Nonlinear Fisher information, corresponding functional inequalities and applications

Kentaro Fujie; Tatsuya Hosono; Tomasz Cie\'slak

REVIEW 5 major objections 5 minor 21 references

Nonlinear Fisher information, corresponding functional inequalities and applications

T0 review · 5 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read For the critical (p,q)=(2,1) case of the 1D Keller–Segel system, the paper proves that a unique classical positive solution exists globally in time for every initial mass.

desk verdict Genuinely new Keller-Segel result and useful Fisher-information machinery, but the printed Theorem 3.1/Lemma 5.3 define Sigma inconsistently and need correction before the proof is checkable. read the letter →

arxiv 2509.01475 v1 pith:QIGDXALV submitted 2025-09-01 math.AP

Tomasz Cieślak , Kentaro Fujie , Tatsuya Hosono This is my paper

classification math.AP MSC 35K5935B4535K9235Q92

keywords FisherinformationnonlineardiffusionfunctionalinequalityKeller–Segelsystemchemotaxisglobalexistencep-Laplaceequationentropydissipation

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's central claim is that a suitably defined nonlinear Fisher information is monotone along nonlinear diffusion flows, and that this monotonicity is powerful enough to settle an open question in chemotaxis: the one-dimensional fully parabolic Keller–Segel system with critical nonlinear diffusion and sensitivity, D(u)=(1+u)^{-2} and S(u)=u(1+u)^{-1}, has a unique classical positive solution that exists globally in time for every initial mass. The mechanism is an exact identity, proven in Section 3, that writes the time derivative of the nonlinear Fisher information as a negative square. Applying this identity to the Keller–Segel system yields a priori estimates, including a bound on the integral of |∂x u|^2 divided by u(1+u)^{p+1}, that prevent finite-time blow-up. The paper also generalizes a known functional inequality to the nonlinear setting and constructs a new monotone functional for the p-Laplace equation in one dimension. If correct, the result shows that in this critical one-dimensional regime there is no critical-mass phenomenon.

What carries the argument

The central object is the nonlinear Fisher information ∫|∂xΣ(u)|^2 dx (and its higher-dimensional analogue ∫|∇Σ(u)|^2 dx), where Σ(s)=∫_1^s a(τ)√τ dτ is a primitive chosen so that ∂xΣ(u)=a(u)√u ∂x u. The load-bearing identity is (3.3), which expresses the time derivative of this quantity as a negative perfect square; the square is obtained by completing the square in the second-order expression u a(u)|∂x((1/√u)∂xΣ(u))|^2. In the Keller–Segel applications the same completed-square mechanism is applied with Σ(s)=∫_1^s D(τ)/√S(τ) dτ, so that the dissipation controls the combination of diffusion and sensitivity that appears in the system. Theorem 4.1's functional inequalities then convert the di

What would settle it

Run a high-resolution numerical scheme for (5.1) with D(u)=(1+u)^{-2}, S(u)=u(1+u)^{-1} and a smooth initial datum of very large mass; if ||u(·,t)||_{L^∞} or ∫ u^2 becomes unbounded before a finite blow-up time, the global-existence theorem is false. A rigorous counterexample would be a classical positive solution on a maximal interval (0,T), T<∞, with ||u(·,t)||_{L^∞}→∞ as t↑T.

Watch

Extended reading notes

Core claim

The core discovery is a nonlinear Fisher information identity. For positive classical solutions of u_t = ∂x(a(u)∂x u) on (0,1) with Neumann conditions, define Σ(s)=∫_1^s a(τ)√τ dτ. Then (3.3) gives (1/2)d/dt∫|∂xΣ(u)|^2 dx = -∫ u a(u)|∂x((1/√u)∂xΣ(u))|^2 dx, a perfect-square dissipation, so the quantity is non-increasing and the entropy is convex in time. Section 5 adapts this calculation to the Keller–Segel system with Σ(s)=∫_1^s D(τ)/√S(τ) dτ, producing the entropy production identity (5.8), the a priori bound of Proposition 5.5, and, for (p,q)=(2,1), Theorem 5.1: a unique classical positive solution exists globally in time for any initial mass. Section 6 shows I[u]=∫|∂x(u^{1-1/[2(p-1)]})|^

Load-bearing premise

The proof needs the cell density u to stay strictly positive up to the boundary of (0,1) for as long as the solution exists, because the central identities divide by u and S(u) and integrate by parts; the paper does not provide a quantitative lower bound on u or rule out u reaching zero at the boundary during the estimates.

