The Landau equation and Fisher information

Luis Silvestre; Nestor Guillen

REVIEW 1 major objections 5 minor 4 cited by

The Landau equation and Fisher information

T0 review · 1 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper establishes that Fisher information is non-increasing along the homogeneous Landau flow, covering potentials $\alpha(r)=r^\gamma$ with $|\gamma|\le\sqrt{22}$, including the Coulomb case $\gamma=-3$, and thereby ruling out…

desk verdict A clear, honest expository account of the Fisher-information monotonicity proof for Landau-Coulomb; the computations check out, and the main load-bearing constant is less fragile than it looks because the Coulomb case only needs a weaker bound. read the letter →

arxiv 2507.05167 v1 pith:R7N4ASBO submitted 2025-07-07 math.AP math-phmath.MPmath.STstat.TH

Nestor Guillen , Luis Silvestre This is my paper · ORCID

classification math.APmath-phmath.MPmath.STstat.TH MSC 35Q2082C40

keywords LandauequationCoulombpotentialFisherinformationmonotonefunctionalblow-upliftedheatonthesphereΓ2inequality

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 argues that the Fisher information $I(f)=\int_{\mathbb{R}^3}|\nabla f|^2/f\,dv$ is a Lyapunov functional for the spatially homogeneous Landau equation: it never increases along a solution. The proof works for power-law potentials $\alpha(r)=r^\gamma$ whenever $|\gamma|\le\sqrt{22}$, which includes the physically central Coulomb case $\gamma=-3$. Because a time-uniform bound on this quantity controls higher regularity, the result rules out finite-time blow-up and yields smooth solutions for all time. The argument runs through a 'lifted equation' in six variables that turns the nonlinear collision operator into a family of heat equations on the sphere, paired with an improved functional inequality for even functions on $S^2$.

What carries the argument

The load-bearing object is the lifted equation: given a solution $f$ of the homogeneous Landau equation, let $F(v,w)=f(v)f(w)$ and solve $\partial_t F=Q(F)$, where $Q(F)=(\partial_{v_i}-\partial_{w_i})a_{ij}(v-w)(\partial_{v_j}-\partial_{w_j})F$ is the linear operator whose marginal reproduces the collision operator to first order. In the coordinates $(z,r,\sigma)$ the lift becomes $\bar Q\bar F=\alpha(r)\Delta_\sigma\bar F$, a heat equation on each sphere with diffusion coefficient depending only on the separation $r$. The functional input is the lifting property of the Fisher information: $I(F)\ge 2i(\pi F)$ with equality at $F=f\otimes f$, together with the decomposition of $I(F)$ into parallel, spherical, and radial Fisher components. The decisive estimate is inequality (19) on $S^2$: for even functions $\bar f(\sigma)=\bar f(-\sigma)$, $\int_{S^2}\Gamma_2(\ln\bar f)\bar f\,d\sigma \ge \lambda\int_{S^2}|\nabla_\sigma\ln\bar f|^2\bar f\,d\sigma$ with $\lambda\ge 11/2$, where $\Gamma_2(\ln\bar f)=|\nabla_\sigma^2\ln\bar f|^2+|\nabla_\sigma\ln\bar f|^2$; this constant is what carries the Coulomb case.

What would settle it

Compute, by exact spectral analysis or high-precision numerical diagonalization, the optimal constant in inequality (19) restricted to even smooth positive functions on $S^2$; if it is found to be below $11/2$, and in particular below $9/4$, the monotonicity argument for the Landau-Coulomb case does not close.

Watch

Extended reading notes

Core claim

The central discovery is that the Fisher information decreases in time because the nonlinear Landau operator can be linearized one dimension up. Writing the collision operator as the marginal of a linear operator $Q$ acting on $f\otimes f$, the paper lifts the equation to $\partial_t F=Q(F)$ in $\mathbb{R}^6$, where in coordinates $v=z+r\sigma$, $w=z-r\sigma$ the operator becomes $\alpha(r)\Delta_\sigma$, a heat equation on each sphere. The Fisher information satisfies $I(f\otimes f)=2i(f)$ and $I(F)\ge 2i(\pi F)$, so monotonicity of $I$ along the lifted flow implies monotonicity of $i$ along the Landau flow. Splitting $I(F)$ into parallel, spherical, and radial parts gives the balance $\frac{d}{dt}i(f)\le -\int \alpha\Gamma_2(\ln\bar F)\bar F + \int ((\sqrt{\alpha})')^2|\nabla_\sigma\ln\bar F|^2\bar F r^2$. The two terms are comparable whenever every even smooth positive function on $S^2$ satisfies the $\Gamma_2$ inequality with constant $\lambda\ge 11/2$, which for $\alpha(r)=r^\gamma$ is exactly the condition $|\gamma|\le\sqrt{22}$; hence solutions starting smooth remain smooth for all time.

