On the Relation of Exact Hydrodynamics to the Chapman-Enskog Series

Florian Kogelbauer; Ilya Karlin

REVIEW 2 major objections 5 minor 1 cited by

On the Relation of Exact Hydrodynamics to the Chapman-Enskog Series

T0 review · 2 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read The paper claims that the Chapman-Enskog series is only a local Taylor expansion around equilibrium, equivalent order by order to the exact spectral closure as the Knudsen number vanishes, and that in a one-dimensional BGK-type model it…

desk verdict A useful conceptual note with a solid explicit example, but the all-orders CE-spectral equivalence is asserted from analytic perturbation theory without actually deriving the coefficient matching. read the letter →

arxiv 2506.17441 v1 pith:LPJIQ2FD submitted 2025-06-20 math-ph math.APmath.MP

Florian Kogelbauer , Ilya Karlin This is my paper · ORCID

classification math-phmath.APmath.MP MSC 82C4076P0535Q2041A60

keywords Chapman-EnskogseriesspectralclosureslowmanifoldKnudsennumberBGKmodelcriticalwaveasymptoticdivergencekinetictheory

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 tries to settle what the Chapman-Enskog series actually is: a local Taylor expansion in the Knudsen number around equilibrium, not a global route to hydrodynamics. It argues that, for linear kinetic equations, this series agrees order by order with the exact spectral closure built from slow kinetic eigenmodes as the Knudsen number tends to zero. In an explicit one-dimensional BGK-type model, the series is shown to be strongly divergent for every nonzero Knudsen number, while the spectrally closed hydrodynamics remain well defined at any Knudsen number. If correct, higher-order Burnett and super-Burnett corrections are asymptotic refinements near equilibrium, and the observed instabilities of those corrections are artifacts of polynomial approximation rather than physical breakdown.

What carries the argument

The central object is the hydrodynamic eigenvalue branch $\lambda(k,\mathrm{Kn})$ of the linearized kinetic operator, together with the scaling relation $\lambda(k,\mathrm{Kn})=(1/\mathrm{Kn})\hat\lambda(k\, \mathrm{Kn})$. This identity ties the Knudsen-number expansion of the Chapman-Enskog series to a Taylor expansion of the exact eigenvalue in wave number, so analytic perturbation theory transfers local analyticity of the branches into all-orders equivalence with the CE series. In the one-dimensional model, the spectral closure is encoded in the transcendental equation $\mathcal{Z}(i\tau k)=i\tau k$ for the plasma dispersion function $\mathcal{Z}$, whose Taylor coefficients coincide with the chord-diagram sequence A000699 and grow factorially.

What would settle it

Compute the Chapman-Enskog coefficients recursively for any linear kinetic model whose exact slow eigenvalue branch is known, and compare them term by term with the Taylor coefficients of that branch about wave number zero; a mismatch at any finite order would disprove the claimed all-orders local equivalence. Conversely, find a kinetic model that exhibits criticality but whose CE series has a nonzero radius of convergence, which would break the compact-support argument for global divergence.

Watch

Extended reading notes

Core claim

The paper establishes that, for linear kinetic equations, the Chapman-Enskog series and the exact spectral closure are locally the same object: the CE series is the small-Knudsen Taylor expansion of the exact slow eigenvalue function $\lambda(k,\mathrm{Kn})$. This equivalence holds to all orders because the scaling $\lambda(k,\mathrm{Kn})=(1/\mathrm{Kn})\hat\lambda(k\, \mathrm{Kn})$ and analytic perturbation theory make the Knudsen expansion literally a Taylor expansion of the eigenvalue branch in wave number. Globally, the branches exist only up to a critical wave number, so the true branch is compactly supported and cannot equal a Taylor series everywhere. In the explicit one-dimensional BGK-type model, the CE coefficients grow like $(2n-1)!!$, hence strong divergence at every nonzero Knudsen number, while the spectral closure through the plasma dispersion function is defined for all Knudsen numbers.

Load-bearing premise

The whole conclusion rests on the assumption that the slow eigenvalues of the linearized kinetic operator scale as $\lambda(k,\mathrm{Kn})=(1/\mathrm{Kn})\hat\lambda(k\, \mathrm{Kn})$ and are analytic in the wave number up to the needed order, so that the Knudsen-number expansion is literally the Taylor expansion of the exact spectral eigenvalue.

