A priori error estimates for the $\theta$-method for the flow of nonsmooth velocity fields

Gennaro Ciampa; Gianluca Crippa; Raffaele D'Ambrosio; Stefano Spirito; Tommaso Cortopassi

REVIEW 2 major objections 4 minor 45 references

A priori error estimates for the $\theta$-method for the flow of nonsmooth velocity fields

T0 review · 2 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper proves that the θ-method on a smoothed divergence-free Sobolev velocity field approximates the unique regular Lagrangian flow with a logarithmic error rate on bounded sets, and the same rate holds for Lagrangian solutions of the…

desk verdict Theorem A is a genuine, carefully proved extension; the abstract's transport-equation rate is overstated because Theorem B only yields convergence with a constant that blows up as the data approximation is refined. read the letter →

arxiv 2506.02747 v1 pith:XUXDD2LT submitted 2025-06-03 math.AP cs.NAmath.NA

Gennaro Ciampa , Tommaso Cortopassi , Gianluca Crippa , Raffaele D'Ambrosio , Stefano Spirito This is my paper

classification math.APcs.NAmath.NA MSC 34A4535A0265L07

keywords RegularLagrangianflowtransportequationθ-methodnonsmoothvelocityfieldsvortex-blobmethodSobolevapriorierrorestimateslogarithmicconvergence

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Velocity fields with only Sobolev (non-Lipschitz) regularity arise naturally in fluid models, but then particle trajectories are not defined pointwise; one must use the regular Lagrangian flow. This paper shows that a standard family of time-stepping schemes, the θ-method, still converges to that flow, provided the velocity is first smoothed at a scale $\varepsilon$ coupled to the time step $h$ by $\varepsilon = h^{1/(2\beta)}$. Under explicit conditions on the smoothing — an $L^1$ approximation rate, a controlled gradient blow-up, and a uniformly bounded divergence — the $L^1$ error on any bounded ball is bounded by a constant divided by $|\log h|$. The same logarithmic rate is transferred to Lagrangian solutions of the transport equation, and numerical experiments around a singular rotating field confirm the predicted decay.

What carries the argument

The load-bearing construction is the regularized θ-method (2.7)–(2.8), which replaces the velocity field by a smooth approximation $b_\varepsilon$ and evaluates time-averaged velocities over each interval. The error analysis runs through three mechanisms: the approximating-sequence bounds (2.3)–(2.6), where the gradient bound $\|\nabla b_\varepsilon\|_{L^\infty} \leq \bar C_1/\varepsilon^\beta$ and the divergence bound $\|\operatorname{div} b_\varepsilon\|_{L^\infty} \leq \bar C_2$ force the coupling $\varepsilon = h^{1/(2\beta)}$; the determinant bounds of Lemma 3.2, obtained by expanding $\det(I+hB) = 1 + h\operatorname{tr} A + \sum_{j=2}^d h^j \sigma_j(\lambda_1,\ldots,\lambda_d)$ and using the divergence control to show $e^{-CT} \leq \det \nabla X_i^{\theta,\varepsilon} \leq e^{CT}$, so the numerical flow cannot compress or expand volume out of control; and the logarithmic functional $Q_{\delta,N,r}(t) = \int_{B_r} \log(1 + |\widetilde{X}^{h,\theta,\varepsilon}(t,x) - X(t,x)|/\delta)\,dx$ of Lemma 3.3, whose growth is controlled via maximal-function estimates for difference quotients of $W^{1,p}$ maps. A superlevel-set argument converts that logarithmic control into the $L^1$ rate $|\log h|^{-1}$. Appendix A shows that the vortex-blob discretization satisfies all hypotheses with $\alpha = 1$ and $\beta = 3$.

What would settle it

Take the singular rotating field (4.1) with initial data near the origin and a mollified velocity satisfying (2.3)–(2.6), run the θ-method with step sizes $h_k = 2^k h_0$ and $\varepsilon = h^{1/(2\beta)}$, and compute $E(h) = \int_{B_r} |\widetilde{X}^{h,\theta,\varepsilon}(t,x) - X(t,x)|\,dx$; if $E(h)\,|\log h|$ is unbounded as $h \to 0$, the logarithmic upper bound of Theorem A fails.

