On vanishing diffusivity selection for the advection equation

Giulia Mescolini; Jules Pitcho; Massimo Sorella

REVIEW 2 major objections 4 minor 22 references

On vanishing diffusivity selection for the advection equation

T0 review · 2 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Vanishing diffusion selects a unique solution of the advection equation for rough divergence-free fields, including the example with infinitely many weak solutions.

desk verdict A genuinely new vanishing-diffusivity selection theorem, with a proof that is sound after a short epsilon-localization repair; worth refereeing seriously. read the letter →

arxiv 2411.12910 v2 pith:MFTENVO4 submitted 2024-11-19 math.AP

Giulia Mescolini , Jules Pitcho , Massimo Sorella This is my paper · ORCID

classification math.AP MSC 35A0235D3035Q4934A12

keywords vanishingdiffusivityadvectionequationtransportBVvectorfieldsroughdivergence-freeselectionprincipleanomalousdissipationdualitymethod

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

The paper asks whether adding a small diffusion term to a linear advection equation singles out one weak solution when the velocity field is too rough for uniqueness. It answers yes for divergence-free fields that are of bounded variation away from the initial time and square-integrable in space-time: for every bounded initial datum, the parabolic solutions converge to one well-defined limit as the diffusivity tends to zero, independent of how the limit is taken. This covers the classical example of a divergence-free field with infinitely many distinct bounded advection solutions. The paper also proves that the selected limit does not dissipate energy anomalously and coincides with the limit obtained by first smoothing the velocity field.

What carries the argument

The machine is the backward problem. For each smooth test function $\chi$, the paper solves the backward transport equation $\partial_t\theta_\chi+\operatorname{div}(b\theta_\chi)+\chi=0$ with zero data at time $T$, and shows that the backward parabolic solutions $\theta_\chi^\nu$ converge to $\theta_\chi$ in $C([0,T];w-L^2(\mathbb{T}^d))$ as $\nu\to 0$. The duality identity $\iint\rho^\nu\chi=\int\rho_{\mathrm{in}}\theta_\chi^\nu(0)\,dx$ transfers this backward uniqueness to the forward family $\rho^\nu$, so every weak-star limit $\rho$ must satisfy the same pairing $\iint\rho\chi=\int\rho_{\mathrm{in}}\theta_\chi(0)\,dx$. Since $\chi$ was arbitrary, the forward limit is uniquely characterized. The no-anomalous-dissipation conclusion then follows from the parabolic energy balance $\nu\int|\nabla\rho^\nu|^2\le \tfrac12(\|\rho_{\mathrm{in}}\|_{L^2}^2-\|\rho^\nu(T)\|_{L^2}^2)$, combined with the $L^2$ conservation of the selected limit on intervals away from time zero. Here BV means bounded variation: the spatial distributional derivative of the velocity field is a finite measure.

What would settle it

Exhibit two distinct bounded weak solutions of the backward problem (BW) for some divergence-free field in the stated class and some smooth $\chi$; the duality characterization in Section 4 would then collapse. A direct numerical check on the classical non-uniqueness example is also decisive: if two sequences $\nu_i\to 0$ produce different weak-star limits of the forward diffusive solutions, Theorem 1.4 is false.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1.4. Let $b$ be a divergence-free vector field in $L^1_{\mathrm{loc}}((0,T];BV(\mathbb{T}^d;\mathbb{R}^d))\cap L^2((0,T)\times\mathbb{T}^d;\mathbb{R}^d)$ and let $\rho_{\mathrm{in}}\in L^\infty(\mathbb{T}^d)$. Then the unique bounded solutions $\rho^\nu$ of the advection--diffusion equation with diffusivity $\nu$ have a weak-star limit as $\nu\to 0$, and that limit is independent of the vanishing sequence: there is a unique vanishing diffusivity solution. Moreover, the same unique limit is obtained by solving the undiffused equation along the mollified fields $b*w_\delta$ for any standard mollifier $w$, and $\limsup_{\nu\to 0}\nu\int_0^T\int_{\mathbb{T}^d}|\nabla\rho^\nu|^2\,dx\,dt=0$. The proof characterizes the limit through the duality formula $\iint\rho\chi=\int\rho_{\mathrm{in}}\theta_\chi(0)\,dx$, where $\theta_\chi$ is the unique bounded solution of the backward advection equation with source $\chi$ and zero terminal datum; uniqueness of $\theta_\chi$ forces all forward approximate solutions to collapse onto one object.