Editorial extensions

If this is right

Higher-order Chapman-Enskog and Burnett corrections should be read as asymptotic expansions around equilibrium, not as a convergent route to hydrodynamic equations at finite Knudsen number.
Bobylev-type sign changes in the dissipation relation are consequences of truncating the true compactly supported eigenvalue branch by a polynomial, not evidence that the underlying kinetic model is unstable.
For linear kinetic models, the spectrally closed hydrodynamics are well defined for every Knudsen number, giving a unique optimal reduction wherever the slow mode exists.
Finite-moment Grad closures can have convergent CE series because moment truncation removes criticality, so their convergence is a special property, not evidence against the general obstruction.
In the explicit model, the CE coefficients grow like double factorials, so no nonzero Knudsen number lies within the radius of convergence; only equilibrium itself is a point of convergence.

Reading between the lines

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

A natural testable extension is that the same local-equivalence and global-divergence picture holds for the full nonlinear Boltzmann equation only if the slow manifold inherits the linear criticality; the paper demonstrates the linear case but does not prove nonlinearity.
The factorial growth of the coefficients ties the CE expansion to Borel-summable zero-dimensional field-theory expansions, so resummation methods could turn the divergent series into a usable global closure; the paper notes the analogy but stops short of proposing one.
One could build global hydrodynamic closures by replacing Taylor truncations with rational or Pade-type approximants of the exact eigenvalue branch, which would by construction keep dissipation sign-definite and avoid Bobylev instabilities.
If criticality is generic, any finite-order hydrodynamic theory obtained as a polynomial in the Knudsen number inherits a finite validity range in wave number, suggesting a quantitative criterion for when Burnett-type equations should be trusted.

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

Quick take: this is a readable note that connects the known divergence of the Chapman–Enskog series for BGK shear flows to the compact-support criticality of spectral closure. The explicit one-dimensional calculation is correct. The soft spot is that the central claim—local equivalence of CE series and spectral closure to all orders—is not proven; Kato analyticity alone does not identify the CE recursion with the Taylor expansion of the eigenvalue. The paper expands the exact dispersion relation and cites prior work for the identification with CE, rather than computing the CE series independently.

What is good: the compact-support argument is a clean way to see why polynomial CE truncations like Burnett cannot represent the true eigenvalue branch globally. The explicit model is worked out properly: the series (10) matches A000699, the (2n−1)!! growth implies factorial divergence away from zero wave number, and figure 2 makes the instability mechanism concrete. The discussion of why Grad-type truncations can converge while the full CE diverges is also helpful.

Where it is soft: the all-orders equivalence is asserted. Analytic perturbation theory gives analyticity of the eigenvalue branches and their Taylor expansions, but not that the Chapman–Enskog normal-solution recursion produces those same coefficients. That identification is standard for linear kinetic models, but it is not derived here, and the paper leans on citations to McLennan and Ellis–Pinsky plus the authors' own spectral closure papers. For the explicit model, the paper never runs the CE recursion; it expands equation (7), the exact spectral solution, and asserts this is the CE series. If the general identification were false in some class of linear kinetic operators, the divergence conclusion would still likely hold for this model (it matches Santos et al.), but the central conceptual claim would need a separate proof.

There is also a minor typesetting glitch in equation (7), where the plasma dispersion function Z is missing from the displayed equation; the intent is clear from context. The self-citation density is high but not abusive; the cited results are real and the paper builds on them transparently.

Net: a useful conceptual note that deserves a serious referee, but the referee should ask for an explicit derivation or a precise pointer to where the CE-to-eigenvalue coefficient matching is proven. I would not cite it as a primary source in my own work in the next year, but I would bring it to a reading group.

Referee Report

2 major / 5 minor

Summary. The paper claims two results. First, for linear kinetic equations, it asserts that the Chapman-Enskog series is locally equivalent, to all orders in the Knudsen number, to the exact spectral closure built from slow hydrodynamic eigenmodes of the linearized collision operator; the stated reason is that the eigenvalue branches are locally analytic in wave number. Second, for the one-dimensional BGK-type shear model (4), the Taylor expansion of the diffusion eigenvalue is identified with the CE series, its coefficients are matched to OEIS A000699, and the factorial growth (11) is used to conclude that the CE series diverges for every nonzero Knudsen number while the spectral closure remains defined up to the critical wave number (9).