Watch

Extended reading notes

Core claim

The central result, Theorem A, states that if $b$ is bounded and divergence-free in $L^1((0,T);W^{1,p}(\mathbb{R}^d))$ with $1<p<\infty$, and $\{b_\varepsilon\}$ is an approximating sequence satisfying (2.3)–(2.6), then the regularized θ-method flow $\widetilde{X}^{h,\theta,\varepsilon}$ defined by (2.7)–(2.8) with $\varepsilon = h^{1/(2\beta)}$ satisfies $\int_{B_r} |\widetilde{X}^{h,\theta,\varepsilon}(t,x) - X(t,x)|\,dx \leq C/|\log h|$ for every $r>0$, where $X$ is the unique regular Lagrangian flow of $b$ and $C$ depends only on $d$, $p$, $r$, $T$, the $L^\infty$ norm of $b$, the $L^1_t L^p_x$ norm of $\nabla b$, and the smoothing constants $\bar C_1$, $\bar C_2$. Theorem B proves that the numerical solution $u_h(t) = u_0(\widetilde{Y}^{h,\theta,\varepsilon}(t,t,\cdot))$, built from the backward θ-method flow, converges in $L^1$ to the unique Lagrangian solution of the transport equation at the same logarithmic rate, up to a term measuring the $L^1$ approximation of the initial datum. The proof converts the quantitative well-posedness estimates of [19] into a convergence rate for the scheme, so the logarithmic rate is inherited from the logarithmic regularity of the exact flow rather than from any smoothness of the numerical integrator.

Load-bearing premise

Everything rests on the existence of a smooth approximating velocity field whose divergence is uniformly bounded in $L^\infty$ and whose gradient grows no faster than a fixed power of the smoothing scale, a hypothesis the authors themselves note is not met by many standard finite element discretizations.

Editorial extensions

If this is right

For every $\theta \in [0,1]$ — explicit, implicit, and midpoint variants alike — the logarithmic rate is the same; the classical linear or quadratic accuracy seen for Lipschitz fields does not survive below Lipschitz regularity.
The numerical scheme provides a selection principle for the transport equation: even when distributional solutions are non-unique, the θ-method flow converges to the unique Lagrangian solution.
Both standard mollification and the vortex-blob space discretization of Appendix A satisfy the hypotheses, so the estimate covers concrete space discretizations beyond spatially smooth problems.
The rate is an explicit a priori bound: the constant $C$ is independent of $h$ and $\varepsilon$ and depends only on norms of $b$ and the smoothing constants, so the estimate can be used before any computation is run.
Because the rate is inherited from the logarithmic regularity of the exact flow, any substantial improvement in the numerical rate would have to come from new analytic estimates on the regularity of the exact flow itself.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The divergence-in-$L^\infty$ hypothesis (2.5) is the practical bottleneck: the paper itself points out that many finite element spaces fail it. A natural extension would be to replace (2.5) by an $L^2$ divergence bound and accept a mild loss in the determinant estimates, potentially trading the $|\log h|^{-1}$ rate for a slightly slower one such as $|\log h|^{-1/2}$.
Because the exact flow of a Sobolev velocity field is only logarithmically regular, the $1/|\log h|$ barrier plausibly applies to any numerical method for this problem, not just the θ-method; a matching lower bound would make the rate optimal among all schemes.
The vortex-blob parameters $\alpha = 1$, $\beta = 3$ force the coupling $\varepsilon = h^{1/6}$, so realistic step sizes require very fine smoothing cores; testing whether a coarser coupling still yields logarithmic convergence would show how sharp the coupling condition really is.
For $p=1$ the paper proves convergence but no explicit rate; a quantitative version of the BV theory (recent work on concatenated exponential rates points in this direction) might close the gap and produce a rate governed by the equi-integrability of $\nabla b$.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Hi — the thing to know: this paper's main flow estimate, Theorem A, holds up. The theta-method, including implicit steps, for divergence-free Sobolev velocity fields gets a clean logarithmic a priori error bound, and the proof is coherent. The weak spot is the abstract. It promises 'the same logarithmic rate' for the transport equation, but Theorem B proves a two-parameter estimate: ||u_h(t)-u_L(t)||_{L^1} ≤ C(||u0^δ - u0||_{L^1} + C_δ/|log h|). For fixed u0 in L^1, the δ term cannot be removed, so the claimed log rate does not follow for general L^1 data. The paper's own statement of (2.10) is honest, but the abstract and Remark 1.2 oversell it. What is actually new: explicit Euler was done in [18]; the Picard scheme in [28] needed p > d. This paper handles the whole theta family, θ in [0,1], under merely p > 1, in the divergence-free Sobolev setting. The main work is the solvability of implicit steps and the Jacobian/determinant control in Lemma 3.2, which are done carefully. The coupling ε = h^{1/(2β)} is consistent. I don't see a fatal gap in the flow proof. The vortex-blob regularization in Appendix A is a useful complement—it gives a concrete discretization satisfying the restrictive L^∞ divergence control that standard finite elements fail. Soft spots, in proportion. The transport-equation overclaim is a genuine mismatch between advertised and proved result; a revision should correct the abstract or strengthen Theorem B to give a uniform rate for a class of data. The assumptions (2.4)-(2.5) are restrictive, but the authors flag that in Remark 2.6, so it's a scope limitation, not an error. The numerics are illustrative—they show slope alignment on selected examples, not rigorous validation; the paper says that too. Self-citation: the Crippa–De Lellis framework is imported, but the numerical proof is from scratch; I don't see a circularity problem. Who this is for: numerical analysts working on transport with rough coefficients, and anyone in the DiPerna–Lions stable. They'll get a solid extension with a clear proof and a useful counterpoint about how to state convergence rates for PDE approximations. Recommendation: send it for peer review. A referee should ask for the abstract to be matched to Theorem B, and could suggest either weakening the claim or extending the result to get a genuine log rate for L^1 data under some additional condition. The core flow result deserves to be in the literature.