Load-bearing premise

The whole identification rests on the backward advection problem having a unique bounded weak solution; the uniqueness step cites a global well-posedness theorem although the field is only locally BV near time zero, a gap that appears repairable but is not written out in the paper.

Editorial extensions

If this is right

For this class of fields, the parabolic regularization limit is a well-defined functional of the initial datum and the velocity field, independent of the sequence $\nu_i\to 0$.
The infinitely many weak solutions of the undiffused equation in the classical non-uniqueness example are not selected by parabolic regularization; the selected solution is singled out by the backward adjoint problem.
Mollifying the velocity field and adding diffusion are equivalent selection mechanisms in this class, so smoothing the field does not change the limiting solution.
The selected solution conserves its $L^2$ norm on every time interval away from the initial singularity and has zero anomalous dissipation, so no energy is lost to unresolved small scales in the limit.

Reading between the lines

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

The same duality route suggests that any forward approximation scheme whose adjoint converges to the unique backward solution $\theta_\chi$ would select the same limit; the paper establishes this for diffusion and for mollification, but the mechanism is general.
Repairing the time-zero regularity gap in the backward uniqueness step by applying the classical theorem on each compact interval $[\varepsilon,T]$ would make the proof robust and likely allow an even stronger initial singularity, since the backward source is supported away from $t=0$.
An immediate testable extension is the continuous-dependence question left open in the paper: if the backward solution map $\chi\mapsto \theta_\chi(0)$ is stable, then the vanishing diffusivity solution should depend continuously on initial data in the weak-$L^2$ topology.
A numerical experiment on the classical non-uniqueness example, computing the diffusive limit for several $\nu$ and comparing with the distinct non-unique weak solutions, would directly exhibit which of the infinitely many solutions parabolic selection chooses.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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

Short version: this is a solid paper and the main theorem is new. For divergence-free b in L^1_loc((0,T];BV) ∩ L^2, the vanishing viscosity limit selects a unique bounded weak solution of the advection equation, and the selected solution is the same one obtained by mollifying b. Depauw's vector field falls in this class, so the theorem answers a real question in the selection program. It also proves no anomalous dissipation.

The architecture is good. They prove backward uniqueness for the non-diffusive problem, get convergence of backward diffusive solutions in C([0,T];w-L^2), and use the duality formula to identify every forward accumulation point. That is a clean, honest strategy, and the paper credits Pitcho and De Rosa-Park appropriately. The parabolic well-posedness for L^2 fields is re-proved in Section 2, which is useful and self-contained.

The soft spots are real but limited. The one-line time-reversal invocation of Ambrosio in Theorem 3.3 Step 2 goes beyond the stated global-BV hypothesis. You need to apply Ambrosio on [ε,T] and vary ε. That works because the terminal datum is zero, but the paper should have written it out. Without that, Lemma 3.8 and the duality identification have a hole. The no-anomalous-dissipation argument also uses the same symbol for the original and restarted parabolic solutions in (4.7)-(4.10), which cost me real effort to untangle. Neither issue is fatal; both are exposition.

The citation pattern is fine. [21] covers mollification selection only, [1] the global BV class, and the counterexamples in [10] and [16] are in different settings. The paper correctly positions itself.