Significance. If the all-orders equivalence were rigorously established, the paper would give a clean geometric explanation of the status of Chapman-Enskog theory: CE is a local asymptotic expansion of the exact spectral closure, Bobylev-type instabilities are artifacts of polynomial approximation, and the exact closure remains valid beyond the radius of convergence of CE. The explicit one-dimensional model is a valuable concrete illustration: it uses an exact transcendental dispersion relation, identifies the Taylor coefficients with a known integer sequence, and quotes an external proof of factorial divergence. The paper contains no fitted parameters and the explicit divergence statement is checkable from Eq. (10) and Eq. (11). The weakness is that the central CE-to-eigenvalue identification is asserted rather than derived, so the significance of the paper hinges on a missing proof.

major comments (2)

[After Eq. (3), paragraph beginning 'The fundamental observation'] The central all-orders equivalence is asserted rather than derived. The paper states that because the eigenvalue branches are locally analytic and because of the coupling (3), analytical spectral perturbation theory guarantees that the slow spectral closure is equivalent to the CE series to all orders. Kato's analytic perturbation theory gives analyticity of isolated eigenvalue branches, but the Chapman-Enskog series is defined by a normal-solution recursion, and no argument is given that this recursion produces the Taylor coefficients of the eigenvalue function (2) for a general linearized collision operator. The cited results of McLennan and Ellis-Pinsky prove analyticity of the branches, not equality of the CE coefficients with the eigenvalue Taylor coefficients. This step is load-bearing: without it, the paper establishes analyticity of the spectral branches but not the claimed local equivalence to CE.
[Eqs. (7)-(10)] For the explicit model, the series (10) is obtained by expanding the exact dispersion relation (7), and this series is then identified with the Chapman-Enskog series without carrying out the CE normal-solution recursion for the kinetic equation (4). The factorial divergence (11) therefore applies to the Taylor expansion of the diffusion eigenvalue; it proves divergence of the CE series only if the coefficient-wise identity between the CE expansion and (10) is demonstrated. Please provide the independent CE calculation for (4), or state the model-specific argument that identifies the two expansions.

minor comments (5)

[Abstract] The word 'expect' should be 'except'.
[Eq. (7)] The displayed formula for Eq. (7) appears to have lost a fraction; please write the argument of the plasma dispersion function explicitly, e.g. as Z((i\tau\lambda_d+1)/(\tau k))=i\tau k.
[After Eq. (10)] The sentence 'At the origin, the coefficients trivially sum up to zero' is inaccurate: the value of the series at \kappa=0 is zero because every term contains a positive power of \kappa, not because the coefficients sum to zero.
[Abstract and Conclusion] The phrase 'defined globally for any Knudsen number' overstates the result: for each \tau the diffusion branch exists only for |k| \le k_{\mathrm{crit}} with k_{\mathrm{crit}} given by (9), so the closure is global in Knudsen number but not in wave number.
[After Eq. (10)] The identification of the coefficients of (10) with OEIS A000699 is made by citing [43]; since this identification is part of the divergence argument, please display the first few coefficients and the relevant recurrence or generating-function check explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the CE/spectral identification is underproved but not constructed from its own output.

full rationale

The derivation chain is not circular under the stated criteria. The scaling relation (3) is an exact rescaling of the linearized operator, not a fitted input. Local analyticity of the eigenvalue branches is supported by external results (McLennan [5], Ellis–Pinsky [24]), and the factorial divergence of the explicit model is grounded in external combinatorial results [42,43,44] and OEIS A000699, not in the paper's own fitted parameters. The authors' self-citations [17,25,26,28,36] supply the spectral-closure framework and the explicit model's dispersion relation, but the load-bearing quantitative facts (analyticity, criticality, coefficient growth) are independently cited and are not outputs of the present paper. The main weakness is that the 'fundamental observation' — that the CE series is the Knudsen expansion of the spectral eigenvalue — is stated rather than proved from the standard CE recursion, so the all-orders equivalence and the example's identification of Eq. (10) as the CE series inherit that proof gap. That is a correctness/rigor concern, not a reduction of the result to its own input by construction; no equation in the paper is fitted to a target prediction, and no self-citation is the sole justification for the central claim. Accordingly the circularity score is 0.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The central claim involves no fitted numbers. It rests on the spectral structure of the linearized Boltzmann or BGK operator, on the scaling Eq. (3), on the dispersion relation Eq. (7) taken from the authors' prior spectral closure work, and on known combinatorial asymptotics of A000699. No new entities are postulated.

