[Paper Review] A version of Hörmander's theorem for the fractional Brownian motion

Fabrice Baudoin, Martin Hairer|arXiv (Cornell University)|May 25, 2006

Stochastic processes and financial applications13 references13 citations

TL;DR

This paper establishes a version of Hörmander's hypoellipticity theorem for stochastic differential equations (SDEs) driven by fractional Brownian motion (fBm) with Hurst parameter H > 1/2. Using an extension of Norris’ lemma to fBm, the authors prove that the solution's law admits a smooth density with respect to Lebesgue measure under the standard Hörmander condition on the drift and diffusion vector fields.

ABSTRACT

It is shown that the law of an SDE driven by fractional Brownian motion with Hurst parameter greater than 1/2 has a smooth density with respect to Lebesgue measure, provided that the driving vector fields satisfy Hörmander's condition. The main new ingredient of the proof is an extension of Norris' lemma to this situation.

Motivation & Objective

• To extend Hörmander’s theorem on hypoellipticity to SDEs driven by fractional Brownian motion with Hurst parameter H > 1/2.
• To address the open problem of proving the existence and smoothness of the probability density for the solution of such SDEs.
• To develop a Malliavin calculus framework tailored for fractional Brownian motion to analyze the regularity of the law of the solution.
• To establish a fractional Brownian version of Norris’ lemma, which quantifies the smallness of the integrand when the stochastic integral is small.
• To apply the result to study the small-time behavior of the density on the diagonal.

Proposed method

• Utilizes Malliavin calculus for fractional Brownian motion, based on a Volterra-type representation of fBm on the Wiener space.
• Defines the Hilbert space H as the closure of step functions under a scalar product derived from the covariance structure R(t,s) = ½(|t|²ᴴ + |s|²ᴴ − |t−s|²ᴴ).
• Establishes a fractional integral representation of the H-inner product using the Riemann–Liouville fractional integral operator I^α.
• Proves a new version of Norris’ lemma for fBm, showing that if a stochastic integral is small, then the integrand must be small in a controlled H-norm.
• Applies the extended Norris’ lemma to prove non-degeneracy of the Malliavin matrix under Hörmander’s condition.
• Uses the non-degeneracy of the Malliavin matrix to deduce the existence and smoothness of the density via the Malliavin calculus criterion.

Experimental results

Research questions

• RQ1 Does the solution to an SDE driven by fractional Brownian motion with H > 1/2 admit a smooth density with respect to Lebesgue measure when Hörmander’s condition is satisfied?
• RQ2 Can Norris’ lemma, which is crucial in the classical Malliavin proof, be extended to the fractional Brownian motion setting?
• RQ3 What is the precise relationship between the smallness of a stochastic integral with respect to fBm and the size of its integrand in the Hilbert space H?
• RQ4 How does the small-time behavior of the density on the diagonal relate to the hypoelliptic structure of the SDE?
• RQ5 To what extent can the classical Malliavin calculus framework be adapted to the non-Markovian, non-semimartingale setting of fBm?

Key findings

• The law of the solution to an SDE driven by fBm with Hurst parameter H > 1/2 admits a smooth density with respect to Lebesgue measure if the vector fields satisfy Hörmander’s condition.
• A new version of Norris’ lemma is established for fBm, showing that if the stochastic integral is small, then the integrand must be small in the H-norm, with explicit control depending on H.
• The Malliavin matrix of the solution is non-degenerate under Hörmander’s condition, which implies the existence of a smooth density.
• The result holds for all H > 1/2, extending previous results that required stronger ellipticity or one-dimensional settings.
• The small-time behavior of the density on the diagonal is analyzed and shown to be consistent with hypoelliptic smoothing.
• The proof relies on a novel extension of the stochastic calculus of variations to fBm, using the Volterra representation and fractional integral operators.