betaFeedback

WorksMathematics

Well-posedness and Quantitative Stability for Degenerate Fokker–Planck Equations with Rough Coefficients

Abstract

We establish quantitative stability and well-posedness for degenerate Fokker–Planck equations with Sobolev-regular and possibly fully degenerate diffusion coefficient. For two classes of rough drifts, stability estimates imply uniqueness, and existence is proved under suitable conditions. First, for anisotropic drifts with partial fractional Sobolev regularity, we prove both existence (as probability densities, essential for the probabilistic interpretation via superposition) and quantitative stability without imposing any growth conditions, covering the full range of fractional exponents and extending previous Lagrangian-flow results. Second, for drifts whose distributional gradients are given by singular integrals (e.g., Biot–Savart velocity fields from L1L^1 vorticity), we obtain stability estimates that allow the diffusion matrix to be time-space dependent and possibly degenerate, removing the prior restrictions of identity diffusion or bounded coefficients. Our results also recover the pure transport regime as a limiting case.

Introduction

We consider the stochastic differential equation (SDE) on Rd\mathbb{R}^d:

dXt=b(t,Xt) dt+σ(t,Xt) dWt,LX0=ν,(1)\mathrm{d}X_t=b(t,X_t)\,\mathrm{d}t+\sigma(t,X_t)\,\mathrm{d}W_t,\qquad\mathcal{L}_{X_0}=\nu, \tag*{(1)}

where ν∈P(Rd)\nu\in\mathcal{P}(\mathbb{R}^d) and LXt\mathcal{L}_{X_t} denotes the law of XtX_t for t≥0t\ge0. Here

b:[0,∞)×Rd→Rd,σ:[0,∞)×Rd→Rd×db:[0,\infty)\times\mathbb{R}^d\to\mathbb{R}^d,\qquad\sigma:[0,\infty)\times\mathbb{R}^d\to\mathbb{R}^{d\times d}

are measurable, and W=(Wt)t≥0W = (W_t)_{t\geq0} is a dd-dimensional standard Brownian motion on a complete filtered probability space (Ω,F,(Ft)t≥0,P)(\Omega,\mathcal{F},(\mathcal{F}_t)_{t\geq0},\mathbb{P}). We write a:=σσ∗a := \sigma\sigma^* for the diffusion matrix. The Fokker–Planck equation (FPE) associated with (1.1) is

∂tμt+div⁡(bμt)=12∑i,j=1d∂ij(aijμt),μ0=ν.(2)\partial_t\mu_t+\operatorname{div}(b\mu_t)=\frac{1}{2}\sum_{i,j=1}^{d}\partial_{ij}(a_{ij}\mu_t),\qquad\mu_0=\nu. \tag*{(2)}

It describes how the law of XtX_t evolves in time, and arises naturally in probability theory, statistical physics, and fluid dynamics; see, for example, [3, 29].

When the diffusion term vanishes, that is, σ=0\sigma=0, the equation reduces to the continuity equation

∂tμt+div⁡(bμt)=0,μ0=ν.(3)\partial_t\mu_t+\operatorname{div}(b\mu_t)=0,\qquad\mu_0=\nu. \tag*{(3)}

This equation is closely related to the ordinary differential equation

dYt=b(t,Yt) dt,Y0=x.(4)\mathrm{d}Y_t=b(t,Y_t)\,\mathrm{d}t,\qquad Y_0=x. \tag*{(4)}

Fix T>0T>0. In the classical Cauchy–Lipschitz setting, say b∈C0,1([0,T]×Rd;Rd)b\in C^{0,1}([0,T]\times\mathbb{R}^d;\mathbb{R}^d), the method of characteristics represents the solution of (1.3) as

μt=ν∘Yt−1,t∈[0,T].\mu_t=\nu\circ Y_t^{-1},\qquad t\in[0,T].

This representation is classical; see, for example, [9]. In many applications, however, the Lipschitz assumption on bb must be weakened. Indeed, the continuity equation is often part of many equations coming from physics. For instance, it is part of the inhomogeneous Navier–Stokes equations, where it expresses conservation of mass. In such models the velocity field is determined by the system itself and is in general not Lipschitz in space. This motivates the study of (1.3) with rough vector fields, to which we now turn.

The theory begins with DiPerna and Lions [10], who proved weak well-posedness of (1.3) in L∞([0,T];Lq(Rd))L^\infty([0,T];L^q(\mathbb{R}^d)) for b∈L1([0,T];W1,p(Rd;Rd))b\in L^1([0,T];W^{1,p}(\mathbb{R}^d;\mathbb{R}^d)) with 1/p+1/q=11/p+1/q=1, under suitable bounds on div⁡b\operatorname{div}b and a mild growth condition. Ambrosio [1] later extended this theory to vector fields of bounded variation. Both works also established a qualitative stability result for (1.3): if a sequence of vector fields bnb_n converges to bb in a suitable sense, then the corresponding solutions μn\mu^n converge to the solution μ\mu. This stability, however, is qualitative rather than quantitative. The reason is that the proofs rely on the renormalization method, which is not suited to deriving quantitative stability estimates.