assumptions (5)

domain assumption The linearized kinetic operator has isolated hydrodynamic eigenvalue branches that are locally analytic in wave number and in Knudsen number near the five-fold zero eigenvalue.
Used for the general local-equivalence claim; sourced to refs [5,24,29], not proven in this note.
domain assumption The scaling lambda(k,Kn)=1/Kn times lambda-hat(k times Kn) of Eq. (3) holds.
This is what turns the CE expansion in Knudsen number into an expansion in wave number; stated without proof for general cross sections.
domain assumption The one-dimensional BGK-type model (4) has a diffusion mode satisfying the plasma dispersion relation (7), with critical wave number (9).
Taken from the authors' earlier spectral closure paper [36]; no derivation is repeated here.
standard math The Taylor coefficients of the diffusion mode are the integer sequence A000699 with asymptotics (2n-1)!!.
The identification relies on external combinatorial results [43,44]; it supplies the factorial divergence.
standard math Analytic perturbation theory applies to the isolated eigenvalue branches.
Standard Kato theory [29] is invoked as the basis for the all-orders local equivalence, but the mapping to CE coefficients is not carried out.

how reviews work

0 comments

Cite this review

Pith. "Pith review of On the Relation of Exact Hydrodynamics to the Chapman-Enskog Series." pith.science (2026). https://pith.science/paper/LPJIQ2FD

@misc{pith2026250617441,
  author       = {Pith},
  title        = {Pith review of: On the Relation of Exact Hydrodynamics to the Chapman-Enskog Series},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LPJIQ2FD}},
  note         = {Machine review of arXiv:2506.17441}
}

read the original abstract

We demonstrate that the Chapman-Enskog series is locally equivalent to the exact spectral closure defined on slow kinetic eigenmodes in the limit of vanishing Knudsen number. We further show that the Chapman-Enskog series diverges everywhere expect at the global equilibrium for an explicit example, while the exact spectrally closed hydrodynamics are defined globally for any Knudsen number.

Figures

Figures reproduced from arXiv: 2506.17441 by the authors.

**Figure 2.** FIG. 2. Comparison of the shear mode (solid black line) to [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

Learning the Optimal Hydrodynamic Closure
physics.flu-dyn 2025-01 conditional novelty 6.0 of 10

A neural network trained on density fluctuations learns generalized transport coefficients that reproduce Shakhov and DSMC spectra up to Knudsen number 10, extending spectral closure hydrodynamics beyond its critical ...

Reference graph

Works this paper leans on

50 extracted references · 42 canonical work pages · cited by 1 Pith paper

[1]

Chapman and T

S. Chapman and T. G. Cowling, The mathematical theory of non-uniform gases: an account of the kinetic theory of viscosity, thermal conduction and diffusion in gases (Cambridge university press, 1990)

1990
[2]

Grad, Communications on pure and applied mathe- matics 2, 331 (1949)

H. Grad, Communications on pure and applied mathe- matics 2, 331 (1949)

work page 1949
[3]

J. H. Ferziger, H. G. Kaper, et al., (1972)

work page 1972
[4]

Grad, The physics of Fluids 6, 147 (1963)

H. Grad, The physics of Fluids 6, 147 (1963)

work page 1963
[5]

J. A. McLennan, The Physics of Fluids 8, 1580 (1965)

work page 1965
[6]

A. N. Gorban and I. V. Karlin, Physical Review Letters 77, 282 (1996)

work page 1996
[7]

Saint-Raymond, Bulletin of the American Mathemat- ical Society 51, 247 (2014)

L. Saint-Raymond, Bulletin of the American Mathemat- ical Society 51, 247 (2014)

work page 2014
[8]

G. S. Denicol and J. Noronha, arXiv preprint arXiv:1608.07869 (2016)

arXiv 2016

Show all 50 references

[9]