Who should read it: anyone working on non-uniqueness and selection for transport equations, or on anomalous dissipation for passive scalars. It is a genuine step forward, not a marginal remark. I would send it to a serious referee—ideally one who has worked with BV vector fields and duality arguments—and would expect the main theorem to survive with the epsilon-localization made explicit.

Referee Report

2 major / 4 minor

Summary. The paper studies the advection equation ∂tρ + div(bρ) = 0 on the torus for divergence-free vector fields b in L1_loc((0,T];BV(Td;Rd)) ∩ L2((0,T)×Td;Rd). Theorem 1.4 claims that for every bounded initial datum there is a unique vanishing diffusivity solution, that the weak-* limit of solutions along mollified fields b*wδ is this same solution for every standard mollifier, and that the diffusive family satisfies limsup_{ν→0} ν∫|∇ρν|2 = 0, i.e., no anomalous dissipation. The proof proceeds by proving uniqueness of a backward advection problem (Theorem 3.3), establishing duality identities for both the diffusive and the mollified problems (Lemma 3.7 and Lemma 3.9), and then using these identities to characterize any forward vanishing diffusivity limit. The final part of the paper adapts an energy argument to rule out anomalous dissipation. The class of fields includes Depauw's example, for which the undiffused equation admits infinitely many bounded weak solutions.

Significance. If correct, the result is significant: it identifies parabolic regularization as a selection principle in a rough-vector-field regime where the inviscid equation is highly nonunique, and it also yields a quantitative statement on the absence of anomalous dissipation. The backward-duality strategy is natural and elegant, and the paper is largely self-contained: Theorem 2.4, the parabolic well-posedness result, is proved in full rather than only cited. The claims are concrete and falsifiable, and the class of admissible fields is clearly delineated. However, as written the proof contains two load-bearing gaps: the uniqueness step for the backward problem applies the global BV well-posedness theorem in a setting where only local BV integrability is available, and the no-anomalous-dissipation argument uses identical notation for three different diffusive solutions. Both issues are repairable and I do not see a substantive threat to the main conclusion, but they need to be fixed in the text.

major comments (2)

[Theorem 3.3, Step 2 (uniqueness of (BW))] The uniqueness step for the backward problem rests on an unstated epsilon-localization. The proof says 'Switching to the time variable t~=T-t, we can refer to Theorem 2.1 to conclude v=0', but Theorem 2.1 requires the vector field to belong to L1((0,T);BV), whereas the hypothesis is only b∈L1_loc((0,T];BV)∩L2; after time reversal the field -b(T-·) need not be in L1((0,T);BV) because the BV norm may blow up near the reversed initial time. Since Lemma 3.8 and the duality identification in Section 4 would collapse if the backward solution θχ were not unique, this step is load-bearing. It is repairable: for each ε>0, b∈L1([ε,T];BV), so -b(T-·)∈L1((0,T-ε);BV); applying Theorem 2.1 on (0,T-ε) with zero initial datum forces the reversed solution to vanish there, and letting ε↓0 gives uniqueness on (0,T]. Please add this argument.
[Section 4, proof of no anomalous dissipation (after Eq. (4.6))] The no-anomalous-dissipation argument uses the symbol ρν_n for three different objects. The displayed inequality after Eq. (4.6) reads 2ν_n∫_δ^T |∇ρν_n|2 ≤ 2ν_n∫_δ^T |∇ρν_n|2 + 2ν_n∫_δ^T |∇(ρν_n-ρν_n)|2, and the subsequent text refers to 'the unique solution to (ν-PDE) on [δ,1] with initial datum ρ(δ,·)' while keeping the same notation. As printed, the inequality is a tautology and the estimates (4.7)-(4.10) cannot be parsed. The intended proof is reconstructible by writing, for example, ρ̄ν_n for the restarted solution on [δ,T] with initial datum ρ(δ,·), and then applying the energy estimates to ρν_n, ρ̄ν_n, and ρν_n-ρ̄ν_n. This correction is essential for the reader to follow the proof of claim (ii).

