[Paper Review] Weak Well Posedness for Hypoelliptic Stochastic Differential Equation with Singular Drift: A Sharp Result

Paul-Éric Chaudru de Raynal|arXiv (Cornell University)|Jun 17, 2016

Stochastic processes and financial applications21 references3 citations

TL;DR

This paper establishes weak well-posedness for hypoelliptic stochastic differential equations with singular drift under a Hölder continuity condition of order strictly greater than 1/3, using a martingale problem formulation and PDE smoothing techniques. It proves the threshold is sharp by constructing a counterexample when the Hölder exponent is below 1/3.

ABSTRACT

In this paper, we prove weak uniqueness of hypoelliptic stochastic differential equation with H{ö}lder drift, with H{ö}lder exponent strictly greater than 1/3. We then extend to a weak framework the previous work [CdR12] where strong uniqueness was proved when the H{ö}lder exponent is strictly greater than 2/3. We also show that this result is sharp, by giving a counter example to weak uniqueness when the H{ö}lder exponent is just below 1/3. Our approach is based on martingale problem formulation of Stroock and Varadhan and is based on smoothing properties of the associated PDE.

Motivation & Objective

• To establish weak well-posedness for a class of hypoelliptic SDEs with singular drift, where the drift is Hölder continuous with exponent strictly greater than 1/3.
• To extend previous strong uniqueness results (valid for Hölder exponent > 2/3) to the weak uniqueness framework.
• To demonstrate that the 1/3 threshold for Hölder continuity is sharp, meaning weak uniqueness fails when the exponent is strictly below 1/3.
• To analyze the system under hypoellipticity and uniformly elliptic diffusion conditions, focusing on regularization by stochastic drift.
• To provide a counterexample to weak uniqueness when the Hölder exponent is less than 1/3, confirming the sharpness of the result.

Proposed method

• Formulates the SDE system as a martingale problem using Stroock and Varadhan's framework to analyze weak solutions.
• Applies PDE smoothing properties of the associated Fokker-Planck equation to derive regularity estimates on the solution semigroup.
• Uses a perturbation analysis based on the transition density and conditional expectations to control the dependence on the degenerate component.
• Employs a decomposition of drift differences into multiple terms involving Hölder continuity and spatial regularity, estimating each via heat kernel-type bounds.
• Introduces a parameterized family of operators (e.g., $\tilde{P}_{t,s}^\xi$) to analyze the evolution of test functions and their derivatives.
• Applies interpolation and scaling arguments to derive Hölder-type estimates on the gradient of the solution, leading to a fixed-point argument for uniqueness.

Experimental results

Research questions

• RQ1 Can weak uniqueness be established for hypoelliptic SDEs with drift that is only Hölder continuous with exponent strictly greater than 1/3?
• RQ2 Is the threshold of 1/3 for the Hölder exponent sharp in the sense that weak uniqueness fails when the exponent drops below this value?
• RQ3 How does regularization by noise from a Brownian motion acting only on the degenerate component affect the well-posedness of the system?
• RQ4 Can the martingale problem approach be used to extend strong uniqueness results to the weak setting under weaker regularity assumptions?
• RQ5 What is the role of hypoellipticity and uniform ellipticity of the diffusion matrix in ensuring weak well-posedness under singular drift?

Key findings

• Weak uniqueness holds for the hypoelliptic SDE system (1.1) when the drift is Hölder continuous with exponent strictly greater than 1/3.
• The result extends previous strong uniqueness results (valid for exponent > 2/3) to the weaker weak uniqueness framework.
• A counterexample is constructed to show that weak uniqueness fails when the Hölder exponent is strictly less than 1/3, proving the sharpness of the threshold.
• The proof relies on PDE smoothing techniques and estimates on the transition density and its derivatives via the heat kernel and conditional expectations.
• The method establishes Hölder regularity of the solution's gradient in the degenerate variable, even when the drift is singular, by exploiting the hypoelliptic structure.
• The analysis shows that the critical exponent 1/3 arises from the interplay between the time decay of heat kernel estimates and the Hölder regularity of the drift in the degenerate direction.