Referee Report

2 major / 4 minor

Summary. The paper studies the numerical approximation of regular Lagrangian flows and transport equations for divergence-free velocity fields with Sobolev regularity b ∈ L^1((0,T); W^{1,p}(R^d)). The authors propose a regularized θ-method, coupling the time step h with a mollification parameter ε, and prove in Theorem A (Eq. (2.9)) an a priori L^1 error estimate of order 1/|log h| for the flow on bounded sets. A second result, Theorem B (Eq. (2.10)), concerns convergence of a backward-flow approximation to the Lagrangian solution of the transport equation. The paper also contains numerical experiments suggesting logarithmic decay, and an appendix constructing a vortex-blob type regularization satisfying the required hypotheses.

Significance. Theorem A is a solid extension of the numerical analysis of ODEs to the DiPerna–Lions setting, and it gives a quantitative logarithmic rate that matches the regularity theory of Crippa–De Lellis. The proof is detailed and uses standard tools (maximal function estimates, Jacobian bounds, change of variables), and the flow part appears correct and self-contained. The advertised transport-equation result, however, is not delivered at the claimed level of generality: the stated estimate in Theorem B contains a parameter δ whose approximation error and constant Cδ cannot be removed to obtain a uniform logarithmic rate for arbitrary L^1 data. This overstatement affects the abstract and Introduction, though it is localized and repairable by revising the claims. The numerical experiments are illustrative rather than rigorous benchmark tests, but they do not undermine the flow theorem.

major comments (2)