Crippa and De Lellis [9] obtained the first quantitative stability estimates for the flow of (1.4) with a bounded Sobolev vector field b∈L∞([0,T]×Rd;Rd)∩L1([0,T];W1,p(Rd;Rd))b\in L^\infty([0,T]\times\mathbb{R}^d;\mathbb{R}^d)\cap L^1([0,T];W^{1,p}(\mathbb{R}^d;\mathbb{R}^d)) for some p>1p>1. Their key idea was to control a logarithmic distance between two flows. More precisely, given two flows Y1Y^1 and Y2Y^2 generated by vector fields b1b^1 and b2b^2, they introduced functionals of the form

gδ(t):=[∫Br(log⁡(1+∣Y1(t,x)−Y2(t,x)∣δ))pdx]1/p.g_\delta(t):=\left[\int_{B_r}\left(\log\left(1+\frac{\lvert Y^1(t,x)-Y^2(t,x)\rvert}{\delta}\right)\right)^p\mathrm{d}x\right]^{1/p}.

where p>1p > 1, r,δ>0r, \delta> 0, and BrB_r is the open ball in Rd\mathbb{R}^d centered at the origin with radius rr. With these functionals, they derived stability estimates of the form

∥Y1(t,⋅)−Y2(t,⋅)∥Lp(Br)≤Ct∣log⁡(∥b1−b2∥L1([0,t];Lp(BR)))∣−1,\left\|Y^1(t,\cdot)-Y^2(t,\cdot)\right\|_{L^p(B_r)} \le C_t\left|\log\left(\left\|b^1-b^2\right\|_{L^1([0,t];L^p(B_R))}\right)\right|^{-1},

where R=r+T∥b∥L∞([0,T]×Rd)R=r+T\|b\|_{L^\infty([0,T]\times\mathbb{R}^d)}, and CtC_t is a positive constant depending not only on tt but also on rr, ∥b1∥L∞([0,t]×Rd)\|b^1\|_{L^\infty([0,t]\times\mathbb{R}^d)}, ∥b2∥L∞([0,t]×Rd)\|b^2\|_{L^\infty([0,t]\times\mathbb{R}^d)} and ∥∇xb1∥L1([0,t];Lp(Rd))\|\nabla_x b^1\|_{L^1([0,t];L^p(\mathbb{R}^d))}. Motivated by this result, Seis [32] used ideas from optimal transport to establish quantitative stability estimates for solutions of the continuity equation. Beyond the Sobolev setting, the regularity of bb has been relaxed in several directions [27, 8, 4, 15, 25]; of particular relevance to us, [6] treated vector fields whose gradients are singular integrals (Assumption 4.5). This approach was recently extended by [7] to vector fields with partial fractional Sobolev regularity, though under a stronger restriction on the fractional order than in Assumption 3.1.

We now turn to the full equation (2), in which diffusion is present. For smooth coefficients, the weak well-posedness theory is classical, since linearity and regularity give existence and uniqueness of weak solutions through standard a priori estimates [26]. The irregular-coefficient case has also been studied extensively [3, 20, 23, 22, 30, 24, 16]. The coefficients considered in this paper, however, fall outside the scope of these results. Weak well-posedness for (2) is therefore established here through the following route. A quantitative stability estimate gives uniqueness, and an approximation argument gives existence.

The stability estimate is the main step. We therefore begin by reviewing the existing quantitative stability results for (2). The first named author and Luo [21] proved such estimates for FPEs with Sobolev regularity. A key step in their analysis was Trevisan’s superposition principle (see Theorem 2.5), which let them recast the stability problem for the FPE as one for the associated SDE. They introduced the Kantorovich–Rubinstein distance