minor comments (4)

[Lemmas 3.8 and 3.9] In both lemmas, the phrase 'by Lemma 3.6, [0,T]∋t↦∫θχφdx is continuous' should refer to Theorem 3.3, not Lemma 3.6; Lemma 3.6 concerns the diffusive backward problem, while the continuity of θχ is part of Theorem 3.3.
[Theorem 3.3, Step 1] The sentence 'we can pass into the limit δ↓0 in the weak formulation of (ν−BW)' should refer to the nondiffusive weak formulation (BW); the same slip appears in Lemma 3.9, where the weak formulation of (3.2) is cited instead of that of (δ-BW).
[Lemma 3.6, Step 2] The mollified equation is written as ∂tvδ+νΔvδ+div(bδvδ)=rδ with rδ:=div(bvδ-(bv)*wδ); the first term on the left is inconsistent with this definition of rδ and should read div(bvδ).
[Section 4, first paragraph] The phrase 'Let ρ be a vanishing viscosity solution' should read 'vanishing diffusivity solution', matching the terminology used in Definition 1.2 and Proposition 2.5.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the vanishing-diffusivity uniqueness claim is proved in-line by a duality argument resting on Ambrosio's external theorem, with the only self-citation used as strategy rather than as a load-bearing premise.

full rationale

The paper's central claim, Theorem 1.4, is not obtained by assuming its conclusion. The uniqueness of the vanishing diffusivity solution is derived through the duality formula (4.2), which characterizes any vanishing diffusivity solution by the unique backward solution theta_chi of (BW). The uniqueness of theta_chi is proved in Theorem 3.3, Step 2, by invoking Ambrosio's Theorem 2.1 after time reversal. This is an external, machine-checkable-style mathematical result, not a self-citation or a definitional identity. The one-line time-reversal step does use Theorem 2.1 in a way that goes beyond its stated global BV hypothesis, since b is only assumed to lie in L^1_loc((0,T];BV); however, this is a correctness gap that is repairable by an epsilon-localization argument, and it is not a circular reduction. The paper's own prior work [21] is cited only for the strategy ('This strategy draws from the work [21] of the second author') and for context, and the mollified-field selection part (i) is reproved in-line via Lemma 3.9 and the duality formula rather than imported from the self-citation. No parameters are fitted, no known result is merely renamed, and no 'uniqueness theorem' is imported from the authors' own prior work as an unexamined premise. The no-anomalous-dissipation argument in Section 4 is a direct energy estimate using Theorem 2.1 on compact time intervals [delta,T], which is legitimate under the local BV assumption. Overall, every load-bearing step is either proved in the paper or rests on an external theorem, so there is no significant circularity.

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

The central claim rests entirely on external theorems and its own hypotheses: no free parameters, no fitted constants, no invented physical entities. The essential ingredients are Ambrosio's BV transport theory (external), the parabolic well-posedness reproved from [5], compactness and smoothing arguments inside the paper, the distributional divergence-free condition on b, and the L^infinity initial datum. The only structural subtlety is that the backward uniqueness theorem uses the local-in-(0,T] BV integrability through an epsilon-localization that the authors compress; this is an assumption about how the reference theory is applied, not a new postulate.

assumptions (4)