Load-bearing premise

The load-bearing premise is a borrowed numerical lemma about symmetric functions on the sphere: every smooth positive even function satisfies the $\Gamma_2$ inequality with constant at least $11/2$, and if the optimal constant were smaller than that—specifically below $9/4$ for the Coulomb exponent—the monotonicity chain would break.

Editorial extensions

If this is right

For the homogeneous Landau equation with Coulomb potential, a classical solution with smooth initial data exists for all times: finite-time blow-up is ruled out.
The initial Fisher information serves as a time-uniform control: $i(f(t))\le i(f(0))$ for all $t>0$ under the stated hypotheses.
The same monotonicity covers every power-law potential with $|\gamma|\le\sqrt{22}$, so the method interpolates continuously from the constant-kernel case $\alpha\equiv 1$ down through and beyond the Coulomb exponent.
The result supplies the missing proof for the numerical observation, reported in [BC98], that the Fisher information decreases for general potentials.

Reading between the lines

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

The range $|\gamma|\le\sqrt{22}$ is tied to the constant $11/2$; determining the exact optimal even constant in (19) would immediately give the sharpest power-law range this method can reach.
The lifting construction invites a search for other functionals with the same superadditivity property, such as higher derivatives of $\ln f$, which could yield new monotone quantities for kinetic equations.
If the lifted equation could be adapted to the inhomogeneous Landau equation, where the paper notes the technique does not yet apply, the same monotone quantity might control the full equation with transport.

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

Dear colleague,

You can safely take this one at face value: it is an expository note, and as exposition it works. Guillen and Silvestre walk the reader through the lifting argument that proves Fisher-information monotonicity for the homogeneous Landau equation with Coulomb potential, and the parts of the proof carried out in the note are clean and consistent. I checked the constant arithmetic: the trivial λ=1 bound reproduces the range |γ|≤2, and the stated λ=11/2 gives |γ|≤√22, covering γ=-3. The derivative computations for the three Fisher components, especially the radial one via the KL divergence and completing the square, look correct.

Nothing here is new, and that is fine. The note attributes every theorem to [GS23], [ISV24], and [Ji24], and says so in the abstract. Its value is pedagogical: the lifting construction, the change of variables to (z,r,σ), and the reduction to heat flow on S² are presented in a way that makes the proof digestible for people who do not work in kinetic theory daily.

The weakest point is Lemma 6, the Bakry-Emery inequality on S² with the improved constant for even functions. It is not proved in the note, and it is the one numerical input the whole argument leans on. That is acceptable for a Notices article, but the note slightly overstates what the stronger constant buys you. The Coulomb case needs only λ≥9/4, and the already-cited [GS23] bound λ≥19/4 exceeds that. So the headline 'no blow-up for Coulomb' does not hang on Ji's improved constant; it already follows from the weaker bound. The improved λ≥11/2 extends the power-law range to |γ|≤√22, but that is a secondary refinement. I would ask the authors to add a sentence making this clear.

Minor presentation issues: the Section 5 headings contain garbled text (probably a rendering artifact, but it looks bad), the list of estimates in Section 5 omits a factor 2 that appears correctly in Section 5.3, and there are a few typos. These are copy-editing, not substance.

The manuscript is honest about its expository status, the computations shown are reproducible, and I found no circular step or fitted constant.

Bottom line: this paper is for a mathematical audience that wants to understand a major recent result without reading the full proof. It deserves a serious referee and, after minor revisions, publication in Notices. I would send it out rather than desk-reject.

Referee Report

1 major / 5 minor

Summary. This expository note explains the mechanism behind the recent proof that solutions of the homogeneous Landau equation with Coulomb potential do not blow up. The central idea is to lift the nonlinear Landau flow to a linear diffusion equation in six variables, which in coordinates (z,r,σ) becomes a family of heat equations on the sphere. The Fisher information is decomposed into parallel, spherical, and radial components, and its time derivative is bounded using a Bakry-Émery Γ2 inequality for even functions on S². The note reduces the monotonicity of the Fisher information to a spectral constant λ ≥ 11/2 and obtains the range |γ| ≤ √22 for power-law potentials α(r) = r^γ, thereby covering the Coulomb case γ = -3.

