain a unique strong solution for \frac{1}{2} -Hölder continuous diffusion coefficients on the real line by employing e.g. the theorem of Yamada-Watanabe. On the other hand, the case of singular kernels (as e.g. of the type (t-s)^{-\alpha} ) and spatially Lipschitz continuous vector fields was investigated in [ 10 ] and [ 11 ] . Furthermore, based on PDE techniques the authors in [ 19 ] establish pathwise uniqueness of one-dimensional SVDEs with singular kernels and \xi -Hölder continuous diffusion coefficients and drift coefficients not depending on the solution. See also [ 21 ] for an extension of the latter result to the case of time-inhomogeneous coefficients. As for the concept of weak solutions of singular SVDEs we refer to [ 2 ] , [ 19 ] and the references therein. While the Malliavin framework is inspired by earlier work in [ 16 , 18 ] , the Volterra framework creates new analytic difficulties that are absent in the Markovian SDE case: - the equation is non-Markovian and lacks the flow/semigroup structure that underlies many classical arguments; - the drift involves time-dependent memory via \int_{0}^{t}b(t,s,\cdot)\,ds , which changes the structure of Picard iterations and derivative representations; - new kernel-dependent estimates are required to control iterated integrals and ensure convergence on small time intervals; - the proof of strong well-posedness under merely measurable drifts in this non-Markovian setting requires a careful combination of approximation, weak convergence, and identification of adapted limits. Thus the present paper constitutes a genuine extension rather than a small technical adaptation and therefore: - establishes strong existence and uniqueness for non‑Markovian SVDEs with merely measurable drift. - proves a regularization by noise phenomenon in the Volterra setting, previously unknown. - derives Malliavin differentiability and Sobolev regularity in the initial condition. In other words, we aim to establish strong well-posedness and regularity for a new class of singular stochastic Volterra equations under minimal assumptions. The results demonstrate a robust regularization-by-noise phenomenon in a non-Markovian framework. The Volterra structure fundamentally alters the analysis since classical semimartingale and Markov techniques are no longer applicable. Finally, let us mention that Volterra equations are a canonical way to model non-Markovian dynamics with memory and that, singular kernels appear naturally in rough and long-memory modelling. See, e.g., [ 1 , 3 ] and the references therein. Further, our focus is on a basic, robust regularization-by-noise phenomenon: additive Brownian noise restores well-posedness even when the drift is merely measurable (in the spatial variable). In this case, we choose the kernel class \mathbb{H} to allow a transparent analytic route to strong solutions and regularity with respect to the initial value; it is not meant as a complete coverage of all rough-volatility kernels, but it provides a rigorous first step for singular SVDEs under minimal drift regularity. The remainder of the paper is organised as follows: We prove that SVDE ( 1 ) has a unique strong solution in Section 2 . In Section 3 , we establish the weak convergence of approximating solutions. We then discuss the existence of a unique weak solution in Section 4 and provide an example to support the main result. 2 Strong uniqueness We are coming to the main result of our paper: Theorem 1 . Suppose that b\in \mathbb{H} . Then there exists for a sufficiently small time horizon T a unique strong solution X_{t}^{x},0\leq t\leq T to the SVDE ( 1 ) for all initial values x . Moreover, X_{t}^{x} is Malliavin differentiable and the associated stochastic flow X_{t}^{\cdot} belongs to the space L^{2}(\Omega;W_{loc}^{1,2}(\mathbb{R}^{d})) for all 0\leq t\leq T . We also need the following auxiliary result for the proof of the main result: Lemma 2 . Let b\in\mathbb{H} given by b(t,s,x)=\sum_{\mu\geq 0}(t-s)^{\mu}\cdot b_{\mu}(s,x) for b_{\mu}\in C_{c}^{\infty}((0,T)\times\mathbb{R}^{d};\mathbb{R}^{d}) , \mu\geq 0, such that \sup_{\mu\geq 0}||Db_{\mu}||_{\infty}<\infty, where D is the spatial derivative of b_{\mu} . Let X_{\cdot}^{x} be the unique strong solution associated with the SVDE ( 1 ). Denote by D_{u} the Malliavin derivative with respect to the Brownian motion B_{\cdot} . Then for a sufficiently small T \sup_{0\leq u\leq T}\big|\big|D_{u}X_{t}^{x}\big|\big|_{L^{2}(\Omega)}\leq C_{d,T}(\sup_{\mu\geq 0}||b_{\mu}||_{\infty}) as well as there exists \alpha=\alpha(s)>0 such that \big|\big|D_{r}X_{s}^{x}-D_{v}X_{s}^{x}\big|\big|_{L^{2}(\Omega)}\leq C_{d,T}(\sup_{\mu\geq 0}\left\|b_{\mu}\right\|_{\infty})\left|r-v\right|^{\alpha} for 0\leq v\leq r\leq T , where C_{d,T}:\left[0,\infty\right)\longrightarrow \left[0,\infty\right) is an increasing, continuous function, \left\|\cdot\right\| a matrix norm and \left\|\cdot\right\|_{\infty} the supremum norm. Proof. We know by assumption that b(t,s,x)=\sum_{\mu\geq 0}(t-s)^{\mu}\cdot b_{\mu}(s,x). So we can use dominated convergence and obtain that Db(t,s,x)=\sum_{\mu\geq 0}(t-s)^{\mu}\cdot Db_{\mu}(s,x) pointwise. By applying the Malliavin derivative D to both sides of the SVDE X_{t}^{x}=x+\int_{0}^{t}b(t,s,X_{s}^{x})ds+B_{t}\text{,} we obtain the equation

Read the complete text