standard math Ambrosio's uniqueness and well-posedness for the transport equation with divergence-free vector fields in L^1((0,T);BV(T^d)) (Theorem 2.1 of the paper).
Used for uniqueness of the backward problem (Theorem 3.3), for L^2 conservation of the transport solution (Lemma 4.1 and equation (4.8)), and for identifying limits of the restarted diffusive family. The paper cites it as external.
domain assumption Well-posedness of forward and backward linear advection-diffusion problems for divergence-free b in L^2, with unique bounded weak solutions in L^infinity intersected with L^2_t H^1_x (Theorem 2.4 and Lemma 3.6); the proofs reuse the commutator and mollification argument of Bonicatto-Ciampa-Crippa…
The vanishing diffusivity family {rho^nu} and the backward family {theta^nu_chi} must be well-defined for the selection and duality argument to make sense. The paper proves Theorem 2.4 in the text and credits [5]; Remark 1.5 notes the L^2 and boundedness hypotheses are not optimal but are needed here.
standard math Standard compactness and smoothing facts: sequential weak-star compactness in L^infinity, Arzela-Ascoli, Riesz representation, and dominated convergence for the mollified commutator terms.
Used in the existence steps of Proposition 2.5, Theorem 3.3, Lemma 3.8 and Lemma 3.9, including the equicontinuity estimates in C([0,T]; w-L^2).
domain assumption The defining hypotheses of the theorem: b divergence-free in the distributional sense, b in L^1_loc((0,T];BV) intersected with L^2((0,T)xT^d), initial datum in L^infinity, and the vanishing diffusivity solution defined as a weak-star accumulation point of {rho^nu}.
These hypotheses delimit what is proved. They are the setting of the paper, not derived from anything; Remark 1.5 explicitly flags that the L^2 assumption is not the optimal exponent pair.

how reviews work

0 comments

Cite this review

Pith. "Pith review of On vanishing diffusivity selection for the advection equation." pith.science (2026). https://pith.science/paper/MFTENVO4

@misc{pith2026241112910,
  author       = {Pith},
  title        = {Pith review of: On vanishing diffusivity selection for the advection equation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MFTENVO4}},
  note         = {Machine review of arXiv:2411.12910}
}

abstract

We study the advection equation along vector fields singular at the initial time. More precisely, we prove that for divergence-free vector fields in $L^1_{loc}((0, T ]; BV (\mathbb{T}^d;\mathbb{R}^d))\cap L^2((0, T ) \times\mathbb{T}^d;\mathbb{R}^d)$, there exists a unique vanishing diffusivity solution. This class includes the vector field constructed by Depauw, for which there are infinitely many distinct bounded solutions to the advection equation.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

22 extracted references · 18 canonical work pages

[21]

Pitcho, J.: A remark on selection of solutions for the transport equation. J. Evol. Equ. 24, 69 (2024)

work page 2024
[1]

Ambrosio, L.: Transport equation and Cauchy problem for BV vector ﬁelds. Invent. Math. 158(2), 227– 260 (2004)

work page 2004
[10]

Colombo, M., Crippa, G., Sorella, M.: Anomalous dissipation and lack of selection in the Obukhov- Corrsin theory of scalar turbulence. Ann. PDE 9, 48 (2023)

work page 2023
[16]

Non-Uniqueness and Inadmissibility of the Vanishing Viscosity Limit of the Passive Scalar Transport Equation

Huysmans, L., Titi, E. S.: Non-uniqueness and inadmissibility of the vanishing viscosity limit of the passive scalar transport equation. arXiv:2307.00809 (2023)

work page Pith review arXiv 2023
[2]

Armstrong, S., Vicol, V .: Anomalous diffusion by fractal homogenization, arXiv:2305.05048 (2023)

arXiv 2023
[3]

Bianchini, S., Bressan, A.: V anishing viscosity solutions of nonlinear hyperbolic systems. Ann. Math. 161, 223–342 (2005)

work page 2005
[4]

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

work page 2022
[5]

Bonicatto, P ., Ciampa, G., Crippa, G.: Weak and parabolic solutions of advection-diffusion equations with rough velocity ﬁeld. J. Evol. Equ. 24(1), 1 (2024)

work page 2024

Show all 22 references

[6]

Bressan, A.: An ill posed cauchy problem for a hyperbolic system in two space dimensions. Rend. Sem. Mat. Univ. Padova 110, 103–117 (2003)

2003
[7]