Editorial extensions

If this is right

For (p,q)=(2,1), the 1D fully parabolic Keller–Segel system has a unique classical positive solution for all time and any nonnegative initial data; the bound on log(1+u) rules out finite-time blow-up regardless of the initial mass.
The nonlinear Fisher information ∫|∂xΣ(u)|^2 dx is non-increasing along the quasilinear heat flow in 1D for every positive diffusion coefficient a, and the entropy ∫H(u)dx is convex in time.
The functional inequalities (4.1) and (4.2) control ∫ a(u)^3/u^3 |∇u|^4 dx and ∫|D^2Σ(u)|^2 dx by the nonlinear Fisher dissipation, extending the linear inequality (1.2) to general nonlinear diffusions.
In the case (p,q)=(1,0), the paper's method gives a shorter proof of the Lyapunov functional that previously established global existence, making the earlier identity in [9, Lemma 2.1] unnecessary.
For the 1D p-Laplace equation with p≥2, the functional I[u]=∫|∂x(u^{1-1/[2(p-1)]})|^p dx is a Lyapunov functional, non-increasing in time.

Reading between the lines

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

If the same dissipation identity can be written for the remaining critical pairs p-q=1 with q outside (1/2,1], the no-critical-mass conclusion would plausibly extend beyond (p,q)=(2,1); the paper's Remark 5.1 leaves those cases open, and the structure of (5.8) suggests the obstruction is the sign of S'' rather than the method itself.
The p-Laplace Lyapunov functional, which the paper establishes formally for classical solutions, could be expected to yield long-time decay rates for L^p-type norms in 1D; the paper does not pursue asymptotics.
The perfect-square form of the dissipation suggests the nonlinear Fisher information may satisfy a differential inequality that quantifies relaxation to equilibrium, analogous to logarithmic Sobolev or Nash inequalities; testing such an inequality would be a natural next step outside the paper.
A numerical check of the a priori bound of Proposition 5.5 for p-q=1, q∈(1/2,1], across a range of initial masses and profiles, would give independent evidence for the global-existence claim and could identify which q-values are genuinely delicate.

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

Colleague,

Read arXiv:2509.01475. The main thing to know: the paper has a genuinely new result—global existence for the 1D fully parabolic Keller–Segel system in the critical (p,q)=(2,1) case—but the central identity as printed is wrong in a way that infects Theorem 3.1, Remark 3.1, and Lemma 5.3. You will need to correct a systematic definition of Sigma before the proof can be checked.

What is actually new and good: Theorem 5.1 resolves an open case noted in Remark 5.1. The alternative Fisher-information identities in Section 3 are a real extension of the Savaré–Toscani approach, and Section 4's functional inequalities are new and likely useful. Section 6's Lyapunov functional for the p-Laplace equation is new, though the authors explicitly say the estimates there are formal. The derivation of the known [9] functional in Section 5.1 is shorter and more direct than the original, which shows the method has power. The paper is self-contained; there are no fitted parameters, and the self-citations are for context, not load-bearing.

The soft spots are real but localized. The stress-test note is correct: Theorem 3.1 defines Sigma(s)=∫ a(τ)√τ dτ, but the proof computes |∂xΣ|² as a²u|u_x|² and then equates it to a²/u|u_x|². The displayed Sigma'' in (3.4) corresponds to the other primitive, Sigma=∫ a/√τ. Remark 3.1 uses the same wrong definition. Lemma 5.3 repeats the error with D/τ instead of D/√τ. Importantly, the main Keller–Segel proof in Section 5.2 uses the correct Sigma = ∫ D/√S, so the new theorem is probably sound—but the foundational identity in Section 3 is not, and the paper needs a correction pass. There is also a minor gap: the estimates divide by u and S(u) without quantifying a lower bound on u. This is likely fixable by standard continuation and the final log(1+u) bound, but it should be stated.

