[Paper Review] Global existence, regularity and a probabilistic scheme for a class of ultraparabolic Cauchy problems

Christian Fries, Joerg Kampen|arXiv (Cornell University)|Feb 26, 2010

Stochastic processes and financial applications15 references3 citations

TL;DR

This paper establishes global existence and regularity for a class of degenerate ultraparabolic Cauchy problems satisfying a weak Hörmander condition on part of the domain, using a constructive approach based on Malliavin calculus and stochastic analysis. It introduces a weighted Monte Carlo scheme with path-dependent weight corrections to compute value functions and their sensitivities (Greeks) in reduced LIBOR market models, even when the Malliavin covariance matrix is not invertible.

ABSTRACT

In this paper we establish a constructive method in order to show global existence and regularity for a class of degenerate parabolic Cauchy problems which satisfy a weak Hoermander condition on a subset of the domain where the data are measurable and which have regular data on the complementary set of the domain. This result has practical incentives related to the computation of Greeks in reduced LIBOR market models, which are standard computable approximations of the HJM-description of interest rate markets. The method leads to a probabilistic scheme for the computation of the value function and its sensitivities based on Malliavin calculus. From a practical perspective the main contribution of the paper is an Monte-Carlo algorithm which includes weight corrections for paths which move in time into a region where a (weak) Hoermander condition holds.

Motivation & Objective

• To address the challenge of computing Greeks in reduced LIBOR market models where standard Malliavin calculus fails due to non-invertible Malliavin covariance matrices.
• To establish global existence and regularity for a class of ultraparabolic Cauchy problems with measurable data on a subset where a weak Hörmander condition holds.
• To develop a probabilistic numerical scheme that remains stable and accurate even when diffusion degenerates in certain regions of the state space.
• To extend the applicability of Monte Carlo methods to problems with Lipschitz-continuous coefficients and non-smooth initial data by incorporating path-dependent weight corrections.

Proposed method

• Constructs a solution via an AD-scheme (alternating direction scheme) for degenerate parabolic equations under weak Hörmander-type conditions.
• Applies Peano’s method adapted to SDEs to ensure global existence and weak regularity under linear growth and Lipschitz continuity of coefficients.
• Introduces a weighted Monte Carlo algorithm where weights are corrected based on whether paths enter or exit a region H where the weak Hörmander condition holds.
• Uses Malliavin weights $ H^j $ and $ H^{jk} $ to compute correction terms $ \Delta C_{t_i} $ for paths transitioning into the degenerate region H.
• Defines four path regimes ($ W^{00}, W^{ii}, W^{0i}, W^{i0} $) to determine when and how to apply weight corrections and correction increments.
• Computes the value function and its sensitivities as $ u(t,x) = \mathbb{E}[f(X_T^0)W_T] + \sum_{i=1}^N \mathbb{E}[\Delta C_{t_i}] $, with $ W_T $ being the product of local weights.

Experimental results

Research questions

• RQ1 Can global existence and regularity be established for ultraparabolic Cauchy problems with measurable data on a subset where a weak Hörmander condition holds?
• RQ2 How can Monte Carlo methods be adapted to compute sensitivities (Greeks) when the Malliavin covariance matrix is not invertible in $ L^p $ for $ p \geq 1 $?
• RQ3 What is a stable and computable probabilistic scheme for value functions and their derivatives in degenerate diffusion models with Lipschitz coefficients?
• RQ4 How can path-dependent weight corrections be systematically derived and implemented in a Monte Carlo framework for such problems?

Key findings

• The paper proves global existence and weak regularity of solutions to a class of degenerate ultraparabolic Cauchy problems under a weak Hörmander condition on part of the domain and measurable data elsewhere.
• A constructive solution method is developed via an alternating direction scheme, ensuring existence and regularity even when standard ellipticity or strong Hörmander conditions fail.
• The proposed weighted Monte Carlo scheme successfully computes value functions and sensitivities by correcting for paths that enter the degenerate region H, where the diffusion is not uniformly non-degenerate.
• Correction terms $ \Delta C_{t_i} $ are derived using first- and second-order Malliavin weights $ H^j $ and $ H^{jk} $, enabling accurate sensitivity estimation despite non-invertible Malliavin matrices.
• The scheme is robust under Lipschitz-continuous coefficients and allows for exponential growth bounds on payoffs, extending beyond bounded or continuous payoffs.
• The method enables practical computation of Greeks in reduced LIBOR market models, where factor reduction leads to degenerate dynamics, by combining stochastic analysis with pathwise weight adjustments.