Bressan, A., Guerra, G., Shen, W.: V anishing viscosity solutions for conservation laws with regulated ﬂux. J. Differ. Equ. 266, 312–351 (2019)

2019
[8]

arXiv:2310.02934 (2023)

Burczak, J., Székelyhidi, L., Wu, B.: Anomalous dissipation and Euler ﬂows. arXiv:2310.02934 (2023)

2023 arXiv
[9]

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

2020
[11]

De Lellis, C., Giri, V .: Smoothing does not give a selection principle for transport equations with bounded autonomous ﬁelds. Ann. Math. Qué. 46(1), 27–39 (2022)

2022
[12]

arXiv:2403.04668 (2024)

De Rosa, L., Park, J.: No anomalous dissipation in two-dimensional incompressible ﬂuids. arXiv:2403.04668 (2024)

2024 arXiv
[13]

Depauw, N.: Non unicité des solutions bornées pour un champ de vecteurs BV en dehors d’un hyperplan. C.R. Math. 337(4), 249–252 (2003)

2003
[14]

DiPerna, R.J., Lions, P .L.: Ordinary differential equations, transport theory and Sobolev spaces. Invent. Math. 98(3), 511–547 (1989)

1989
[15]

Drivas, T.D., Elgindi, T.M., Iyer, G., Jeong, I.-J.: Anomalous dissipation in passive scalar transport. Arch. Ration. Mech. Anal. 243(3), 1151–1180 (2022)

2022
[17]

Johansson, C. J. P ., Sorella, M.: Anomalous dissipation via spontaneous stochasticity with a two- dimensional autonomous velocity ﬁeld. arXiv:2409.03599 (2024)

2024
[18]

Kružkov, S.N.: First order quasilinear equations in several independent variables. Math. USSR-Sbornik 10, 217 (1970)

1970
[19]

arXiv:2404.19687 (2024)

Pitcho, J.: On the lack of selection for the transport equation over a dense set of vector ﬁelds. arXiv:2404.19687 (2024)

2024 arXiv
[20]

Pitcho, J.: On the stochastic selection of integral curves of a rough vector ﬁeld, arXiv:2407.02364 (2024)

2024 arXiv
[22]

Rowan, K.: On anomalous diffusion in the Kraichnan model and correlated-in-time variants, arXiv:2311.12147 (2024) Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. 123

2024 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [21]

Pitcho, J.: A remark on selection of solutions for the transport equation. J. Evol. Equ. 24, 69 (2024)

work page 2024

[2] [1]

Ambrosio, L.: Transport equation and Cauchy problem for BV vector ﬁelds. Invent. Math. 158(2), 227– 260 (2004)

work page 2004

[3] [10]

Colombo, M., Crippa, G., Sorella, M.: Anomalous dissipation and lack of selection in the Obukhov- Corrsin theory of scalar turbulence. Ann. PDE 9, 48 (2023)

work page 2023

[4] [16]

Non-Uniqueness and Inadmissibility of the Vanishing Viscosity Limit of the Passive Scalar Transport Equation

Huysmans, L., Titi, E. S.: Non-uniqueness and inadmissibility of the vanishing viscosity limit of the passive scalar transport equation. arXiv:2307.00809 (2023)

work page Pith review arXiv 2023

[5] [2]

Armstrong, S., Vicol, V .: Anomalous diffusion by fractal homogenization, arXiv:2305.05048 (2023)

arXiv 2023

[6] [3]

Bianchini, S., Bressan, A.: V anishing viscosity solutions of nonlinear hyperbolic systems. Ann. Math. 161, 223–342 (2005)

work page 2005

[7] [4]

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

work page 2022

[8] [5]

Bonicatto, P ., Ciampa, G., Crippa, G.: Weak and parabolic solutions of advection-diffusion equations with rough velocity ﬁeld. J. Evol. Equ. 24(1), 1 (2024)

work page 2024