Significance. If the computations are correct, this is a valuable expository account of an important recent result. The paper's strengths are its clear conceptual reduction of a nonlinear nonlocal problem to a linear heat equation, its explicit reduction of the monotonicity question to a spectral constant on S², and its honest attribution of the delicate spectral estimates to [GS23] and [Ji24]. The note does not claim to prove the spectral lemma itself, which is appropriate for an expository article. The final monotonicity criterion and the Coulomb conclusion are structurally robust, provided the constants in the displayed estimates are corrected as described below.

major comments (1)

[Section 5, decomposition before Theorem 5] The identity I(F) = 1/2 (Ipar(F) + Isph(F) + Irad(F)) is inconsistent with the change of variables v = z + rσ, w = z - rσ. The Jacobian of this map is 8r² dσ dr dz, so with the displayed definitions one obtains I(F) = 4(Ipar(F) + Isph(F) + Irad(F)), not 1/2(...). Consequently Theorem 5 is misstated by a common factor 4 in the two terms on the right-hand side: the estimate that follows from the component computations is d/dt i(f) ≤ -4∫ α Γ2(ln Fbar) Fbar dσdrdz + 4∫ ((√α)′)² |∇σ ln Fbar|² Fbar r² dσdrdz. The comparison in Section 6 is unaffected because the factor 4 is common, so the main monotonicity criterion and the Coulomb conclusion survive, but the theorem as stated is not the one proved and must be corrected.

minor comments (5)

[Section 5.3] In the integration-by-parts display following the radial KL computation, the term α′(∇σ∂r ln Fbar)·∇σ Fbar needs a minus sign; with the minus sign the completion of the square gives the stated bound. Also, the local inequality should read 1/2 d/dt ∫ (∂r Fbar)²/Fbar dσ ≤ ∫ ((√α)′)² |∇σ ln Fbar|² Fbar dσ; the displayed formula is missing the factor Fbar, and after integrating in r² dz dr the estimate for Irad carries an additional factor 2.
[Section 6, Lemma 6] The phrasing 'There is a λ ≥ 11/2 such that ...' is imprecise; it should say that the inequality holds with λ = 11/2, or equivalently that the optimal constant is at least 11/2. The subsequent sentence about [Ji24] shows the intended meaning, but the lemma as written is ambiguous.
[Section 6 and Section 7] The note emphasizes that the Coulomb case requires the improved constant λ ≥ 11/2, but in fact the Coulomb case γ = -3 needs only λ ≥ 9/4, which the already-cited [GS23] bound λ ≥ 19/4 supplies. The stronger [Ji24] constant extends the range but is not needed for the headline claim; this should be stated to avoid making the main conclusion appear dependent on an unproved constant.
[Sections 5.1, 5.2, 5.3] There are several nonsensical inserted strings in the text (for example, 'Marrow hoooooo Ta m o r a Maoooo ...' in Section 5.1, 'addLose addLose addLoss' in Section 5.2, and 'added IT wat wat ...' in Section 5.3). These appear to be text-extraction artifacts and must be removed before publication.
[Section 2, Proposition 2] The proposition states an equality, but the proof only establishes the inequality direction 2⟨j′(f), q(f)⟩ ≤ ⟨J′(f⊗f), Q(f⊗f)⟩. Since the subsequent Fisher information argument uses only this inequality, the exposition would be clearer if the proposition stated the inequality as the proved statement and mentioned the equality condition separately.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the main derivation is self-contained up to an imported, independently stated Bakry-Émery constant, and no fitted input or self-referential reduction is present.

full rationale

The paper's derivation chain is not circular. Lemma 4 explicitly identifies the lifted Landau operator as Q = α(r)Δσ in (z, r, σ) coordinates, and Section 5 computes the time derivatives of the three Fisher-information components Ipar, Isph, and Irad by direct calculation, yielding Theorem 5. The only remaining input is inequality (19) with λ ≥ Λ(α), which is exactly what Lemma 6 supplies for even functions on S2 with λ ≥ 11/2. Lemma 6 is not proved in the note, but it is an independent spectral inequality with stated provenance: the weaker bound λ ≥ 19/4 is credited to [GS23] and the improved bound λ ≥ 11/2 to [Ji24], with the upper bound ≈ 5.73892 also stated. Nothing is fitted to force the conclusion: the constant λ is not adjusted to make the Landau-Coulomb case close, and the Coulomb case γ = -3 requires only λ ≥ 9/4, which is below both bounds already cited. The Maxwell molecules case α ≡ 1 is checked against the earlier results of McKean, Toscani, and Villani, giving an external benchmark. The note is explicitly expository and defers the full proof of Lemma 6 to the cited works; citing one's own prior paper for a complete proof is not circularity when that paper supplies independent evidence, and the stronger constant used for |γ| ≤ √22 comes from [Ji24], whose author is not among the note's authors. No self-definitional identity, fitted-input prediction, or renaming of a known result occurs. The deferral of Lemma 6 is a verifiability limitation of this note, not a circular step.