Who it is for: PDE/chemotaxis people using entropy and Fisher-information methods. It deserves a serious referee, but with required revisions that fix the Sigma definitions and re-derive the affected identities. I would want to see the corrected version before relying on Theorem 5.1.

Referee Report

5 major / 5 minor

Summary. The paper introduces a nonlinear Fisher information for quasilinear heat equations, proves related functional inequalities, and applies them to one-dimensional fully parabolic Keller–Segel systems. The headline result is Theorem 5.1: for (p,q)=(2,1), i.e. D(u)=(1+u)^{-2} and S(u)=u(1+u)^{-1}, the 1D critical Keller–Segel system has a unique global classical positive solution for arbitrary initial mass. The paper also gives a new Lyapunov functional for the 1D p-Laplace equation. The methods are based on dissipation identities for nonlinear Fisher information and on Cieślak–Fuest–Hajduk–Sierżęga-type functional inequalities.

Significance. If the main result is correct, it resolves an open question about absence of critical-mass blow-up in a critical quasilinear 1D Keller–Segel system, and the proposed nonlinear Fisher-information framework could be a useful tool for other PDE systems. The paper also provides self-contained derivations of known functionals and a new monotone quantity for the p-Laplace equation. These are potentially valuable contributions. However, the manuscript as written contains multiple inconsistent definitions of the transformed variable Σ, and the central identities are not checkable until these are corrected. The intended correction is recognizable and appears local, but it affects the proofs of the main functional inequalities and of the Keller–Segel estimates.

major comments (5)

[Theorem 3.1, Eqs. (3.1)–(3.4)] The definition Σ(s)=∫_1^s a(τ)√τ dτ is incompatible with identity (3.2). With this definition |∂xΣ|^2 = a(u)^2 u |∂xu|^2, whereas the computed entropy production is ∫ a(u)^2/u |∂xu|^2. Thus (3.2) is false as printed. The proof itself uses Σ′(s)=a(s)/√s (see the Σ″ formula in (3.4), Remark 3.1, and Section 5.2), not Σ′=a√s. This is not a notational nuisance: the correct primitive is Σ(s)=∫_1^s a(τ)/√τ dτ, and all subsequent identities that rely on (3.1)–(3.4) need to be re-derived with that definition.
[Lemma 5.3, Eqs. (5.4)–(5.7)] The printed definition Σ(u)=∫_1^u D(τ)/τ dτ makes the claimed equivalence between (5.4) and (5.5) false: with that Σ, |∂xΣ|^2 = D(u)^2/u^2 |∂xu|^2, whereas the left-hand side of (5.4) is (1/2)d/dt∫ D(u)^2/u |∂xu|^2. The needed primitive is Σ(u)=∫_1^u D(τ)/√τ dτ. The displayed chain in (5.7) also mixes the two normalizations. Since this lemma is presented as the basis of the short derivation in §5.1 and the same pattern is used in §5.2, the authors should correct the definition and the intermediate identities.
[Theorem 4.1, Eqs. (4.1)–(4.2)] The inequality is stated for Σ defined as in (3.1), i.e. with Σ′(s)=a(s)√s, but the proof uses the relation 1/√u ∇Σ(u)=∇Λ(u), which holds only when Σ′(s)=a(s)/√s. Consequently the statement of Theorem 4.1 does not match the proof. The intended version appears to be with the corrected Σ, but as written the theorem is not established.
[Proposition 5.6, Eqs. (5.17)–(5.18)] In the completed-square step, the coefficient of |∂xΣ|^4 in (5.17) is (2SS″−(S′)²)/(4DS), which forces the cross-term denominator to contain √S, not S. The written identity √D∂xxΣ − S′/(2√D S)|∂xΣ|² = √(DS)∂x(1/√S ∂xΣ) is false for general S; replacing the second S by √S makes the identity correct. This step is load-bearing in the proof of the entropy-production identity (5.8) used for Theorem 5.1.
[Section 5, proof of Theorem 5.1] The proof works with positive classical solutions on a maximal interval and divides by u and S(u). No quantitative lower bound away from zero is established, so the continuation argument is incomplete as written. This can likely be repaired by standard positivity/continuation arguments after the L∞ bound, but the step should be stated explicitly.