[Abstract and Section 5, Theorem B (Eq. (2.10))] The abstract claims that the θ-method yields 'the same logarithmic rate of convergence' for Lagrangian solutions of the transport equation. However, Theorem B only proves ∥u_h(t,·)-u_L(t,·)∥_{L^1} ≤ C(∥u_0^δ-u_0∥_{L^1} + C_δ/|log h|), where C_δ → ∞ as δ→0. For a fixed u_0 ∈ L^1, one must first choose δ and then let h→0, so no uniform O(1/|log h|) bound follows from (2.10). The Step 2 density argument in the proof of Theorem B confirms this: the convergence rate is controlled by the L^1 approximation modulus of u_0, which can be arbitrarily slow. Equation (1.5) in the Introduction, which states a bound ≲ 1/|log h| with an unspecified dependence on the modulus of continuity of u_0, is therefore misleading. The statement of Theorem B and the corresponding abstract/introduction claims should be revised either to assert convergence without a rate for general L^1 data or to state an explicit rate in terms of an approximation modulus.
[Section 5, Step 1 (page 16-17)] In the proof of Theorem B for compactly supported Lipschitz u_0, the paper estimates ∫_{R^d} |u_0(Ỹ(t,t,x))-u_0(Y(t,t,x))| dx ≤ Lip(u_0) ∫_{B_R} |Ỹ(t,t,x)-Y(t,t,x)| dx. This inequality is not justified as written: the integrand is nonzero only when Ỹ(t,t,x) or Y(t,t,x) lies in supp u_0 ⊂ B_R, and since Y(t,t,·) is the backward flow, the set {x : Y(t,t,x) ∈ B_R} is generally not B_R but a transported set of the same measure. The proof would need a change of variables and the compressibility bounds of Lemma 3.2 to control the measure of the relevant set. This is a technical gap in the proof of Theorem B that should be repaired, independently of the rate issue raised above.

minor comments (4)

[Section 3, Proposition 3.1 vs Lemma 3.2] Proposition 3.1 defines ε(h) = (2 C_1 h)^{1/β}, while Lemma 3.2, Lemma 3.3, and Theorem A set ε(h) = h^{1/(2β)}. These are different parametrizations; the latter is compatible with Proposition 3.1 for small h because h = ε^{2β} implies h ≤ ε^β/(2C_1), but the inconsistency in the text should be removed.
[Section 3, Eq. (3.8)] The display in (3.8) appears to have a typesetting error: the term written as 'h ε^{-2β}' should presumably be 'h^2 ε^{-2β}'. With h = ε^{2β}, the second-order term is ε^{4β-2β} = ε^{2β} = h, so the conclusion |r_1| ≤ C̄h is correct, but the displayed formula as printed is confusing.
[Section 4, Figure captions] The figure captions write 'step-sizes 2kh0' where the intended meaning is '2^k h_0'; the missing superscript should be corrected.
[Section 1.5, Notation] The notation paragraph states that I is the d×d identity matrix in R^{2d}; this should read R^d.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Theorem A is proved from the scheme's own error-accumulation estimates; the self-citations to [19] and [1] are independent mathematical results, not fitted inputs.

full rationale

The paper's central derivation chain is not circular. Theorem A's estimate (2.9) is obtained by proving Lemma 3.2 (determinant bounds for the numerical flow, using only the approximating-sequence assumptions (2.4)-(2.5)), Lemma 3.3 (the logarithmic integral estimate via the maximal-function Lipschitz estimate Lemma 3.6 and a change of variables), and then optimizing the auxiliary parameters δ and η. No quantity appearing in the final error bound is fitted to the target error; the coupling ε = h^{1/(2β)} is forced by the stated bounds (2.3)-(2.4), and the logarithmic rate emerges from δ(ε)=max{ε^α,h(ε)}. The citation to [19], coauthored by a present author, explains the origin of the logarithmic regularity framework, but the proof of Lemma 3.3 is re-derived in the paper from standard harmonic analysis, and [19] is an independent theorem whose assumptions do not include the theta-method conclusion; it therefore does not make the argument circular. The citation to [1] concerns optimality and is not used to prove (2.9). Theorem B is a direct corollary of the backward-flow version of Theorem A, with a density argument for general L^1 data. There is a real scope concern: the abstract claims 'the same logarithmic rate' for the transport equation, while (2.10) only yields ∥u_h-u_L∥ ≤ C(∥u0^δ-u0∥ + C_δ/|log h|) with C_δ → ∞, so no uniform logarithmic rate for fixed u0 ∈ L^1 follows. That is an overclaim about the strength of the result, not a circular dependence of the result on its assumptions.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claim is a theorem about a numerical scheme. It assumes standard DiPerna-Lions well-posedness, classical harmonic analysis results, and explicit scale bounds on the regularization. No parameters are fitted to data and no new physical or mathematical entities are introduced. The only fragile inputs are the L-infinity gradient and divergence bounds on b_epsilon, which hold for mollification and the vortex-blob example but may fail for other discretizations.