Assumptions & free parameters 0 free parameters · 6 assumptions · 2 invented entities

No free parameters: the constant 11/2 is a proved lower bound on a spectral constant, not a fit to data, and the note states the competing upper bound 5.73892 from [Ji24]. All numerical claims are derived from stated assumptions. The axioms list the standard background (Bochner, Jensen, Bakry-Émery) and the two imported results (Lemma 6 from [GS23]/[Ji24]; classical Landau/Boltzmann identifications). The invented entities are mathematical tools with no empirical content; they are validated by the theorems they prove.

assumptions (6)

standard math Bochner identity: Δ|∇g|² = 2|D²g|² + 2∇g·∇Δg (Eq. 13).
Used in Section 3 to derive the decay formula (12) for the Fisher information under the heat flow, the model computation for the whole paper.
standard math Jensen's inequality applied to the conditional expectation of ∇ ln F, together with the divergence theorem.
Proves Lemma 3, the lifting inequality 2i(f) ≤ I(F), which the note attributes to Carlen [Car91] and which converts the Landau problem into a problem about the lifted equation.
standard math Bakry-Émery Γ2 criterion for log-Sobolev inequalities on the sphere.
Background for inequality (19) in Section 6; the criterion reduces the needed estimate to a spectral constant λ for even functions on S².
ad hoc to paper Lemma 6: for even functions f on S², λ∫|∇ ln f|²f ≤ ∫Γ2(ln f)f holds with λ ≥ 11/2, and the optimal λ is at most ≈ 5.73892.
The load-bearing numerical estimate, imported from [GS23] (weaker, λ ≥ 19/4) and [Ji24] (λ ≥ 11/2). It is not proved in this note; without the even-symmetry improvement beyond the classical λ = 2, the Coulomb case needing λ ≥ 9/4 would not close.
domain assumption The grazing-collision limit of the Boltzmann operator yields the Landau operator, and the forms (4), integral, divergence, and nondivergence coincide.
Section 1.1; standard kinetic theory used to justify studying equation (1) in the Coulomb case γ = -3.
domain assumption Local smooth solutions of the homogeneous Landau equation exist and can be differentiated in time; global existence follows once monotone quantities are controlled.
The monotonicity proof operates pointwise in time on classical solutions; the final step from Fisher information monotonicity to the absence of blow-up is stated in Section 7 but not sketched in this note and is carried out in [GS23].

invented entities (2)

Lifted equation ∂tF = Q(F) on R6 with Q(F) = (∂vi - ∂wi) aij(v-w) (∂vj - ∂wj) F independent evidence
purpose: Linear auxiliary equation whose marginal πF matches the Landau flow to first order at t = 0, turning the nonlinear collision operator into a linear diffusion for analysis (Eq. 10, Section 2.1).
A mathematical device, not a new physical object. Its worth is verifiable: it yields the monotonicity theorem and, via [GS23], global existence, both of which are checkable results external to the device itself.
Lifting functional J(F), with J(f⊗f) = 2j(f) and J(F) ≥ 2j(πF) for symmetric F independent evidence
purpose: Transfers functional inequalities from R^{2d} to R^d; instantiated for entropy and Fisher information (Definition 1, Proposition 2, Section 2.1).
Abstract construct whose validity rests on the same theorems; the note notes that few other functionals have natural liftings, which is itself a research pointer.

how reviews work

0 comments

Cite this review

Pith. "Pith review of The Landau equation and Fisher information." pith.science (2026). https://pith.science/paper/R7N4ASBO

@misc{pith2026250705167,
  author       = {Pith},
  title        = {Pith review of: The Landau equation and Fisher information},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/R7N4ASBO}},
  note         = {Machine review of arXiv:2507.05167}
}

read the original abstract

In this expository note (submitted to Notices of the AMS) we present the ideas used in our recent work ruling out blow up for the Landau equation with Coulomb potential. Blow up is ruled out by the discovery that the Fisher information is not increasing in time along a solution. This monotonicity is established by means of a new ``lifted equation'' which is an auxiliary linear equation in double the number of variables that encodes the nonlinear nonlocal collision operator. For the Landau equation in particular this lifted equation amounts to a family of heat equations over the sphere. Some background on kinetic equations and the Fisher information, and connections to Bakry-Emery theory is also discussed.