minor comments (5)

[Corollary 2.3] The sentence 'Suppose assumptions as in Theorem 3.1' should refer to Theorem 2.1, since Theorem 3.1 is introduced later and concerns the 1D case.
[Eq. (5.8)] The formula contains '∂xux' in the second integral; it should be '∂x u'.
[Section 2 heading] The heading contains a typo: ‘Sa v aré’ should be ‘Savaré’.
[References] Reference [14] is incomplete as formatted; the publisher information should be completed.
[Throughout] Several displayed identities contain inconsistent normalization of √u; the authors should carefully check all occurrences of Σ′ and ∂xΣ after the definition is corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the main estimates and identities are derived in-line from the PDE, with no fitted parameters and no load-bearing self-citation.

full rationale

The paper's central derivation chain is self-contained. Theorem 3.1 proves the nonlinear Fisher information identities directly from (1.1) by integration by parts; Theorem 4.1 proves the functional inequalities from pointwise algebraic identities; Proposition 5.6 is re-proved in the text (modulo a small algebraic step referred to [16, Lemma 2.2], which is a published external lemma not dependent on the present data); and Lemmas 5.10–5.13 provide in-text a priori estimates leading to Theorem 5.1. No free parameters are fitted, and no 'prediction' is statistically forced by construction. The self-citations to [9], [11], and [16] are used as context or for standard algebraic completions, but the load-bearing identities are re-derived here. There is, however, a notable mathematical inconsistency in the printed definition of Σ in Theorem 3.1: the statement defines Σ(s)=∫_1^s a(τ)√τ dτ, while the proof (3.4) and the entropy identity (3.2) require Σ(s)=∫_1^s a(τ)/√τ dτ. This is a correctness/typographical issue, not a circularity, because it does not make any conclusion equivalent to an input; it makes the printed identity false as stated. The same type of mismatch appears in Lemma 5.3, where Σ is defined with D(τ)/τ but the surrounding identities require D(τ)/√τ. These should be corrected, but they do not constitute circular reasoning.

Assumptions & free parameters 0 free parameters · 8 assumptions · 0 invented entities

The paper's results rest on classical PDE tools and the structural assumption that solutions are positive classical solutions. No empirical constants or fitted parameters appear. The new functional Σ is an analytical construction, not a postulated physical entity.

assumptions (8)

standard math Bochner's formula (Lemma 2.1)
Used in the proof of Theorem 2.1 to relate Δ|∇Λ|^2 and |D^2Λ|^2.
standard math Lemma 2.2: ∂|∇f|^2/∂ν ≤ 0 on convex boundary for f with zero Neumann data
Cited from Evans; gives the non-positive boundary term in Theorem 2.1.
standard math Pointwise estimate |Δf|^2 ≤ n|D^2f|^2
Used in Corollary 2.3 and in the proof of Theorem 4.1.
standard math 1D Sobolev embedding W^{1,1} ⊂ L^∞
Used at the end of the proof of Theorem 5.1 to bound ||log(1+u)||_∞.
domain assumption Semigroup estimate Lemma 5.8 for the heat equation with mass-conserved source
Gives uniform L^r bounds on v and ∂xv used in Lemma 5.10.
domain assumption Local existence and uniqueness of classical positive solutions to (5.1), cited to [2]
The paper assumes such solutions exist on a maximal time interval.
domain assumption Structural assumptions D(u)=(1+u)^{-p}, S(u)=u(1+u)^{-q}, with p-q=1 and q∈(1/2,1] for the main existence result
Restricts the class of Keller-Segel systems considered; q>1/2 is used to bound a power of (1+u) against mass conservation.
domain assumption Convexity of Ω for Theorem 2.1
Required for Lemma 2.2 to control the boundary term.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Nonlinear Fisher information, corresponding functional inequalities and applications." pith.science (2026). https://pith.science/paper/QIGDXALV