assumptions (4)

standard math DiPerna-Lions existence and uniqueness of the regular Lagrangian flow for bounded divergence-free vector fields in L^1((0,T); W^{1,p}(R^d)), p>1.
Invoked in Definition 2.1 and in the statements of Theorem A and Theorem B; the exact flow X is defined through this theory and the numerical solution is compared against it.
standard math Strong (p,p) boundedness of the Hardy-Littlewood maximal function for 1<p≤infinity and the pointwise estimate |f(x)-f(y)| ≤ C |x-y| (M grad f(x) + M grad f(y)) for W^{1,p} functions.
Used in Lemma 3.3 to control difference quotients of the velocity field; the rate and constants in (3.18)-(3.20) depend on these classical results.
domain assumption The approximating sequence b_epsilon satisfies the scale bounds (2.3)-(2.6), including ||grad b_epsilon||_{L-infinity} ≤ C1 / epsilon^beta and ||div b_epsilon||_{L-infinity} ≤ C2.
This is the hypothesis that makes the determinant bounds (3.1) and the error accumulation (3.12) quantitative; it restricts the admissible spatial discretizations, as Remark 2.6 notes for finite elements.
domain assumption The numerical maps X_i^{theta,epsilon} obtained from the inverse function theorem are smooth enough (C^1 diffeomorphisms) for the change-of-variables estimates (3.2) to hold.
Lemma 3.2 differentiates the maps and uses det grad X_i^{theta,epsilon}, while Proposition 3.1 states only continuity; the standard global inverse function theorem supplies the missing regularity, but the paper does not spell this out.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A priori error estimates for the $\theta$-method for the flow of nonsmooth velocity fields." pith.science (2026). https://pith.science/paper/XUXDD2LT

@misc{pith2026250602747,
  author       = {Pith},
  title        = {Pith review of: A priori error estimates for the $\theta$-method for the flow of nonsmooth velocity fields},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XUXDD2LT}},
  note         = {Machine review of arXiv:2506.02747}
}

abstract

Velocity fields with low regularity (below the Lipschitz threshold) naturally arise in many models from mathematical physics, such as the inhomogeneous incompressible Navier-Stokes equations, and play a fundamental role in the analysis of nonlinear PDEs. The DiPerna-Lions theory ensures existence and uniqueness of the flow associated with a divergence-free velocity field with Sobolev regularity. In this paper, we establish a priori error estimates showing a logarithmic rate of convergence of numerical solutions, constructed via the $\theta$-method, towards the exact (analytic) flow for a velocity field with Sobolev regularity. In addition, we derive analogous a priori error estimates for Lagrangian solutions of the associated transport equation, exhibiting the same logarithmic rate of convergence. Our theoretical results are supported by numerical experiments, which confirm the predicted logarithmic behavior.

Figures

Figures reproduced from arXiv: 2506.02747 by the authors.

**Figure 1.** displays the pattern of the error estimate and highlight its decay in comparison with the reference logarithmic slope. 0 0.5 1 1.5 2 2.5 3 3.5 10-3 0 0.05 0.1 0.15 0.2 error estimate reference slope (a) θ = 0.2, p = 3, α = 0.36. 2 4 6 8 10 12 14 10-4 0 0.04 0.08 0.12 0.16 error estimate reference slope (b) θ = 0, p = 3, α = 0.35 [PITH_FULL_IMAGE:figures/full_fig_p014_1.png] view at source ↗

**Figure 2.** Error estimate computed according to (2.9), arising from the application of the θ-method to Equation (4.3) (on the left) and Equation (4.4) (on the right). The reference solution is computed with step-size h0 and the displayed error decay results from the application of the same method with step-sizes 2 kh0, k = 0, 1, 2, 3. 5. Lagrangian solutions of the transport equation In this section we apply the Lagrangian the… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

45 extracted references · 43 canonical work pages

[18]

Cortopassi : An explicit Euler method for Sobolev vector fields with applications to the continuity equation on non cartesian grids

T. Cortopassi : An explicit Euler method for Sobolev vector fields with applications to the continuity equation on non cartesian grids. J. Math. Pures Appl. 199 , 103722 (2025)

work page 2025
[28]