D~δ(μ,ν):=inf⁡π∈C(μ,ν)∫Rd×Rdlog⁡(1+∣x−y∣2δ2) π(dx,dy),δ>0, μ,ν∈P(Rd),\widetilde{\mathcal{D}}_\delta(\mu,\nu) := \inf_{\pi\in\mathcal{C}(\mu,\nu)} \int_{\mathbb{R}^d\times\mathbb{R}^d} \log\left(1+\frac{|x-y|^2}{\delta^2}\right)\,\pi(\mathrm{d}x,\mathrm{d}y), \qquad\delta>0,\ \mu,\nu\in\mathcal{P}(\mathbb{R}^d),

where C(μ,ν)\mathcal{C}(\mu,\nu) denotes the set of all couplings of μ\mu and ν\nu. Combining this distance with a synchronous coupling of the two associated SDEs, they obtained quantitative stability estimates for weak solutions of the FPE. Using the Zvonkin transformation, they further extended their analysis to singular drifts satisfying the Ladyzhenskaya–Prodi–Serrin condition. In this setting, the diffusion coefficient is the identity matrix, that is, σ=Id×d\sigma=I_{d\times d}, and the drift coefficient satisfies

b∈Lq([0,T];Lp(Rd;Rd)),with p>d, q>2 such that dp+2q<1.b\in L^q([0,T];L^p(\mathbb{R}^d;\mathbb{R}^d)), \qquad\text{with } p>d,\ q>2\ \text{such that }\frac{d}{p}+\frac{2}{q}<1.

Two other works [23, 28] studied vector fields whose gradients can be written as a singular integral. Such vector fields were first studied at the level of the ODE (4) by [6], and their analysis was later extended to the FPE by [28] and by [23]. The former requires the diffusion matrix to be the identity, a=Id×da=I_{d\times d}, while the latter allows a more general σ=σ(t,x)\sigma=\sigma(t,x) with Sobolev regularity, at the cost of an additional boundedness assumption on bb and σ\sigma.

The aim of the present paper is to establish the well-posedness and extend the quantitative stability theory for degenerate FPEs in two directions.

First, we treat anisotropic drift vector fields with partial fractional Sobolev regularity of order ss. For this class, we prove quantitative stability estimates for weak solutions of degenerate FPEs, extending the theory of [7] for Lagrangian flows of the ODE (4) to the diffusive setting. Compared with that work, our treatment introduces two key improvements: we drop the growth assumption on the drift bb, and we cover the full range s∈(0,1)s \in(0,1), rather than the more restrictive range s∈(1/2,1)s \in(1/2,1) treated there. As a special case, when the diffusion coefficient vanishes (σ=0\sigma= 0), the equation reduces to the continuity equation (3), and our result gives a quantitative stability estimate for it, valid for initial data that are absolutely continuous with respect to the Lebesgue measure.

Second, we study drift vector fields whose distributional derivatives are given by singular integrals. The typical example is the velocity field reconstructed from an L1L^1 vorticity through the Biot–Savart law,

b(x)=(kBS∗w)(x),w∈L1(Rd),b(x) = (k_{\mathrm{BS}} * w)(x), \qquad w \in L^1(\mathbb{R}^d),