@misc{pith2026250901475,
  author       = {Pith},
  title        = {Pith review of: Nonlinear Fisher information, corresponding functional inequalities and applications},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QIGDXALV}},
  note         = {Machine review of arXiv:2509.01475}
}

abstract

We study the evolution of the nonlinear version of the Fisher information along the quasilinear heat equation. We also provide a nonlinear version of a recent functional inequality (Cie\'slak--Fuest--Hajduk--Sier\.z\k{e}ga, 2024), corresponding to the nonlinear heat equation. Next, applications of our version of nonlinear Fisher information to the 1D critical quasilinear fully parabolic Keller--Segel system are given. In particular, the global existence of solutions to the critical nonlinear diffusion/nonlinear sensitivity 1D fully parabolic Keller--Segel system is obtained for certain type of diffusion. Last, but not least, we also study the version of the Fisher information along the $p$-Laplace equation.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

21 extracted references · 21 canonical work pages

[9]

Cieślak and K

T. Cieślak and K. Fujie, No critical nonlinear diﬀusion in 1D quasilinear fully para - bolic chemotaxis system, Proc. Amer. Math. Soc. 146 (2018), no. 6, 2529–2540

work page 2018
[1]

Bakry and M

D. Bakry and M. Émery, Diﬀusions hypercontractives, in Séminaire de probabilités, XIX, 1983/84 , 177–206, Lecture Notes in Math., 1123, Springer, Berlin

work page 1983
[2]

Bellomo, A

N. Bellomo, A. Bellouquid, Y. Tao and M. Winkler, Toward a mathematical theory of Keller-Segel models of pattern formation in biological tis sues, Math. Models Methods Appl. Sci. 25 (2015), no. 9, 1663–1763

work page 2015
[3]

Bieganowski, T

B. Bieganowski, T. Cieślak, K. Fujie and T. Senba, Boundedness of solutions to the critical fully parabolic quasilinear one-dimensional Kel ler-Segel system, Math. Nachr. 292 (2019), no. 4, 724–732

work page 2019
[4]

P. M. Bies and T. Cieślak, Global-in-time regular unique solutions with positive tem - perature to one-dimensional thermoelasticity, SIAM J. Math. Anal. 55 (2023), no. 6, 7024–7038

work page 2023
[5]

P. M. Bies and T. Cieślak, Time-asymptotics of a heated string, Math. Ann. 391 (2025), no. 4, 5941–5964

work page 2025
[6]

P. M. Bies, T. Cieślak, M. Fuest, J. Lankeit, B. Muha and S. Trifunović, Existence, uniqueness, and long-time asymptotic behavior of regular s olutions in multidimen- sional thermoelasticity, arXiv:2507.20794v1

work page Pith review arXiv
[7]

J. A. Carrillo, A. Jüngel, P. A. Markowich, G. Toscani and A. Unterreiter, Entropy dissipation methods for degenerate parabolic problems and generalized Sobolev inequal- ities, Monatsh. Math. 133 (2001), no. 1, 1–82

work page 2001

Show all 21 references

[8]

J. A. Carrillo and G. Toscani, Asymptotic L1-decay of solutions of the porous medium equation to self-similarity, Indiana Univ. Math. J. 49 (2000), 113–142

2000
[10]

Cieślak and K

T. Cieślak and K. Fujie, Global existence in the 1D quasilinear parabolic-elliptic chemotaxis system with critical nonlinearity, Discrete Contin. Dyn. Syst. Ser. S 13 (2020), no. 2, 165–176

2020
[11]

Cieślak, M

T. Cieślak, M. Fuest, K. Hajduk and M. Sierżęga, On the existence of global solutions for the 3D chemorepulsion system, Z. Anal. Anwend. 43 (2024), no. 1-2, 49–65

2024
[12]

Cieślak, M

T. Cieślak, M. Gaczkowski and W. Kryński, A functional inequality between Hessians in spaces with non-zero curvature, arXiv:2506.07292

arXiv
[13]

Edgeworth, On the probable errors of frequency-constants, J

F. Edgeworth, On the probable errors of frequency-constants, J. Roy. Stat. Soc. 71 (1908), 381–397, 499–512, 651–678

1908
[14]