Galeati : Almost-everywhere uniqueness of Lagrangian trajectories for 3D Navier--Stokes revisited

L. Galeati : Almost-everywhere uniqueness of Lagrangian trajectories for 3D Navier--Stokes revisited. J. Math. Pures Appl. 200 , 103723 (2025)

work page 2025
[1]

Alberti, G

G. Alberti, G. Crippa, A. L. Mazzucato : Exponential self-similar mixing by incompressible flows. J. Amer. Math. Soc. 32 , 445-490 (2019)

work page 2019
[2]

Ambrosio : Transport equation and Cauchy problem for BV vector fields

L. Ambrosio : Transport equation and Cauchy problem for BV vector fields. Invent. Math. 158 , 227-260 (2004)

work page 2004
[3]

Ambrosio, F

L. Ambrosio, F. Bouchut, C. De Lellis : Well-posedness for a class of hyperbolic systems of conservation laws in several space dimensions. Comm. Partial Differ. Equ. 29 no. 9-10, 1635-1651 (2004)

work page 2004
[4]

J. T. Beale : The approximation of weak solutions to the 2-D Euler equations by vortex elements . In: Glimm, J., Majda, A.J. (eds) Multidimensional Hyperbolic Problems and Computations. The IMA Volumes in Mathematics and Its Applications 29 . Springer, New York, NY (1991)

work page 1991
[5]

Ben Belgacem, P.-E

F. Ben Belgacem, P.-E. Jabin : Convergence of numerical approximations to nonlinear continuity equations with rough force fields. Arch. Rational Mech. Anal. 234 (2), 509-547 (2019)

work page 2019
[6]

Bonicatto, G

P. Bonicatto, G. Ciampa, G. Crippa : On the advection-diffusion equation with rough coefficients: weak solutions and vanishing viscosity. J. Math. Pures Appl. 167 , 204-224 (2022)

work page 2022

Show all 45 references

[7]

Bouchut, G

F. Bouchut, G. Crippa : Lagrangian flows for vector fields with gradient given by a singular integral. J. Hyperbolic Diff. Equ. 10 , 235-282 (2013)

2013
[8]

Boyer : Analysis of the upwind finite volume method for general initial- and boundary-value transport problems

F. Boyer : Analysis of the upwind finite volume method for general initial- and boundary-value transport problems. IMA J. Numer. Anal. 32 (4), 1404-1439 (2012)

2012
[9]

Bru\`e, M

E. Bru\`e, M. Colombo, C. De Lellis : Positive solutions of transport equations and classical nonuniqueness of characteristic curves. Arch. Rational Mech. Anal. 240 , 1055-1090 (2021)

2021
[10]

Bru\`e, M

E. Bru\`e, M. Colombo, A. Kumar : Sharp Nonuniqueness in the Transport Equation with Sobolev Velocity Field. https://arxiv.org/abs/2405.01670

arXiv
[11]

Caravenna, G

L. Caravenna, G. Crippa : A directional Lipschitz extension lemma, with applications to uniqueness and Lagrangianity for the continuity equation. Commun. Partial Differ. Equ. 46 (8), 1488-1520 (2021)

2021
[12]

Cheskidov, X

A. Cheskidov, X. Luo : Nonuniqueness of weak solutions for the transport equation at critical space regularity. Annals of PDE 7 , 2 (2021)

2021
[13]

Cheskidov, X

A. Cheskidov, X. Luo : Extreme temporal intermittency in the linear Sobolev transport: almost smooth nonunique solutions. Analysis and PDE 17 (6), 2161-2177 (2024)

2024
[14]

Ciampa, G

G. Ciampa, G. Crippa, S. Spirito : Smooth approximation is not a selection principle for the transport equation with rough vector field. Calc. Var. 59 , 13 (2020)

2020
[15]

Ciampa, G

G. Ciampa, G. Crippa, S. Spirito : Weak solutions obtained by the vortex method for the 2D Euler equations are Lagrangian and conserve the energy . J. Nonlinear Sci. 30 , 2787-2820 (2020)

2020
[16]