Figures

Figures reproduced from arXiv: 2507.05167 by the authors.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Kac's Program for the Landau Equation
math.AP 2025-06 conditional novelty 8.0 of 10

The k-particle velocity marginals of Kac's particle system converge to the factorized law of the Landau solution for all power-law potentials, including Coulomb collisions, in weak, Wasserstein, entropic, and strong L...
The Homogeneous Landau Equation with Regularised Thermal Noise
math.AP 2026-07 conditional novelty 7.0 of 10

Weak solutions exist for a regularised fluctuating homogeneous Landau equation with moderately soft potentials, small conservative noise in Landau-divergence form, and a refined entropy decay when the noise modes are ...
The fuzzy Landau equation: global well-posedness and Fisher information
math.AP 2025-07 conditional novelty 6.0 of 10

The fuzzy Landau equation has unique global smooth solutions for moderately soft potentials, and its spatial Fisher information decreases monotonically over time.
On a fuzzy Landau Equation: Part II. Solvability results
math.AP 2025-07 conditional novelty 5.0 of 10

Global existence of entropy-dissipating H-solutions is established for the fuzzy Landau equation with power-law and very-soft kernels, together with a priori propagation estimates for moments and Lp norms.

Reference graph

Works this paper leans on

4 extracted references · 2 canonical work pages · cited by 4 Pith papers

[1]

Com- put

[BC98] Christophe Buet and Stephane Cordier, Conserva- tive and entropy decaying numerical scheme for the isotropic Fokker–Planck–Landau equation , J. Com- put. Phys. 145 (1998), no. 1, 228–245. [B´E06] Dominique Bakry and Michel ´Emery, Diffusions hypercontractives, S´ eminaire de probabilit´ es xix 1983/84: Proceedings, 2006, pp. 177–206. [Bol72] Ludwig...

work page 1998
[4]

[Tos92] Giuseppe Toscani, New a priori estimates for the spatially homogeneous Boltzmann equation , Contin. Mech. Thermodyn. 4 (1992), no. 2, 81–93. [Vil02] C´ edric Villani,A review of mathematical topics in collisional kinetic theory, Handbook of mathematical fluid dynamics 1 (2002), 71–74. [Vil25] , Fisher information in kinetic theory , arXiv preprint...

arXiv 1992
[1973]

[PP96] AR Plastino and A Plastino, Symmetries of the Fokker-Planck equation and the Fisher-Frieden ar- row of time , Physical Review E 54 (1996), no

work page 1996
[1975]

Series A, containing papers of a mathematical or physical character 222 (1922), no

[Fis22] Ronald A Fisher, On the mathematical foundations of theoretical statistics, Philosophical transactions of the Royal Society of London. Series A, containing papers of a mathematical or physical character 222 (1922), no. 594-604, 309–368. [GS23] Nestor Guillen and Luis Silvestre, The Landau equa- tion does not blow up, Acta Mathematica (to appear) (...

arXiv 1922

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

Com- put

[BC98] Christophe Buet and Stephane Cordier, Conserva- tive and entropy decaying numerical scheme for the isotropic Fokker–Planck–Landau equation , J. Com- put. Phys. 145 (1998), no. 1, 228–245. [B´E06] Dominique Bakry and Michel ´Emery, Diffusions hypercontractives, S´ eminaire de probabilit´ es xix 1983/84: Proceedings, 2006, pp. 177–206. [Bol72] Ludwig...

work page 1998

[2] [4]

[Tos92] Giuseppe Toscani, New a priori estimates for the spatially homogeneous Boltzmann equation , Contin. Mech. Thermodyn. 4 (1992), no. 2, 81–93. [Vil02] C´ edric Villani,A review of mathematical topics in collisional kinetic theory, Handbook of mathematical fluid dynamics 1 (2002), 71–74. [Vil25] , Fisher information in kinetic theory , arXiv preprint...

arXiv 1992

[3] [1973]

[PP96] AR Plastino and A Plastino, Symmetries of the Fokker-Planck equation and the Fisher-Frieden ar- row of time , Physical Review E 54 (1996), no

work page 1996

[4] [1975]

Series A, containing papers of a mathematical or physical character 222 (1922), no

[Fis22] Ronald A Fisher, On the mathematical foundations of theoretical statistics, Philosophical transactions of the Royal Society of London. Series A, containing papers of a mathematical or physical character 222 (1922), no. 594-604, 309–368. [GS23] Nestor Guillen and Luis Silvestre, The Landau equa- tion does not blow up, Acta Mathematica (to appear) (...

arXiv 1922