where the Biot–Savart kernel kBSk_{\mathrm{BS}} is given by

References

  1. [1]L. Ambrosio, Transport equation and Cauchy problem for BV vector fields, Invent. Math. 158 (2004), 227–260.DOI
  2. [2]L. Ambrosio, N. Gigli, G. Savaré, Gradient flows in metric spaces and in the space of probability measures, 2nd ed., Lectures in Mathematics ETH Zürich, Birkhäuser Verlag, Basel, 2008.
  3. [3]V. I. Bogachev, N. V. Krylov, M. Röckner, S. V. Shaposhnikov, Folker–Planck–Kolmogorov Equations, Vol. 207, American Mathematical Society, Providence, RI, 2022.
  4. [4]A. Bohun, F. Bouchut, G. Crippa, Lagrangian flows for vector fields with anisotropic regularity, Ann. Inst. H. Poincaré C Anal. Non Linéaire. 33 (6) (2016), 1409–1429.arxiv.org/abs/1412.2513
  5. [5]H. Brézis, Functional Analysis, Sobolev Spaces and Partial Differential Equations, Springer, New York, 2011.DOI
  6. [6]F. Bouchut, G. Crippa, Lagrangian flows for vector fields with gradient given by a singular integral, J. Hyperbolic Differ. Equ. 10 (2) (2013), 235–282.
  7. [7]G. Crippa, S. Ligabue, A note on the Lagrangian flow associated to a partially regular vector field, Differ. Equ. Dyn. Syst. 31 (4) (2020), 767–786.arxiv.org/abs/1907.13389
  8. [8]G. Crippa, C. Nobili, C. Seis, and S. Spirito, Eulerian and Lagrangian solutions to the continuity and Euler equations with L¹-vorticity, SIAM J. Math. Anal. 49 (5) (2017), 3973–3998.arxiv.org/abs/1705.06188
  9. [9]G. Crippa, C. De Lellis, Estimates and regularity results for the DiPerna–Lions flow, J. Reine Angew. Math. 616 (2008), 15–46.DOI
  10. [10]R. J. DiPerna, P.-L. Lions, Ordinary differential equations, transport theory and Sobolev spaces, Invent. Math. 98 (1989), no. 3, 511–547.DOI
  11. [11]R. J. DiPerna, P.-L. Lions, On the Fokker–Planck–Boltzmann equation, Comm. Math. Phys. 120 (1988), 1–23.DOI
  12. [12]R. J. DiPerna, P.-L. Lions, On the Cauchy problem for Boltzmann equations: global existence and weak stability, Ann. of Math. 130 (1989), no. 2, 321–366.DOI
  13. [13]L. C. Evans, Measure theory and fine properties of functions, 2nd ed., Chapman and Hall/CRC, New York, 2025.
  14. [14]S. Fang, D. Luo, A. Thalmaier, Stochastic differential equations with coefficients in Sobolev spaces, J. Funct. Anal. 259 (5) (2010), 1129–1168.arxiv.org/abs/1001.3007
  15. [15]F. Flandoli, M. Gubinelli, E. Priola, Well-posedness of the transport equation by stochastic perturbation. Invent. Math. 180 (2010), 1–53.DOI
  16. [16]A. Figalli, Existence and uniqueness of martingale solutions for SDEs with rough or degenerate coefficients, J. Funct. Anal. 254 (2008), 109–153.DOI
  17. [17]L. Grafakos, Classical Fourier analysis. 3rd ed., Graduate Texts in Mathematics, 249. Springer, New York, 2014.
  18. [18]N. Ikeda, S. Watanabe, Stochastic Differential Equations and Diffusion Processes, 2nd ed., North-Holland Mathematical Library, vol. 24, North-Holland, Amsterdam, 1989.
  19. [19]P. E. Jabin, Differential equations with singular fields, J. Math. Pures Appl. (9) 94 (2010), 597–621.arxiv.org/abs/1003.5845
  20. [20]C. Le Bris, P.-L. Lions, Parabolic equations with irregular data and related issues: applications to stochastic differential equations, De Gruyter Series in Applied and Numerical Mathematics, vol 4, De Gruyter, Berlin, 2019.
  21. [21]H. Li, D. Luo, Quantitative stability estimates for Fokker–Planck equations, J. Math. Pures Appl. (9) 122 (2019), 125–163.DOI
  22. [22]D. Luo, Fokker–Planck type equations with Sobolev diffusion coefficients and BV drift coefficients, Acta Math. Sin. Engl. Ser. (2) 29 (2013), 303–314.
  23. [23]D. Luo, Uniqueness of degenerate Fokker–Planck equations with weakly differentiable drift whose gradient is given by a singular integral. Electron. Commun. Probab. 19 (2014), no. 43, 1–14.DOI
  24. [24]C. Le Bris, P.L. Lions, Existence and uniqueness of solutions to Fokker–Planck type equations with irregular coefficients, Commun. Partial Differ. Equ. 33 (2008), 1272–1317.DOI
  25. [25]C. Le Bris, P.L. Lions, Renormalized solutions of some transport equations with partially W^{1,1} velocities and applications, Ann. Mat. Pura Appl. 183 (2004), 97–130.DOI
  26. [26]O. A. Ladyženskaja, V. A. Solonnikov, N. N. Ural’ceva, Linear and quasilinear equations of parabolic type, Translations of Mathematical Monographs, vol. 23, American Mathematical Society, Providence, R.I., 1968.
  27. [27]Q.-H. Nguyen, Quantitative estimates for regular Lagrangian flows with BV vector fields, Comm. Pure Appl. Math. 74 (2021), no. 6, 1129–1192.
  28. [28]V. Navarro-Fernández, A. Schlichting, C. Seis, Optimal stability estimates and a new uniqueness result for advection–diffusion equations, Pure Appl. Anal. 4 (3) (2022), 571–596.
  29. [29]H. Risken and T. Frank, The Fokker–Planck Equation: Methods of Solution and Applications, 2nd ed., Springer, Berlin, 1989.
  30. [30]M. Röckner, X. Zhang, Weak uniqueness of Fokker–Planck equations with degenerate and bounded coefficients, C. R. Math. Acad. Sci. Paris. 348, (2010) 435–438.DOI
  31. [31]D. W. Stroock and S. R. S. Varadhan, Multidimensional Diffusion Processes, Classics in Mathematics, Springer, Berlin, 2006. Reprint of the 1997 edition.
  32. [32]C. Seis, A quantitative theory for the continuity equation, Ann. Inst. Henri Poincaré, Anal. Non Linéaire. 34, (2017) 1837–1850.DOI
  33. [33]D. Trevisan, Well-posedness of multidimensional diffusion processes with weakly differentiable coefficients, Electron. J. Probab. (22) 21, (2016), 41 pp.arxiv.org/abs/1507.01357

Paper details

vlnc:W_84pka9qg
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 2

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

ChatGPT (OpenAI)

View statement
Looking for related papers…
Contents