Colombo, R

M. Colombo, R. Colombo, A. Kumar : A convex integration scheme for the continuity equation past the Sobolev embedding threshold. https://arxiv.org/abs/2504.03578

arXiv
[17]

Colombo, G

M. Colombo, G. Crippa, M. Sorella : Anomalous dissipation and lack of selection in the Obukhov-Corrsin theory of scalar turbulence. Ann. PDE 9 , 21 (2023)

2023
[19]

Crippa, C

G. Crippa, C. De Lellis : Estimates and regularity results for the DiPerna-Lions flow. J. Reine Angew. Math. 616 , 15-46 (2008)

2008
[20]

D’Ambrosio : Numerical Approximation of Ordinary Differential Problems

R. D’Ambrosio : Numerical Approximation of Ordinary Differential Problems. From Deterministic to Stochastic Numerical Methods. Springer (2023)

2023
[21]

De Lellis, V

C. De Lellis, V. Giri : Smoothing does not give a selection principle for transport equations with bounded autonomous fields. Ann. Math. Qu\'ebec 46 , 27–39 (2022)

2022
[22]

R. J. DiPerna, P.-L. Lions : Ordinary differential equations, transport theory and Sobolev spaces. Invent. Math. 98 , 511-547 (1989)

1989
[23]

R. J. DiPerna, P.-L. Lions : On the Cauchy problem for Boltzmann equations: global existence and weak stability. Ann. of Math. 130 , 321-366 (1989)

1989
[24]

R. J. DiPerna, P.-L. Lions : Global weak solutions of kinetic equations. Rend. Semin. Mat. Univ. Politec. Torino 46 , 259-288 (1990)

1990
[25]

R. J. DiPerna, A. Majda : Concentrations in regularizations for 2-D incompressible flow. Comm. Pure Appl. Math. 40 , 301-345 (1987)

1987
[26]

Feireisl : Dynamics of viscous compressible fluids

E. Feireisl : Dynamics of viscous compressible fluids. Oxford Lecture Series in Mathematics and its Applications, 26. Oxford University Press, Oxford, 2004

2004
[27]

U. S. Fjordholm, K. H. Karlsen, P. H. C. Pang : Convergent finite difference schemes for stochastic transport equations. SIAM J. Numer. Anal. 63 (2025)

2025
[29]

V. Giri, M. Sorella : Non-uniqueness of integral curves for autonomous Hamiltonian vector fields. Differential Integral Equations 35(7/8) , 411-436 (2022)

2022
[30]

Grothendieck : La théorie de Fredholm

A. Grothendieck : La théorie de Fredholm. Bull. Soc. Math. France 84 , 319-384 (1956)

1956
[31]

Huysmans, A

L. Huysmans, A. R. Said : Mixing Estimates for Passive Scalar Transport by BV Vector Fields. https://arxiv.org/abs/2504.03023

arXiv
[32]

Jabin, D

P.-E. Jabin, D. Zhou : Discretizing advection equations with rough velocity fields on non- C artesian grids . Quart. Appl. Math. 82 (2), 229-303 (2024)

2024
[33]

Kumar : Nonuniqueness of trajectories on a set of full measure for Sobolev vector fields

A. Kumar : Nonuniqueness of trajectories on a set of full measure for Sobolev vector fields. Arch. Rational Mech. Anal. 248 , 114 (2024)

2024
[34]

Lions , Mathematical topics in fluid mechanics

P.-L. Lions , Mathematical topics in fluid mechanics. Vol. 1. Incompressible models. Oxford Lecture Series in Mathematics and its Applications, 3. Oxford Science Publications. The Clarendon Press, Oxford University Press, New York, 1996

1996
[35]

Lions , Mathematical topics in fluid mechanics

P.-L. Lions , Mathematical topics in fluid mechanics. Vol. 2. Compressible models. Oxford Lecture Series in Mathematics and its Applications, 10. Oxford Science Publications. The Clarendon Press, Oxford University Press, New York, 1998
[36]

Modena, G

S. Modena, G. Sattig : Convex integration solutions to the transport equation with full dimensional concentration. Ann. Inst. H. Poincaré C Anal. Non Linéaire 37 , no. 5, 1075–1108 (2020)

2020
[37]