L. C. Evans, Entropy and partial diﬀerential equations. Create Space Independent Publishing Platform, South Carolina, USA, 2014

2014
[15]

Fisher, Theory of statistical estimation, Proc

R. Fisher, Theory of statistical estimation, Proc. Camb. Phil. Soc. 22 (1925), 700–725

1925
[16]

K. Fujie, Energy-like functional in a quasilinear parabolic chemota xis system, in Geo- metric properties for parabolic and elliptic PDEs , 67–77, Springer INdAM Ser., 47, Springer, Cham
[17]

Jüngel, Entropy methods for diﬀusive partial diﬀerential equation s, SpringerBriefs in Math., Springer, 2016

A. Jüngel, Entropy methods for diﬀusive partial diﬀerential equation s, SpringerBriefs in Math., Springer, 2016

2016
[18]

Li and J

S. Li and J. Yang, Characterisations for the depletion of reactant in a one-di mensional dynamic combustion model, Phys. D 461 (2024), Paper No. 134099, 9 pp

2024
[19]

Savaré and G

G. Savaré and G. Toscani, The concavity of Rényi entropy power, IEEE Trans. Inform. Theory 60 (2014), no. 5, 2687–2693

2014
[20]

C. E. Shannon, A mathematical theory of communication, Bell System Tech. J. 27 (1948), 379–423, 623–656. NONLINEAR FISHER INFORMATION 33

1948
[21]

Winkler, Global large-data solutions in a chemotaxis-(Navier-)Sto kes system mod- eling cellular swimming in ﬂuid drops, Comm

M. Winkler, Global large-data solutions in a chemotaxis-(Navier-)Sto kes system mod- eling cellular swimming in ﬂuid drops, Comm. Partial Diﬀerential Equations 37 (2012), no. 2, 319–351. Institute of Ma thema tics Polish Academy of Sciences Śniadeckich 8, 00-656 W arsza w a, P...

2012

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [9]

Cieślak and K

T. Cieślak and K. Fujie, No critical nonlinear diﬀusion in 1D quasilinear fully para - bolic chemotaxis system, Proc. Amer. Math. Soc. 146 (2018), no. 6, 2529–2540

work page 2018

[2] [1]

Bakry and M

D. Bakry and M. Émery, Diﬀusions hypercontractives, in Séminaire de probabilités, XIX, 1983/84 , 177–206, Lecture Notes in Math., 1123, Springer, Berlin

work page 1983

[3] [2]

Bellomo, A

N. Bellomo, A. Bellouquid, Y. Tao and M. Winkler, Toward a mathematical theory of Keller-Segel models of pattern formation in biological tis sues, Math. Models Methods Appl. Sci. 25 (2015), no. 9, 1663–1763

work page 2015

[4] [3]

Bieganowski, T

B. Bieganowski, T. Cieślak, K. Fujie and T. Senba, Boundedness of solutions to the critical fully parabolic quasilinear one-dimensional Kel ler-Segel system, Math. Nachr. 292 (2019), no. 4, 724–732

work page 2019

[5] [4]

P. M. Bies and T. Cieślak, Global-in-time regular unique solutions with positive tem - perature to one-dimensional thermoelasticity, SIAM J. Math. Anal. 55 (2023), no. 6, 7024–7038

work page 2023

[6] [5]

P. M. Bies and T. Cieślak, Time-asymptotics of a heated string, Math. Ann. 391 (2025), no. 4, 5941–5964

work page 2025

[7] [6]

P. M. Bies, T. Cieślak, M. Fuest, J. Lankeit, B. Muha and S. Trifunović, Existence, uniqueness, and long-time asymptotic behavior of regular s olutions in multidimen- sional thermoelasticity, arXiv:2507.20794v1

work page Pith review arXiv

[8] [7]

J. A. Carrillo, A. Jüngel, P. A. Markowich, G. Toscani and A. Unterreiter, Entropy dissipation methods for degenerate parabolic problems and generalized Sobolev inequal- ities, Monatsh. Math. 133 (2001), no. 1, 1–82

work page 2001