[Paper Review] Harnack Inequalities for Degenerate Diffusions

Charles L. Epstein, Camelia A. Pop|arXiv (Cornell University)|Jun 18, 2014

Advanced Mathematical Modeling in Engineering23 references3 citations

TL;DR

This paper establishes a scale-invariant Harnack inequality for nonnegative solutions to the parabolic Kimura equation with degenerate diffusion and singular drift coefficients, using a probabilistic approach based on stochastic representations and Girsanov's transformation. The authors generalize Feynman-Kac formulas to degenerate settings and prove the Harnack inequality via a novel adaptation of Sturm's method for lower-order perturbations of model operators.

ABSTRACT

We study various probabilistic and analytical properties of a class of degenerate diffusion operators arising in Population Genetics, the so-called generalized Kimura diffusion operators. Our main results is a stochastic representation of weak solutions to a degenerate parabolic equation with singular lower-order coefficients, and the proof of the scale-invariant Harnack inequality for nonnegative solutions to the Kimura parabolic equation. The stochastic representation of solutions that we establish is a considerable generalization of the classical results on Feynman-Kac formulas concerning the assumptions on the degeneracy of the diffusion matrix, the boundedness of the drift coefficients, and on the a priori regularity of the weak solutions.

Motivation & Objective

• To establish a scale-invariant Harnack inequality for nonnegative solutions to the Kimura parabolic equation with degenerate diffusion and singular drift coefficients.
• To generalize the Feynman-Kac formula to weak solutions of degenerate parabolic equations with unbounded drift and degenerate diffusion matrices.
• To construct a continuous Markov process associated with the generalized Kimura operator and solve the corresponding martingale problem.
• To prove the existence and regularity of solutions to the Kimura stochastic differential equation with singular drift using time-changed processes and Girsanov's formula.
• To connect semigroup theory with stochastic processes and derive a stochastic representation for weak solutions of the parabolic equation.

Proposed method

• Adapt Sturm's probabilistic method for Harnack inequalities, treating the generalized Kimura operator as a lower-order perturbation of a model operator with known Harnack properties.
• Construct a continuous Markov process via the solution to the martingale problem associated with the generalized Kimura operator on the domain $ S_{n,m} = \mathbb{R}^n_+ \times \mathbb{R}^m $.
• Use Girsanov's formula to transform the SDE with singular drift into a time-changed diffusion process, enabling the application of probabilistic estimates.
• Employ weighted Sobolev and anisotropic H"older spaces to control the regularity of solutions and derive a priori bounds.
• Derive a stochastic representation of weak solutions using expectations over paths of the diffusion process, extending classical Feynman-Kac formulas to degenerate settings.
• Establish a chain of estimates involving exit times and time-changed processes to bound the $ L^1 $-norm of $ u^{1/3} $, leading to the Harnack inequality.

Experimental results

Research questions

• RQ1 Can a scale-invariant Harnack inequality be established for nonnegative solutions to the Kimura parabolic equation with degenerate diffusion and singular drift?
• RQ2 How can the Feynman-Kac formula be extended to weak solutions of degenerate parabolic equations with unbounded and degenerate coefficients?
• RQ3 What is the relationship between the semigroup generated by the generalized Kimura operator and the solution to the associated stochastic differential equation?
• RQ4 How do logarithmic-type singularities in the drift affect the regularity and stochastic representation of solutions?
• RQ5 To what extent can probabilistic techniques like Girsanov's transformation and time-changed processes be used to analyze degenerate parabolic equations with boundary degeneracy?

Key findings

• The paper proves a scale-invariant Harnack inequality for nonnegative solutions to the Kimura parabolic equation $ u_t - \widehat{L}u = 0 $, as stated in Theorem 7.8.
• A stochastic representation of weak solutions is established for the degenerate parabolic equation with singular drift, generalizing classical Feynman-Kac formulas.
• The Harnack inequality is derived via a probabilistic method based on Sturm's approach, using estimates on exit times and time-changed processes.
• The authors prove that $ \mathbb{E}_{\mathbb{P}^z}\left[ \frac{1}{R} \int_0^R u^{1/3}(s - r^2 \wedge \tau_r, Z(r^2 \wedge \tau_r)) \mathbf{1}_{\{s - r^2 \wedge \tau_r \leq s - 2R^2/3\}} \, dr \right] \leq C \cdot u^{1/3}(t,w) $ for $ (t,w) \in Q^R_{cR}(s,z) $, with $ c \in (\sqrt{2}/\sqrt{3}, 1) $, leading to the Harnack inequality.
• The proof relies on a chain of estimates involving $ L^1 $-norms of $ u^{1/3} $, logarithmic transformations, and integration over time-changed paths, ultimately yielding a uniform bound $ \int_c^1 \frac{\ln I(\beta)}{\beta} \, d\beta \leq C $, which implies $ I(c) \leq C $.