Modena, L

S. Modena, L. Sz\`ekelyhidi Jr : Nonuniqueness for the transport equation with Sobolev vector fields. Ann. PDE 4 , 18 (2018)

2018
[38]

Navarro-Fern\'andez, A

V. Navarro-Fern\'andez, A. Schlichting : Error estimates for a finite volume scheme for advection-diffusion equations with rough coefficients. ESAIM Math. Model. Numer. Anal. 57 (4), 2131-2158 (2023)

2023
[39]

Navarro-Fern\'andez, A

V. Navarro-Fern\'andez, A. Schlichting, C. Seis : Optimal stability estimates and a new uniqueness result for advection-diffusion equations. Pure Appl. Anal. 4 (3), 571-596 (2022)

2022
[40]

Pitcho, M

J. Pitcho, M. Sorella : Almost everywhere nonuniqueness of integral curves for divergence-free sobolev vector fields. SIAM J. Math. Anal. 55 (5), 4640-4663 (2023)

2023
[41]

Schlichting, C

A. Schlichting, C. Seis : Convergence rates for upwind schemes with rough coefficients. SIAM J. Numer. Anal. 55 , 812-840 (2017)

2017
[42]

Schlichting, C

A. Schlichting, C. Seis : Analysis of the implicit upwind finite volume scheme with rough coefficients. Numerische Mathematik 139 , 155-186 (2018)

2018
[43]

Seis : A quantitative theory for the continuity equation

C. Seis : A quantitative theory for the continuity equation. Ann. Inst. H. Poincar\'e C Anal. Non Lin\'eaire 34 , 1837--1850 (2017)

2017
[44]

E. M. Stein : Singular integrals and differentiability properties of functions. Princeton Mathematical Series, 1970

1970
[45]

N. J. Walkington : Convergence of the discontinuous G alerkin method for discontinuous solutions . SIAM J. Numer. Anal. 42 (5), 1801-1817 (2005)

2005

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [18]

Cortopassi : An explicit Euler method for Sobolev vector fields with applications to the continuity equation on non cartesian grids

T. Cortopassi : An explicit Euler method for Sobolev vector fields with applications to the continuity equation on non cartesian grids. J. Math. Pures Appl. 199 , 103722 (2025)

work page 2025

[2] [28]

Galeati : Almost-everywhere uniqueness of Lagrangian trajectories for 3D Navier--Stokes revisited

L. Galeati : Almost-everywhere uniqueness of Lagrangian trajectories for 3D Navier--Stokes revisited. J. Math. Pures Appl. 200 , 103723 (2025)

work page 2025

[3] [1]

Alberti, G

G. Alberti, G. Crippa, A. L. Mazzucato : Exponential self-similar mixing by incompressible flows. J. Amer. Math. Soc. 32 , 445-490 (2019)

work page 2019

[4] [2]

Ambrosio : Transport equation and Cauchy problem for BV vector fields

L. Ambrosio : Transport equation and Cauchy problem for BV vector fields. Invent. Math. 158 , 227-260 (2004)

work page 2004

[5] [3]

Ambrosio, F

L. Ambrosio, F. Bouchut, C. De Lellis : Well-posedness for a class of hyperbolic systems of conservation laws in several space dimensions. Comm. Partial Differ. Equ. 29 no. 9-10, 1635-1651 (2004)

work page 2004

[6] [4]

J. T. Beale : The approximation of weak solutions to the 2-D Euler equations by vortex elements . In: Glimm, J., Majda, A.J. (eds) Multidimensional Hyperbolic Problems and Computations. The IMA Volumes in Mathematics and Its Applications 29 . Springer, New York, NY (1991)

work page 1991

[7] [5]

Ben Belgacem, P.-E

F. Ben Belgacem, P.-E. Jabin : Convergence of numerical approximations to nonlinear continuity equations with rough force fields. Arch. Rational Mech. Anal. 234 (2), 509-547 (2019)

work page 2019

[8] [6]

Bonicatto, G

P. Bonicatto, G. Ciampa, G. Crippa : On the advection-diffusion equation with rough coefficients: weak solutions and vanishing viscosity. J. Math. Pures Appl. 167 , 204-224 (2022)

work page 2022