Santos, Physical review letters 100, 078003 (2008)

A. Santos, Physical review letters 100, 078003 (2008). 5

2008
[10]

Burnett, Proceedings of the London Mathematical So- ciety 2, 382 (1936)

D. Burnett, Proceedings of the London Mathematical So- ciety 2, 382 (1936)

1936
[11]

A. V. Bobylev, in Akademiia Nauk SSSR Doklady , Vol. 262 (1982) pp. 71–75

1982
[12]

Santos, J

A. Santos, J. J. Brey, and J. W. Dufty, Physical review letters 56, 1571 (1986)

1986
[13]

The moment closure problem [2] arises when we try to derive a self-consistent evolution of the macro- scopic observables from a kinetic description

and Boltzmann [14] that lead to the modern devel- opment of kinetic theory consist in the abstraction evolu- tion of a distribution function from which the evolution of physical observables such as the temperature or the velocity field of a gas can be inferred by taking expec-...

2025 arXiv
[14]

J. C. Maxwell, Philosophical transactions of the Royal Society of London , 49 (1867)
[15]

Boltzmann, Weitere Studien ¨ uber das W¨ armegleichgewicht unter Gasmolek¨ ulen, Vol

L. Boltzmann, Weitere Studien ¨ uber das W¨ armegleichgewicht unter Gasmolek¨ ulen, Vol. 66 (Aus der kk Hot-und Staatsdruckerei, 1872)
[16]

Hilbert, in Integralgleichungen und Gleichungen mit unendlich vielen Unbekannten (Springer, 1989) pp

D. Hilbert, in Integralgleichungen und Gleichungen mit unendlich vielen Unbekannten (Springer, 1989) pp. 6– 169

1989
[17]

A. N. Gorban and I. V. Karlin, Invariant manifolds for physical and chemical kinetics, Vol. 660 (Springer, 2005)

2005
[18]

Kogelbauer and I

F. Kogelbauer and I. Karlin, Physical Review E 110, 055105 (2024)

2024
[19]

E. N. Lorenz, Journal of Atmospheric Sciences 43, 1547 (1986)

1986
[20]

MacKay, in Energy localisation and Transfer (World Scientific, 2004) pp

R. MacKay, in Energy localisation and Transfer (World Scientific, 2004) pp. 149–192

2004
[21]

Cabr´ e, E

X. Cabr´ e, E. Fontich, and R. De la Llave, Indiana Uni- versity mathematics journal , 283 (2003)

2003
[22]

Y. Wang, W. Ubachs, and W. Van De Water, The Jour- nal of chemical physics 150 (2019)

2019
[23]

Siewert, Physics of Fluids 15, 1696 (2003)

C. Siewert, Physics of Fluids 15, 1696 (2003)

2003
[24]

Cercignani and C

C. Cercignani and C. Cercignani, The Boltzmann equa- tion (Springer, 1988)

1988
[25]

R. S. Ellis and M. A. Pinsky, J. Math. Pures Appl 54, 125 (1975)

1975
[26]

Kogelbauer and I

F. Kogelbauer and I. Karlin, Continuum Mechanics and Thermodynamics 36, 1685 (2024)

2024
[27]

Kogelbauer and I

F. Kogelbauer and I. Karlin, Continuum Mechanics and Thermodynamics 37, 45 (2025)

2025
[28]

Nishida and K

T. Nishida and K. Imai, Publications of the Research Institute for Mathematical Sciences 12, 229 (1976)

1976
[29]

Kogelbauer, C

F. Kogelbauer, C. Zheng, and I. Karlin, arXiv preprint arXiv:2501.13938 (2025)

2025 arXiv
[30]

Kato, Perturbation theory for linear operators , Vol

T. Kato, Perturbation theory for linear operators , Vol. 132 (Springer Science & Business Media, 2013)

2013
[31]

Rudin, Real and complex analysis (McGraw-Hill, Inc., 1987)

W. Rudin, Real and complex analysis (McGraw-Hill, Inc., 1987)

1987
[32]

Dudy´ nski, Journal of Statistical Physics 153, 1084 (2013)

M. Dudy´ nski, Journal of Statistical Physics 153, 1084 (2013)

2013
[33]

Karlin, S

I. Karlin, S. Chikatamarla, and M. Kooshkbaghi, Phys- ica A: Statistical Mechanics and its Applications 403, 189 (2014)

2014
[34]

Kogelbauer, Continuum Mechanics and Thermody- namics 32, 1141 (2020)

F. Kogelbauer, Continuum Mechanics and Thermody- namics 32, 1141 (2020)

2020
[35]

Cercignani and A

C. Cercignani and A. Daneri, Journal of Applied Physics 34, 3509 (1963)

1963
[36]

P. L. Bhatnagar, E. P. Gross, and M. Krook, Physical review 94, 511 (1954)

1954
[37]

Kogelbauer, Continuum Mechanics and Thermody- namics 33, 431 (2021)

F. Kogelbauer, Continuum Mechanics and Thermody- namics 33, 431 (2021)

2021
[38]

B. D. Fried and S. D. Conte, The plasma dispersion func- tion: the Hilbert transform of the Gaussian (Academic press, 2015)

2015
[39]

Abramowitz and I

M. Abramowitz and I. A. Stegun, Handbook of mathe- matical functions with formulas, graphs, and mathemati- cal tables, Vol. 55 (US Government printing office, 1948)

1948
[40]

L. S. Garc´ ıa-Col´ ın, R. M. Velasco, and F. J. Uribe, Physics Reports 465, 149 (2008)

2008
[41]

Uribe, R

F. Uribe, R. Velasco, and L. Garcia-Colin, Physical Re- view E 62, 5835 (2000)

2000
[42]

The On-Line Encyclopedia of Integer Sequences,

OEIS Foundation Inc., “The On-Line Encyclopedia of Integer Sequences,” (2025), published electronically at http://oeis.org/A000699

2025
[43]

Flajolet and M

P. Flajolet and M. Noy, in Formal Power Series and Al- gebraic Combinatorics: 12th International Conference, FPSAC’00, Moscow, Russia, June 2000, Proceedings (Springer, 2000) pp. 191–201

2000
[44]

Marie and K

N. Marie and K. Yeats, arXiv preprint arXiv:1210.5457 (2012)

2012 arXiv
[45]

Borinsky, The Electronic Journal of Combinatorics , P4 (2018)

M. Borinsky, The Electronic Journal of Combinatorics , P4 (2018)

2018
[46]

C. M. Bender and T. T. Wu, Physical Review 184, 1231 (1969)

1969
[47]

D. J. Kleitman, Studies in Applied Mathematics 49, 297 (1970)

1970
[48]

Glimm and A

J. Glimm and A. Jaffe, Quantum physics: a functional integral point of view (Springer Science & Business Me- dia, 2012)

2012
[49]

Rosenau, Physical Review A 40, 7193 (1989)

P. Rosenau, Physical Review A 40, 7193 (1989)

1989
[50]

I. V. Karlin and S. Ansumali, Physical Review E—Statistical, Nonlinear, and Soft Matter Physics 76, 025701 (2007)

2007

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

Chapman and T

S. Chapman and T. G. Cowling, The mathematical theory of non-uniform gases: an account of the kinetic theory of viscosity, thermal conduction and diffusion in gases (Cambridge university press, 1990)

1990

[2] [2]

Grad, Communications on pure and applied mathe- matics 2, 331 (1949)

H. Grad, Communications on pure and applied mathe- matics 2, 331 (1949)

work page 1949

[3] [3]

J. H. Ferziger, H. G. Kaper, et al., (1972)

work page 1972

[4] [4]

Grad, The physics of Fluids 6, 147 (1963)

H. Grad, The physics of Fluids 6, 147 (1963)

work page 1963

[5] [5]

J. A. McLennan, The Physics of Fluids 8, 1580 (1965)

work page 1965

[6] [6]

A. N. Gorban and I. V. Karlin, Physical Review Letters 77, 282 (1996)

work page 1996

[7] [7]

Saint-Raymond, Bulletin of the American Mathemat- ical Society 51, 247 (2014)

L. Saint-Raymond, Bulletin of the American Mathemat- ical Society 51, 247 (2014)

work page 2014

[8] [8]

G. S. Denicol and J. Noronha, arXiv preprint arXiv:1608.07869 (2016)

arXiv 2016